implement our own sharding for PageCache (16 HashMaps based on lower 4 bits of blkno)

1. Compute may generate WAL on shutdown. The test assumes that after
shutdown,
no further ingest happens. Tweak the compute shutdown to make the
assumption true.
2. Assertion of local layer count post cold migration is not right since
we may have downloaded
layers due to ingest. Remove it.

Closes https://github.com/neondatabase/neon/issues/11587

.github/actionlint.yml

@@ -6,6 +6,7 @@ self-hosted-runner:
     - small
     - small-metal
     - small-arm64
+    - unit-perf
     - us-east-2
 config-variables:
   - AWS_ECR_REGION

.github/actions/allure-report-generate/action.yml

@@ -70,6 +70,7 @@ runs:
 
     - name: Install Allure
       shell: bash -euxo pipefail {0}
+      working-directory: /tmp
       run: |
         if ! which allure; then
           ALLURE_ZIP=allure-${ALLURE_VERSION}.zip

.github/workflows/_create-release-pr.yml

@@ -53,10 +53,13 @@ jobs:
             || inputs.component-name == 'Compute' && 'release-compute'
           }}
       run: |
-        today=$(date +'%Y-%m-%d')
-        echo "title=${COMPONENT_NAME} release ${today}" | tee -a ${GITHUB_OUTPUT}
-        echo "rc-branch=rc/${RELEASE_BRANCH}/${today}"  | tee -a ${GITHUB_OUTPUT}
-        echo "release-branch=${RELEASE_BRANCH}"         | tee -a ${GITHUB_OUTPUT}
+        now_date=$(date -u +'%Y-%m-%d')
+        now_time=$(date -u +'%H-%M-%Z')
+        {
+          echo "title=${COMPONENT_NAME} release ${now_date}"
+          echo "rc-branch=rc/${RELEASE_BRANCH}/${now_date}_${now_time}"
+          echo "release-branch=${RELEASE_BRANCH}"
+        } | tee -a ${GITHUB_OUTPUT}
 
     - name: Configure git
       run: |

.github/workflows/build_and_test.yml

@@ -284,7 +284,7 @@ jobs:
       statuses: write
       contents: write
       pull-requests: write
-    runs-on: [ self-hosted, small-metal ]
+    runs-on: [ self-hosted, unit-perf ]
     container:
       image: ${{ needs.build-build-tools-image.outputs.image }}-bookworm
       credentials:
@@ -1271,7 +1271,7 @@ jobs:
           exit 1
 
   deploy:
-    needs: [ check-permissions, push-neon-image-dev, push-compute-image-dev, push-neon-image-prod, push-compute-image-prod, meta, build-and-test-locally, trigger-custom-extensions-build-and-wait ]
+    needs: [ check-permissions, push-neon-image-dev, push-compute-image-dev, push-neon-image-prod, push-compute-image-prod, meta, trigger-custom-extensions-build-and-wait ]
     # `!failure() && !cancelled()` is required because the workflow depends on the job that can be skipped: `push-neon-image-prod` and `push-compute-image-prod`
     if: ${{ contains(fromJSON('["push-main", "storage-release", "proxy-release", "compute-release"]'), needs.meta.outputs.run-kind) && !failure() && !cancelled() }}
     permissions:

Cargo.lock

@@ -1416,6 +1416,7 @@ name = "control_plane"
 version = "0.1.0"
 dependencies = [
  "anyhow",
+ "base64 0.13.1",
  "camino",
  "clap",
  "comfy-table",
@@ -1425,10 +1426,12 @@ dependencies = [
  "humantime",
  "humantime-serde",
  "hyper 0.14.30",
+ "jsonwebtoken",
  "nix 0.27.1",
  "once_cell",
  "pageserver_api",
  "pageserver_client",
+ "pem",
  "postgres_backend",
  "postgres_connection",
  "regex",
@@ -1437,6 +1440,8 @@ dependencies = [
  "scopeguard",
  "serde",
  "serde_json",
+ "sha2",
+ "spki 0.7.3",
  "storage_broker",
  "thiserror 1.0.69",
  "tokio",
@@ -2817,6 +2822,7 @@ dependencies = [
  "hyper 0.14.30",
  "itertools 0.10.5",
  "jemalloc_pprof",
+ "jsonwebtoken",
  "metrics",
  "once_cell",
  "pprof",
@@ -2837,6 +2843,7 @@ dependencies = [
  "utils",
  "uuid",
  "workspace_hack",
+ "x509-cert",
 ]
 
 [[package]]
@@ -4268,6 +4275,7 @@ dependencies = [
  "hyper 0.14.30",
  "indoc",
  "itertools 0.10.5",
+ "jsonwebtoken",
  "md5",
  "metrics",
  "nix 0.27.1",
@@ -5684,9 +5692,9 @@ dependencies = [
 
 [[package]]
 name = "ring"
-version = "0.17.13"
+version = "0.17.14"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "70ac5d832aa16abd7d1def883a8545280c20a60f523a370aa3a9617c2b8550ee"
+checksum = "a4689e6c2294d81e88dc6261c768b63bc4fcdb852be6d1352498b114f61383b7"
 dependencies = [
  "cc",
  "cfg-if",
@@ -5987,6 +5995,7 @@ dependencies = [
  "humantime",
  "hyper 0.14.30",
  "itertools 0.10.5",
+ "jsonwebtoken",
  "metrics",
  "once_cell",
  "pageserver_api",
@@ -7871,6 +7880,7 @@ dependencies = [
  "metrics",
  "nix 0.27.1",
  "once_cell",
+ "pem",
  "pin-project-lite",
  "postgres_connection",
  "pprof",

Cargo.toml

@@ -141,6 +141,7 @@ parking_lot = "0.12"
 parquet = { version = "53", default-features = false, features = ["zstd"] }
 parquet_derive = "53"
 pbkdf2 = { version = "0.12.1", features = ["simple", "std"] }
+pem = "3.0.3"
 pin-project-lite = "0.2"
 pprof = { version = "0.14", features = ["criterion", "flamegraph", "frame-pointer", "prost-codec"] }
 procfs = "0.16"
@@ -174,6 +175,7 @@ signal-hook = "0.3"
 smallvec = "1.11"
 smol_str = { version = "0.2.0", features = ["serde"] }
 socket2 = "0.5"
+spki = "0.7.3"
 strum = "0.26"
 strum_macros = "0.26"
 "subtle"  = "2.5.0"

compute/patches/pgvector.patch

@@ -15,7 +15,7 @@ index 7a4b88c..56678af 100644
  HEADERS = src/halfvec.h src/sparsevec.h src/vector.h
  
 diff --git a/src/hnswbuild.c b/src/hnswbuild.c
-index b667478..dc95d89 100644
+index b667478..1298aa1 100644
 --- a/src/hnswbuild.c
 +++ b/src/hnswbuild.c
 @@ -843,9 +843,17 @@ HnswParallelBuildMain(dsm_segment *seg, shm_toc *toc)
@@ -36,7 +36,7 @@ index b667478..dc95d89 100644
  	/* Close relations within worker */
  	index_close(indexRel, indexLockmode);
  	table_close(heapRel, heapLockmode);
-@@ -1100,12 +1108,39 @@ BuildIndex(Relation heap, Relation index, IndexInfo *indexInfo,
+@@ -1100,13 +1108,25 @@ BuildIndex(Relation heap, Relation index, IndexInfo *indexInfo,
  	SeedRandom(42);
  #endif
  
@@ -48,32 +48,17 @@ index b667478..dc95d89 100644
  
  	BuildGraph(buildstate, forkNum);
  
--	if (RelationNeedsWAL(index) || forkNum == INIT_FORKNUM)
 +#ifdef NEON_SMGR
 +	smgr_finish_unlogged_build_phase_1(RelationGetSmgr(index));
 +#endif
 +
-+	if (RelationNeedsWAL(index) || forkNum == INIT_FORKNUM) {
+ 	if (RelationNeedsWAL(index) || forkNum == INIT_FORKNUM)
  		log_newpage_range(index, forkNum, 0, RelationGetNumberOfBlocksInFork(index, forkNum), true);
-+#ifdef NEON_SMGR
-+		{
-+#if PG_VERSION_NUM >= 160000
-+			RelFileLocator rlocator = RelationGetSmgr(index)->smgr_rlocator.locator;
-+#else
-+			RelFileNode rlocator = RelationGetSmgr(index)->smgr_rnode.node;
-+#endif
-+			if (set_lwlsn_block_range_hook)
-+				set_lwlsn_block_range_hook(XactLastRecEnd, rlocator,
-+										   MAIN_FORKNUM, 0, RelationGetNumberOfBlocks(index));
-+			if (set_lwlsn_relation_hook)
-+				set_lwlsn_relation_hook(XactLastRecEnd, rlocator, MAIN_FORKNUM);
-+		}
-+#endif
-+	}
-+
+ 
 +#ifdef NEON_SMGR
 +	smgr_end_unlogged_build(RelationGetSmgr(index));
 +#endif
- 
++
  	FreeBuildState(buildstate);
  }
+ 

compute/patches/rum.patch

@@ -1,5 +1,5 @@
 diff --git a/src/ruminsert.c b/src/ruminsert.c
-index 255e616..7a2240f 100644
+index 255e616..1c6edb7 100644
 --- a/src/ruminsert.c
 +++ b/src/ruminsert.c
 @@ -628,6 +628,10 @@ rumbuild(Relation heap, Relation index, struct IndexInfo *indexInfo)
@@ -24,24 +24,12 @@ index 255e616..7a2240f 100644
  	/*
  	 * Write index to xlog
  	 */
-@@ -713,6 +721,22 @@ rumbuild(Relation heap, Relation index, struct IndexInfo *indexInfo)
+@@ -713,6 +721,10 @@ rumbuild(Relation heap, Relation index, struct IndexInfo *indexInfo)
  		UnlockReleaseBuffer(buffer);
  	}
  
 +#ifdef NEON_SMGR
-+	{
-+#if PG_VERSION_NUM >= 160000
-+		RelFileLocator rlocator = RelationGetSmgr(index)->smgr_rlocator.locator;
-+#else
-+		RelFileNode rlocator = RelationGetSmgr(index)->smgr_rnode.node;
-+#endif
-+		if (set_lwlsn_block_range_hook)
-+			set_lwlsn_block_range_hook(XactLastRecEnd, rlocator, MAIN_FORKNUM, 0, RelationGetNumberOfBlocks(index));
-+		if (set_lwlsn_relation_hook)
-+			set_lwlsn_relation_hook(XactLastRecEnd, rlocator, MAIN_FORKNUM);
-+
-+		smgr_end_unlogged_build(index->rd_smgr);
-+	}
++	smgr_end_unlogged_build(index->rd_smgr);
 +#endif
 +
  	/*

compute_tools/src/bin/compute_ctl.rs

@@ -29,13 +29,12 @@
 //! ```sh
 //! compute_ctl -D /var/db/postgres/compute \
 //!             -C 'postgresql://cloud_admin@localhost/postgres' \
-//!             -S /var/db/postgres/specs/current.json \
+//!             -c /var/db/postgres/configs/config.json \
 //!             -b /usr/local/bin/postgres \
 //!             -r http://pg-ext-s3-gateway \
 //! ```
 use std::ffi::OsString;
 use std::fs::File;
-use std::path::Path;
 use std::process::exit;
 use std::sync::mpsc;
 use std::thread;
@@ -43,8 +42,7 @@ use std::time::Duration;
 
 use anyhow::{Context, Result};
 use clap::Parser;
-use compute_api::responses::ComputeCtlConfig;
-use compute_api::spec::ComputeSpec;
+use compute_api::responses::ComputeConfig;
 use compute_tools::compute::{
     BUILD_TAG, ComputeNode, ComputeNodeParams, forward_termination_signal,
 };
@@ -118,8 +116,10 @@ struct Cli {
     #[arg(long)]
     pub set_disk_quota_for_fs: Option<String>,
 
-    #[arg(short = 'S', long, group = "spec-path")]
-    pub spec_path: Option<OsString>,
+    // TODO(tristan957): remove alias after compatibility tests are no longer
+    // an issue
+    #[arg(short = 'c', long, alias = "spec-path")]
+    pub config: Option<OsString>,
 
     #[arg(short = 'i', long, group = "compute-id")]
     pub compute_id: String,
@@ -127,8 +127,9 @@ struct Cli {
     #[arg(
         short = 'p',
         long,
-        conflicts_with = "spec-path",
-        value_name = "CONTROL_PLANE_API_BASE_URL"
+        conflicts_with = "config",
+        value_name = "CONTROL_PLANE_API_BASE_URL",
+        requires = "compute-id"
     )]
     pub control_plane_uri: Option<String>,
 }
@@ -138,7 +139,7 @@ fn main() -> Result<()> {
 
     let scenario = failpoint_support::init();
 
-    // For historical reasons, the main thread that processes the spec and launches postgres
+    // For historical reasons, the main thread that processes the config and launches postgres
     // is synchronous, but we always have this tokio runtime available and we "enter" it so
     // that you can use tokio::spawn() and tokio::runtime::Handle::current().block_on(...)
     // from all parts of compute_ctl.
@@ -154,7 +155,7 @@ fn main() -> Result<()> {
 
     let connstr = Url::parse(&cli.connstr).context("cannot parse connstr as a URL")?;
 
-    let cli_spec = try_spec_from_cli(&cli)?;
+    let config = get_config(&cli)?;
 
     let compute_node = ComputeNode::new(
         ComputeNodeParams {
@@ -175,8 +176,7 @@ fn main() -> Result<()> {
             #[cfg(target_os = "linux")]
             vm_monitor_addr: cli.vm_monitor_addr,
         },
-        cli_spec.spec,
-        cli_spec.compute_ctl_config,
+        config,
     )?;
 
     let exit_code = compute_node.run()?;
@@ -201,27 +201,17 @@ async fn init() -> Result<()> {
     Ok(())
 }
 
-fn try_spec_from_cli(cli: &Cli) -> Result<CliSpecParams> {
-    // First, read spec from the path if provided
-    if let Some(ref spec_path) = cli.spec_path {
-        let file = File::open(Path::new(spec_path))?;
-        return Ok(CliSpecParams {
-            spec: Some(serde_json::from_reader(file)?),
-            compute_ctl_config: ComputeCtlConfig::default(),
-        });
+fn get_config(cli: &Cli) -> Result<ComputeConfig> {
+    // First, read the config from the path if provided
+    if let Some(ref config) = cli.config {
+        let file = File::open(config)?;
+        return Ok(serde_json::from_reader(&file)?);
     }
 
-    if cli.control_plane_uri.is_none() {
-        panic!("must specify --control-plane-uri");
-    };
-
-    // If the spec wasn't provided in the CLI arguments, then retrieve it from
+    // If the config wasn't provided in the CLI arguments, then retrieve it from
     // the control plane
-    match get_spec_from_control_plane(cli.control_plane_uri.as_ref().unwrap(), &cli.compute_id) {
-        Ok(resp) => Ok(CliSpecParams {
-            spec: resp.0,
-            compute_ctl_config: resp.1,
-        }),
+    match get_config_from_control_plane(cli.control_plane_uri.as_ref().unwrap(), &cli.compute_id) {
+        Ok(config) => Ok(config),
         Err(e) => {
             error!(
                 "cannot get response from control plane: {}\n\
@@ -233,13 +223,6 @@ fn try_spec_from_cli(cli: &Cli) -> Result<CliSpecParams> {
     }
 }
 
-struct CliSpecParams {
-    /// If a spec was provided via CLI or file, the [`ComputeSpec`]
-    spec: Option<ComputeSpec>,
-    #[allow(dead_code)]
-    compute_ctl_config: ComputeCtlConfig,
-}
-
 fn deinit_and_exit(exit_code: Option<i32>) -> ! {
     // Shutdown trace pipeline gracefully, so that it has a chance to send any
     // pending traces before we exit. Shutting down OTEL tracing provider may

compute_tools/src/compute.rs

@@ -11,7 +11,7 @@ use std::{env, fs};
 use anyhow::{Context, Result};
 use chrono::{DateTime, Utc};
 use compute_api::privilege::Privilege;
-use compute_api::responses::{ComputeCtlConfig, ComputeMetrics, ComputeStatus};
+use compute_api::responses::{ComputeConfig, ComputeCtlConfig, ComputeMetrics, ComputeStatus};
 use compute_api::spec::{
     ComputeAudit, ComputeFeature, ComputeMode, ComputeSpec, ExtVersion, PgIdent,
 };
@@ -303,11 +303,7 @@ struct StartVmMonitorResult {
 }
 
 impl ComputeNode {
-    pub fn new(
-        params: ComputeNodeParams,
-        cli_spec: Option<ComputeSpec>,
-        compute_ctl_config: ComputeCtlConfig,
-    ) -> Result<Self> {
+    pub fn new(params: ComputeNodeParams, config: ComputeConfig) -> Result<Self> {
         let connstr = params.connstr.as_str();
         let conn_conf = postgres::config::Config::from_str(connstr)
             .context("cannot build postgres config from connstr")?;
@@ -315,8 +311,8 @@ impl ComputeNode {
             .context("cannot build tokio postgres config from connstr")?;
 
         let mut new_state = ComputeState::new();
-        if let Some(cli_spec) = cli_spec {
-            let pspec = ParsedSpec::try_from(cli_spec).map_err(|msg| anyhow::anyhow!(msg))?;
+        if let Some(spec) = config.spec {
+            let pspec = ParsedSpec::try_from(spec).map_err(|msg| anyhow::anyhow!(msg))?;
             new_state.pspec = Some(pspec);
         }
 
@@ -327,7 +323,7 @@ impl ComputeNode {
             state: Mutex::new(new_state),
             state_changed: Condvar::new(),
             ext_download_progress: RwLock::new(HashMap::new()),
-            compute_ctl_config,
+            compute_ctl_config: config.compute_ctl_config,
         })
     }
 
@@ -523,11 +519,14 @@ impl ComputeNode {
 
         let pspec = compute_state.pspec.as_ref().expect("spec must be set");
         info!(
-            "starting compute for project {}, operation {}, tenant {}, timeline {}, features {:?}, spec.remote_extensions {:?}",
+            "starting compute for project {}, operation {}, tenant {}, timeline {}, project {}, branch {}, endpoint {}, features {:?}, spec.remote_extensions {:?}",
             pspec.spec.cluster.cluster_id.as_deref().unwrap_or("None"),
             pspec.spec.operation_uuid.as_deref().unwrap_or("None"),
             pspec.tenant_id,
             pspec.timeline_id,
+            pspec.spec.project_id.as_deref().unwrap_or("None"),
+            pspec.spec.branch_id.as_deref().unwrap_or("None"),
+            pspec.spec.endpoint_id.as_deref().unwrap_or("None"),
             pspec.spec.features,
             pspec.spec.remote_extensions,
         );
@@ -631,19 +630,23 @@ impl ComputeNode {
             });
         }
 
-        // Configure and start rsyslog for HIPAA if necessary
-        if let ComputeAudit::Hipaa = pspec.spec.audit_log_level {
-            let remote_endpoint = std::env::var("AUDIT_LOGGING_ENDPOINT").unwrap_or("".to_string());
-            if remote_endpoint.is_empty() {
-                anyhow::bail!("AUDIT_LOGGING_ENDPOINT is empty");
+        // Configure and start rsyslog for compliance audit logging
+        match pspec.spec.audit_log_level {
+            ComputeAudit::Hipaa | ComputeAudit::Extended | ComputeAudit::Full => {
+                let remote_endpoint =
+                    std::env::var("AUDIT_LOGGING_ENDPOINT").unwrap_or("".to_string());
+                if remote_endpoint.is_empty() {
+                    anyhow::bail!("AUDIT_LOGGING_ENDPOINT is empty");
+                }
+
+                let log_directory_path = Path::new(&self.params.pgdata).join("log");
+                let log_directory_path = log_directory_path.to_string_lossy().to_string();
+                configure_audit_rsyslog(log_directory_path.clone(), "hipaa", &remote_endpoint)?;
+
+                // Launch a background task to clean up the audit logs
+                launch_pgaudit_gc(log_directory_path);
             }
-
-            let log_directory_path = Path::new(&self.params.pgdata).join("log");
-            let log_directory_path = log_directory_path.to_string_lossy().to_string();
-            configure_audit_rsyslog(log_directory_path.clone(), "hipaa", &remote_endpoint)?;
-
-            // Launch a background task to clean up the audit logs
-            launch_pgaudit_gc(log_directory_path);
+            _ => {}
         }
 
         // Configure and start rsyslog for Postgres logs export

compute_tools/src/config.rs

@@ -89,6 +89,15 @@ pub fn write_postgres_conf(
             escape_conf_value(&s.to_string())
         )?;
     }
+    if let Some(s) = &spec.project_id {
+        writeln!(file, "neon.project_id={}", escape_conf_value(s))?;
+    }
+    if let Some(s) = &spec.branch_id {
+        writeln!(file, "neon.branch_id={}", escape_conf_value(s))?;
+    }
+    if let Some(s) = &spec.endpoint_id {
+        writeln!(file, "neon.endpoint_id={}", escape_conf_value(s))?;
+    }
 
     // tls
     if let Some(tls_config) = tls_config {
@@ -169,7 +178,7 @@ pub fn write_postgres_conf(
     // and don't allow the user or the control plane admin to change them.
     match spec.audit_log_level {
         ComputeAudit::Disabled => {}
-        ComputeAudit::Log => {
+        ComputeAudit::Log | ComputeAudit::Base => {
             writeln!(file, "# Managed by compute_ctl base audit settings: start")?;
             writeln!(file, "pgaudit.log='ddl,role'")?;
             // Disable logging of catalog queries to reduce the noise
@@ -193,16 +202,20 @@ pub fn write_postgres_conf(
             }
             writeln!(file, "# Managed by compute_ctl base audit settings: end")?;
         }
-        ComputeAudit::Hipaa => {
+        ComputeAudit::Hipaa | ComputeAudit::Extended | ComputeAudit::Full => {
             writeln!(
                 file,
                 "# Managed by compute_ctl compliance audit settings: begin"
             )?;
-            // This log level is very verbose
-            // but this is necessary for HIPAA compliance.
-            // Exclude 'misc' category, because it doesn't contain anythig relevant.
-            writeln!(file, "pgaudit.log='all, -misc'")?;
-            writeln!(file, "pgaudit.log_parameter=on")?;
+            // Enable logging of parameters.
+            // This is very verbose and may contain sensitive data.
+            if spec.audit_log_level == ComputeAudit::Full {
+                writeln!(file, "pgaudit.log_parameter=on")?;
+                writeln!(file, "pgaudit.log='all'")?;
+            } else {
+                writeln!(file, "pgaudit.log_parameter=off")?;
+                writeln!(file, "pgaudit.log='all, -misc'")?;
+            }
             // Disable logging of catalog queries
             // The catalog doesn't contain sensitive data, so we don't need to audit it.
             writeln!(file, "pgaudit.log_catalog=off")?;

compute_tools/src/http/middleware/authorize.rs

@@ -1,7 +1,7 @@
-use std::{collections::HashSet, net::SocketAddr};
+use std::collections::HashSet;
 
 use anyhow::{Result, anyhow};
-use axum::{RequestExt, body::Body, extract::ConnectInfo};
+use axum::{RequestExt, body::Body};
 use axum_extra::{
     TypedHeader,
     headers::{Authorization, authorization::Bearer},
@@ -11,7 +11,7 @@ use futures::future::BoxFuture;
 use http::{Request, Response, StatusCode};
 use jsonwebtoken::{Algorithm, DecodingKey, TokenData, Validation, jwk::JwkSet};
 use tower_http::auth::AsyncAuthorizeRequest;
-use tracing::warn;
+use tracing::{debug, warn};
 
 use crate::http::{JsonResponse, extract::RequestId};
 
@@ -54,8 +54,8 @@ impl AsyncAuthorizeRequest<Body> for Authorize {
         Box::pin(async move {
             let request_id = request.extract_parts::<RequestId>().await.unwrap();
 
-            // TODO: Remove this stanza after teaching neon_local and the
-            // regression tests to use a JWT + JWKS.
+            // TODO(tristan957): Remove this stanza after teaching neon_local
+            // and the regression tests to use a JWT + JWKS.
             //
             // https://github.com/neondatabase/neon/issues/11316
             if cfg!(feature = "testing") {
@@ -64,19 +64,6 @@ impl AsyncAuthorizeRequest<Body> for Authorize {
                 return Ok(request);
             }
 
-            let connect_info = request
-                .extract_parts::<ConnectInfo<SocketAddr>>()
-                .await
-                .unwrap();
-
-            // In the event the request is coming from the loopback interface,
-            // allow all requests
-            if connect_info.ip().is_loopback() {
-                warn!(%request_id, "Bypassed authorization because request is coming from the loopback interface");
-
-                return Ok(request);
-            }
-
             let TypedHeader(Authorization(bearer)) = request
                 .extract_parts::<TypedHeader<Authorization<Bearer>>>()
                 .await
@@ -92,7 +79,7 @@ impl AsyncAuthorizeRequest<Body> for Authorize {
             if data.claims.compute_id != compute_id {
                 return Err(JsonResponse::error(
                     StatusCode::UNAUTHORIZED,
-                    "invalid claims in authorization token",
+                    "invalid compute ID in authorization token claims",
                 ));
             }
 
@@ -112,12 +99,16 @@ impl Authorize {
         token: &str,
         validation: &Validation,
     ) -> Result<TokenData<ComputeClaims>> {
+        debug_assert!(!jwks.keys.is_empty());
+
+        debug!("verifying token {}", token);
+
         for jwk in jwks.keys.iter() {
             let decoding_key = match DecodingKey::from_jwk(jwk) {
                 Ok(key) => key,
                 Err(e) => {
                     warn!(
-                        "Failed to construct decoding key from {}: {}",
+                        "failed to construct decoding key from {}: {}",
                         jwk.common.key_id.as_ref().unwrap(),
                         e
                     );
@@ -130,7 +121,7 @@ impl Authorize {
                 Ok(data) => return Ok(data),
                 Err(e) => {
                     warn!(
-                        "Failed to decode authorization token using {}: {}",
+                        "failed to decode authorization token using {}: {}",
                         jwk.common.key_id.as_ref().unwrap(),
                         e
                     );
@@ -140,6 +131,6 @@ impl Authorize {
             }
         }
 
-        Err(anyhow!("Failed to verify authorization token"))
+        Err(anyhow!("failed to verify authorization token"))
     }
 }

compute_tools/src/metrics.rs

@@ -19,13 +19,13 @@ pub(crate) static INSTALLED_EXTENSIONS: Lazy<UIntGaugeVec> = Lazy::new(|| {
 // but for all our APIs we defined a 'slug'/method/operationId in the OpenAPI spec.
 // And it's fair to call it a 'RPC' (Remote Procedure Call).
 pub enum CPlaneRequestRPC {
-    GetSpec,
+    GetConfig,
 }
 
 impl CPlaneRequestRPC {
     pub fn as_str(&self) -> &str {
         match self {
-            CPlaneRequestRPC::GetSpec => "GetSpec",
+            CPlaneRequestRPC::GetConfig => "GetConfig",
         }
     }
 }

compute_tools/src/spec.rs

@@ -3,9 +3,8 @@ use std::path::Path;
 
 use anyhow::{Result, anyhow, bail};
 use compute_api::responses::{
-    ComputeCtlConfig, ControlPlaneComputeStatus, ControlPlaneSpecResponse,
+    ComputeConfig, ControlPlaneComputeStatus, ControlPlaneConfigResponse,
 };
-use compute_api::spec::ComputeSpec;
 use reqwest::StatusCode;
 use tokio_postgres::Client;
 use tracing::{error, info, instrument};
@@ -21,7 +20,7 @@ use crate::params::PG_HBA_ALL_MD5;
 fn do_control_plane_request(
     uri: &str,
     jwt: &str,
-) -> Result<ControlPlaneSpecResponse, (bool, String, String)> {
+) -> Result<ControlPlaneConfigResponse, (bool, String, String)> {
     let resp = reqwest::blocking::Client::new()
         .get(uri)
         .header("Authorization", format!("Bearer {}", jwt))
@@ -29,14 +28,14 @@ fn do_control_plane_request(
         .map_err(|e| {
             (
                 true,
-                format!("could not perform spec request to control plane: {:?}", e),
+                format!("could not perform request to control plane: {:?}", e),
                 UNKNOWN_HTTP_STATUS.to_string(),
             )
         })?;
 
     let status = resp.status();
     match status {
-        StatusCode::OK => match resp.json::<ControlPlaneSpecResponse>() {
+        StatusCode::OK => match resp.json::<ControlPlaneConfigResponse>() {
             Ok(spec_resp) => Ok(spec_resp),
             Err(e) => Err((
                 true,
@@ -69,40 +68,35 @@ fn do_control_plane_request(
     }
 }
 
-/// Request spec from the control-plane by compute_id. If `NEON_CONTROL_PLANE_TOKEN`
-/// env variable is set, it will be used for authorization.
-pub fn get_spec_from_control_plane(
-    base_uri: &str,
-    compute_id: &str,
-) -> Result<(Option<ComputeSpec>, ComputeCtlConfig)> {
+/// Request config from the control-plane by compute_id. If
+/// `NEON_CONTROL_PLANE_TOKEN` env variable is set, it will be used for
+/// authorization.
+pub fn get_config_from_control_plane(base_uri: &str, compute_id: &str) -> Result<ComputeConfig> {
     let cp_uri = format!("{base_uri}/compute/api/v2/computes/{compute_id}/spec");
-    let jwt: String = match std::env::var("NEON_CONTROL_PLANE_TOKEN") {
-        Ok(v) => v,
-        Err(_) => "".to_string(),
-    };
+    let jwt: String = std::env::var("NEON_CONTROL_PLANE_TOKEN").unwrap_or_default();
     let mut attempt = 1;
 
-    info!("getting spec from control plane: {}", cp_uri);
+    info!("getting config from control plane: {}", cp_uri);
 
     // Do 3 attempts to get spec from the control plane using the following logic:
     // - network error -> then retry
     // - compute id is unknown or any other error -> bail out
     // - no spec for compute yet (Empty state) -> return Ok(None)
-    // - got spec -> return Ok(Some(spec))
+    // - got config -> return Ok(Some(config))
     while attempt < 4 {
         let result = match do_control_plane_request(&cp_uri, &jwt) {
-            Ok(spec_resp) => {
+            Ok(config_resp) => {
                 CPLANE_REQUESTS_TOTAL
                     .with_label_values(&[
-                        CPlaneRequestRPC::GetSpec.as_str(),
+                        CPlaneRequestRPC::GetConfig.as_str(),
                         &StatusCode::OK.to_string(),
                     ])
                     .inc();
-                match spec_resp.status {
-                    ControlPlaneComputeStatus::Empty => Ok((None, spec_resp.compute_ctl_config)),
+                match config_resp.status {
+                    ControlPlaneComputeStatus::Empty => Ok(config_resp.into()),
                     ControlPlaneComputeStatus::Attached => {
-                        if let Some(spec) = spec_resp.spec {
-                            Ok((Some(spec), spec_resp.compute_ctl_config))
+                        if config_resp.spec.is_some() {
+                            Ok(config_resp.into())
                         } else {
                             bail!("compute is attached, but spec is empty")
                         }
@@ -111,7 +105,7 @@ pub fn get_spec_from_control_plane(
             }
             Err((retry, msg, status)) => {
                 CPLANE_REQUESTS_TOTAL
-                    .with_label_values(&[CPlaneRequestRPC::GetSpec.as_str(), &status])
+                    .with_label_values(&[CPlaneRequestRPC::GetConfig.as_str(), &status])
                     .inc();
                 if retry {
                     Err(anyhow!(msg))
@@ -122,7 +116,7 @@ pub fn get_spec_from_control_plane(
         };
 
         if let Err(e) = &result {
-            error!("attempt {} to get spec failed with: {}", attempt, e);
+            error!("attempt {} to get config failed with: {}", attempt, e);
         } else {
             return result;
         }
@@ -133,13 +127,13 @@ pub fn get_spec_from_control_plane(
 
     // All attempts failed, return error.
     Err(anyhow::anyhow!(
-        "Exhausted all attempts to retrieve the spec from the control plane"
+        "Exhausted all attempts to retrieve the config from the control plane"
     ))
 }
 
 /// Check `pg_hba.conf` and update if needed to allow external connections.
 pub fn update_pg_hba(pgdata_path: &Path) -> Result<()> {
-    // XXX: consider making it a part of spec.json
+    // XXX: consider making it a part of config.json
     let pghba_path = pgdata_path.join("pg_hba.conf");
 
     if config::line_in_file(&pghba_path, PG_HBA_ALL_MD5)? {
@@ -153,7 +147,7 @@ pub fn update_pg_hba(pgdata_path: &Path) -> Result<()> {
 
 /// Create a standby.signal file
 pub fn add_standby_signal(pgdata_path: &Path) -> Result<()> {
-    // XXX: consider making it a part of spec.json
+    // XXX: consider making it a part of config.json
     let signalfile = pgdata_path.join("standby.signal");
 
     if !signalfile.exists() {

compute_tools/src/spec_apply.rs

@@ -278,12 +278,12 @@ impl ComputeNode {
             // so that all config operations are audit logged.
             match spec.audit_log_level
             {
-                ComputeAudit::Hipaa => {
+                ComputeAudit::Hipaa | ComputeAudit::Extended | ComputeAudit::Full => {
                     phases.push(CreatePgauditExtension);
                     phases.push(CreatePgauditlogtofileExtension);
                     phases.push(DisablePostgresDBPgAudit);
                 }
-                ComputeAudit::Log => {
+                ComputeAudit::Log | ComputeAudit::Base => {
                     phases.push(CreatePgauditExtension);
                     phases.push(DisablePostgresDBPgAudit);
                 }

control_plane/Cargo.toml

@@ -6,13 +6,16 @@ license.workspace = true
 
 [dependencies]
 anyhow.workspace = true
+base64.workspace = true
 camino.workspace = true
 clap.workspace = true
 comfy-table.workspace = true
 futures.workspace = true
 humantime.workspace = true
+jsonwebtoken.workspace = true
 nix.workspace = true
 once_cell.workspace = true
+pem.workspace = true
 humantime-serde.workspace = true
 hyper0.workspace = true
 regex.workspace = true
@@ -20,6 +23,8 @@ reqwest = { workspace = true, features = ["blocking", "json"] }
 scopeguard.workspace = true
 serde.workspace = true
 serde_json.workspace = true
+sha2.workspace = true
+spki.workspace = true
 thiserror.workspace = true
 toml.workspace = true
 toml_edit.workspace = true

control_plane/src/bin/neon_local.rs

@@ -552,6 +552,7 @@ enum EndpointCmd {
     Start(EndpointStartCmdArgs),
     Reconfigure(EndpointReconfigureCmdArgs),
     Stop(EndpointStopCmdArgs),
+    GenerateJwt(EndpointGenerateJwtCmdArgs),
 }
 
 #[derive(clap::Args)]
@@ -699,6 +700,13 @@ struct EndpointStopCmdArgs {
     mode: String,
 }
 
+#[derive(clap::Args)]
+#[clap(about = "Generate a JWT for an endpoint")]
+struct EndpointGenerateJwtCmdArgs {
+    #[clap(help = "Postgres endpoint id")]
+    endpoint_id: String,
+}
+
 #[derive(clap::Subcommand)]
 #[clap(about = "Manage neon_local branch name mappings")]
 enum MappingsCmd {
@@ -1528,6 +1536,16 @@ async fn handle_endpoint(subcmd: &EndpointCmd, env: &local_env::LocalEnv) -> Res
                 .with_context(|| format!("postgres endpoint {endpoint_id} is not found"))?;
             endpoint.stop(&args.mode, args.destroy)?;
         }
+        EndpointCmd::GenerateJwt(args) => {
+            let endpoint_id = &args.endpoint_id;
+            let endpoint = cplane
+                .endpoints
+                .get(endpoint_id)
+                .with_context(|| format!("postgres endpoint {endpoint_id} is not found"))?;
+            let jwt = endpoint.generate_jwt()?;
+
+            println!("{jwt}");
+        }
     }
 
     Ok(())

control_plane/src/endpoint.rs

@@ -29,7 +29,7 @@
 //!     compute.log               - log output of `compute_ctl` and `postgres`
 //!     endpoint.json             - serialized `EndpointConf` struct
 //!     postgresql.conf           - postgresql settings
-//!     spec.json                 - passed to `compute_ctl`
+//!     config.json                 - passed to `compute_ctl`
 //!     pgdata/
 //!         postgresql.conf       - copy of postgresql.conf created by `compute_ctl`
 //!         zenith.signal
@@ -42,20 +42,30 @@ use std::path::PathBuf;
 use std::process::Command;
 use std::str::FromStr;
 use std::sync::Arc;
-use std::time::{Duration, Instant, SystemTime, UNIX_EPOCH};
+use std::time::{Duration, Instant};
 
 use anyhow::{Context, Result, anyhow, bail};
-use compute_api::requests::ConfigurationRequest;
-use compute_api::responses::{ComputeCtlConfig, ComputeStatus, ComputeStatusResponse};
+use compute_api::requests::{ComputeClaims, ConfigurationRequest};
+use compute_api::responses::{
+    ComputeConfig, ComputeCtlConfig, ComputeStatus, ComputeStatusResponse, TlsConfig,
+};
 use compute_api::spec::{
     Cluster, ComputeAudit, ComputeFeature, ComputeMode, ComputeSpec, Database, PgIdent,
     RemoteExtSpec, Role,
 };
+use jsonwebtoken::jwk::{
+    AlgorithmParameters, CommonParameters, EllipticCurve, Jwk, JwkSet, KeyAlgorithm, KeyOperations,
+    OctetKeyPairParameters, OctetKeyPairType, PublicKeyUse,
+};
 use nix::sys::signal::{Signal, kill};
 use pageserver_api::shard::ShardStripeSize;
+use pem::Pem;
 use reqwest::header::CONTENT_TYPE;
 use safekeeper_api::membership::SafekeeperGeneration;
 use serde::{Deserialize, Serialize};
+use sha2::{Digest, Sha256};
+use spki::der::Decode;
+use spki::{SubjectPublicKeyInfo, SubjectPublicKeyInfoRef};
 use tracing::debug;
 use url::Host;
 use utils::id::{NodeId, TenantId, TimelineId};
@@ -80,6 +90,7 @@ pub struct EndpointConf {
     drop_subscriptions_before_start: bool,
     features: Vec<ComputeFeature>,
     cluster: Option<Cluster>,
+    compute_ctl_config: ComputeCtlConfig,
 }
 
 //
@@ -135,6 +146,37 @@ impl ComputeControlPlane {
             .unwrap_or(self.base_port)
     }
 
+    /// Create a JSON Web Key Set. This ideally matches the way we create a JWKS
+    /// from the production control plane.
+    fn create_jwks_from_pem(pem: &Pem) -> Result<JwkSet> {
+        let spki: SubjectPublicKeyInfoRef = SubjectPublicKeyInfo::from_der(pem.contents())?;
+        let public_key = spki.subject_public_key.raw_bytes();
+
+        let mut hasher = Sha256::new();
+        hasher.update(public_key);
+        let key_hash = hasher.finalize();
+
+        Ok(JwkSet {
+            keys: vec![Jwk {
+                common: CommonParameters {
+                    public_key_use: Some(PublicKeyUse::Signature),
+                    key_operations: Some(vec![KeyOperations::Verify]),
+                    key_algorithm: Some(KeyAlgorithm::EdDSA),
+                    key_id: Some(base64::encode_config(key_hash, base64::URL_SAFE_NO_PAD)),
+                    x509_url: None::<String>,
+                    x509_chain: None::<Vec<String>>,
+                    x509_sha1_fingerprint: None::<String>,
+                    x509_sha256_fingerprint: None::<String>,
+                },
+                algorithm: AlgorithmParameters::OctetKeyPair(OctetKeyPairParameters {
+                    key_type: OctetKeyPairType::OctetKeyPair,
+                    curve: EllipticCurve::Ed25519,
+                    x: base64::encode_config(public_key, base64::URL_SAFE_NO_PAD),
+                }),
+            }],
+        })
+    }
+
     #[allow(clippy::too_many_arguments)]
     pub fn new_endpoint(
         &mut self,
@@ -152,6 +194,10 @@ impl ComputeControlPlane {
         let pg_port = pg_port.unwrap_or_else(|| self.get_port());
         let external_http_port = external_http_port.unwrap_or_else(|| self.get_port() + 1);
         let internal_http_port = internal_http_port.unwrap_or_else(|| external_http_port + 1);
+        let compute_ctl_config = ComputeCtlConfig {
+            jwks: Self::create_jwks_from_pem(&self.env.read_public_key()?)?,
+            tls: None::<TlsConfig>,
+        };
         let ep = Arc::new(Endpoint {
             endpoint_id: endpoint_id.to_owned(),
             pg_address: SocketAddr::new(IpAddr::from(Ipv4Addr::LOCALHOST), pg_port),
@@ -179,6 +225,7 @@ impl ComputeControlPlane {
             reconfigure_concurrency: 1,
             features: vec![],
             cluster: None,
+            compute_ctl_config: compute_ctl_config.clone(),
         });
 
         ep.create_endpoint_dir()?;
@@ -198,6 +245,7 @@ impl ComputeControlPlane {
                 reconfigure_concurrency: 1,
                 features: vec![],
                 cluster: None,
+                compute_ctl_config,
             })?,
         )?;
         std::fs::write(
@@ -240,7 +288,6 @@ impl ComputeControlPlane {
 
 ///////////////////////////////////////////////////////////////////////////////
 
-#[derive(Debug)]
 pub struct Endpoint {
     /// used as the directory name
     endpoint_id: String,
@@ -269,6 +316,9 @@ pub struct Endpoint {
     features: Vec<ComputeFeature>,
     // Cluster settings
     cluster: Option<Cluster>,
+
+    /// The compute_ctl config for the endpoint's compute.
+    compute_ctl_config: ComputeCtlConfig,
 }
 
 #[derive(PartialEq, Eq)]
@@ -331,6 +381,7 @@ impl Endpoint {
             drop_subscriptions_before_start: conf.drop_subscriptions_before_start,
             features: conf.features,
             cluster: conf.cluster,
+            compute_ctl_config: conf.compute_ctl_config,
         })
     }
 
@@ -578,6 +629,13 @@ impl Endpoint {
         Ok(safekeeper_connstrings)
     }
 
+    /// Generate a JWT with the correct claims.
+    pub fn generate_jwt(&self) -> Result<String> {
+        self.env.generate_auth_token(&ComputeClaims {
+            compute_id: self.endpoint_id.clone(),
+        })
+    }
+
     #[allow(clippy::too_many_arguments)]
     pub async fn start(
         &self,
@@ -619,87 +677,101 @@ impl Endpoint {
             remote_extensions = None;
         };
 
-        // Create spec file
-        let mut spec = ComputeSpec {
-            skip_pg_catalog_updates: self.skip_pg_catalog_updates,
-            format_version: 1.0,
-            operation_uuid: None,
-            features: self.features.clone(),
-            swap_size_bytes: None,
-            disk_quota_bytes: None,
-            disable_lfc_resizing: None,
-            cluster: Cluster {
-                cluster_id: None, // project ID: not used
-                name: None,       // project name: not used
-                state: None,
-                roles: if create_test_user {
-                    vec![Role {
+        // Create config file
+        let config = {
+            let mut spec = ComputeSpec {
+                skip_pg_catalog_updates: self.skip_pg_catalog_updates,
+                format_version: 1.0,
+                operation_uuid: None,
+                features: self.features.clone(),
+                swap_size_bytes: None,
+                disk_quota_bytes: None,
+                disable_lfc_resizing: None,
+                cluster: Cluster {
+                    cluster_id: None, // project ID: not used
+                    name: None,       // project name: not used
+                    state: None,
+                    roles: if create_test_user {
+                        vec![Role {
+                            name: PgIdent::from_str("test").unwrap(),
+                            encrypted_password: None,
+                            options: None,
+                        }]
+                    } else {
+                        Vec::new()
+                    },
+                    databases: if create_test_user {
+                        vec![Database {
+                            name: PgIdent::from_str("neondb").unwrap(),
+                            owner: PgIdent::from_str("test").unwrap(),
+                            options: None,
+                            restrict_conn: false,
+                            invalid: false,
+                        }]
+                    } else {
+                        Vec::new()
+                    },
+                    settings: None,
+                    postgresql_conf: Some(postgresql_conf.clone()),
+                },
+                delta_operations: None,
+                tenant_id: Some(self.tenant_id),
+                timeline_id: Some(self.timeline_id),
+                project_id: None,
+                branch_id: None,
+                endpoint_id: Some(self.endpoint_id.clone()),
+                mode: self.mode,
+                pageserver_connstring: Some(pageserver_connstring),
+                safekeepers_generation: safekeepers_generation.map(|g| g.into_inner()),
+                safekeeper_connstrings,
+                storage_auth_token: auth_token.clone(),
+                remote_extensions,
+                pgbouncer_settings: None,
+                shard_stripe_size: Some(shard_stripe_size),
+                local_proxy_config: None,
+                reconfigure_concurrency: self.reconfigure_concurrency,
+                drop_subscriptions_before_start: self.drop_subscriptions_before_start,
+                audit_log_level: ComputeAudit::Disabled,
+                logs_export_host: None::<String>,
+            };
+
+            // this strange code is needed to support respec() in tests
+            if self.cluster.is_some() {
+                debug!("Cluster is already set in the endpoint spec, using it");
+                spec.cluster = self.cluster.clone().unwrap();
+
+                debug!("spec.cluster {:?}", spec.cluster);
+
+                // fill missing fields again
+                if create_test_user {
+                    spec.cluster.roles.push(Role {
                         name: PgIdent::from_str("test").unwrap(),
                         encrypted_password: None,
                         options: None,
-                    }]
-                } else {
-                    Vec::new()
-                },
-                databases: if create_test_user {
-                    vec![Database {
+                    });
+                    spec.cluster.databases.push(Database {
                         name: PgIdent::from_str("neondb").unwrap(),
                         owner: PgIdent::from_str("test").unwrap(),
                         options: None,
                         restrict_conn: false,
                         invalid: false,
-                    }]
-                } else {
-                    Vec::new()
-                },
-                settings: None,
-                postgresql_conf: Some(postgresql_conf.clone()),
-            },
-            delta_operations: None,
-            tenant_id: Some(self.tenant_id),
-            timeline_id: Some(self.timeline_id),
-            mode: self.mode,
-            pageserver_connstring: Some(pageserver_connstring),
-            safekeepers_generation: safekeepers_generation.map(|g| g.into_inner()),
-            safekeeper_connstrings,
-            storage_auth_token: auth_token.clone(),
-            remote_extensions,
-            pgbouncer_settings: None,
-            shard_stripe_size: Some(shard_stripe_size),
-            local_proxy_config: None,
-            reconfigure_concurrency: self.reconfigure_concurrency,
-            drop_subscriptions_before_start: self.drop_subscriptions_before_start,
-            audit_log_level: ComputeAudit::Disabled,
-            logs_export_host: None::<String>,
+                    });
+                }
+                spec.cluster.postgresql_conf = Some(postgresql_conf);
+            }
+
+            ComputeConfig {
+                spec: Some(spec),
+                compute_ctl_config: self.compute_ctl_config.clone(),
+            }
         };
 
-        // this strange code is needed to support respec() in tests
-        if self.cluster.is_some() {
-            debug!("Cluster is already set in the endpoint spec, using it");
-            spec.cluster = self.cluster.clone().unwrap();
-
-            debug!("spec.cluster {:?}", spec.cluster);
-
-            // fill missing fields again
-            if create_test_user {
-                spec.cluster.roles.push(Role {
-                    name: PgIdent::from_str("test").unwrap(),
-                    encrypted_password: None,
-                    options: None,
-                });
-                spec.cluster.databases.push(Database {
-                    name: PgIdent::from_str("neondb").unwrap(),
-                    owner: PgIdent::from_str("test").unwrap(),
-                    options: None,
-                    restrict_conn: false,
-                    invalid: false,
-                });
-            }
-            spec.cluster.postgresql_conf = Some(postgresql_conf);
-        }
-
+        // TODO(tristan957): Remove the write to spec.json after compatibility
+        // tests work themselves out
         let spec_path = self.endpoint_path().join("spec.json");
-        std::fs::write(spec_path, serde_json::to_string_pretty(&spec)?)?;
+        std::fs::write(spec_path, serde_json::to_string_pretty(&config.spec)?)?;
+        let config_path = self.endpoint_path().join("config.json");
+        std::fs::write(config_path, serde_json::to_string_pretty(&config)?)?;
 
         // Open log file. We'll redirect the stdout and stderr of `compute_ctl` to it.
         let logfile = std::fs::OpenOptions::new()
@@ -707,6 +779,16 @@ impl Endpoint {
             .append(true)
             .open(self.endpoint_path().join("compute.log"))?;
 
+        // TODO(tristan957): Remove when compatibility tests are no longer an
+        // issue
+        let old_compute_ctl = {
+            let mut cmd = Command::new(self.env.neon_distrib_dir.join("compute_ctl"));
+            let help_output = cmd.arg("--help").output()?;
+            let help_output = String::from_utf8_lossy(&help_output.stdout);
+
+            !help_output.contains("--config")
+        };
+
         // Launch compute_ctl
         let conn_str = self.connstr("cloud_admin", "postgres");
         println!("Starting postgres node at '{}'", conn_str);
@@ -725,9 +807,18 @@ impl Endpoint {
         ])
         .args(["--pgdata", self.pgdata().to_str().unwrap()])
         .args(["--connstr", &conn_str])
+        // TODO(tristan957): Change this to --config when compatibility tests
+        // are no longer an issue
         .args([
             "--spec-path",
-            self.endpoint_path().join("spec.json").to_str().unwrap(),
+            self.endpoint_path()
+                .join(if old_compute_ctl {
+                    "spec.json"
+                } else {
+                    "config.json"
+                })
+                .to_str()
+                .unwrap(),
         ])
         .args([
             "--pgbin",
@@ -739,16 +830,7 @@ impl Endpoint {
         ])
         // TODO: It would be nice if we generated compute IDs with the same
         // algorithm as the real control plane.
-        .args([
-            "--compute-id",
-            &format!(
-                "compute-{}",
-                SystemTime::now()
-                    .duration_since(UNIX_EPOCH)
-                    .unwrap()
-                    .as_secs()
-            ),
-        ])
+        .args(["--compute-id", &self.endpoint_id])
         .stdin(std::process::Stdio::null())
         .stderr(logfile.try_clone()?)
         .stdout(logfile);
@@ -846,6 +928,7 @@ impl Endpoint {
                     self.external_http_address.port()
                 ),
             )
+            .bearer_auth(self.generate_jwt()?)
             .send()
             .await?;
 
@@ -870,10 +953,12 @@ impl Endpoint {
         stripe_size: Option<ShardStripeSize>,
         safekeepers: Option<Vec<NodeId>>,
     ) -> Result<()> {
-        let mut spec: ComputeSpec = {
-            let spec_path = self.endpoint_path().join("spec.json");
-            let file = std::fs::File::open(spec_path)?;
-            serde_json::from_reader(file)?
+        let (mut spec, compute_ctl_config) = {
+            let config_path = self.endpoint_path().join("config.json");
+            let file = std::fs::File::open(config_path)?;
+            let config: ComputeConfig = serde_json::from_reader(file)?;
+
+            (config.spec.unwrap(), config.compute_ctl_config)
         };
 
         let postgresql_conf = self.read_postgresql_conf()?;
@@ -920,10 +1005,11 @@ impl Endpoint {
                 self.external_http_address.port()
             ))
             .header(CONTENT_TYPE.as_str(), "application/json")
+            .bearer_auth(self.generate_jwt()?)
             .body(
                 serde_json::to_string(&ConfigurationRequest {
                     spec,
-                    compute_ctl_config: ComputeCtlConfig::default(),
+                    compute_ctl_config,
                 })
                 .unwrap(),
             )

control_plane/src/local_env.rs

@@ -12,6 +12,7 @@ use std::{env, fs};
 
 use anyhow::{Context, bail};
 use clap::ValueEnum;
+use pem::Pem;
 use postgres_backend::AuthType;
 use reqwest::Url;
 use serde::{Deserialize, Serialize};
@@ -56,6 +57,7 @@ pub struct LocalEnv {
 
     // used to issue tokens during e.g pg start
     pub private_key_path: PathBuf,
+    /// Path to environment's public key
     pub public_key_path: PathBuf,
 
     pub broker: NeonBroker,
@@ -758,11 +760,11 @@ impl LocalEnv {
 
     // this function is used only for testing purposes in CLI e g generate tokens during init
     pub fn generate_auth_token<S: Serialize>(&self, claims: &S) -> anyhow::Result<String> {
-        let private_key_path = self.get_private_key_path();
-        let key_data = fs::read(private_key_path)?;
-        encode_from_key_file(claims, &key_data)
+        let key = self.read_private_key()?;
+        encode_from_key_file(claims, &key)
     }
 
+    /// Get the path to the private key.
     pub fn get_private_key_path(&self) -> PathBuf {
         if self.private_key_path.is_absolute() {
             self.private_key_path.to_path_buf()
@@ -771,6 +773,29 @@ impl LocalEnv {
         }
     }
 
+    /// Get the path to the public key.
+    pub fn get_public_key_path(&self) -> PathBuf {
+        if self.public_key_path.is_absolute() {
+            self.public_key_path.to_path_buf()
+        } else {
+            self.base_data_dir.join(&self.public_key_path)
+        }
+    }
+
+    /// Read the contents of the private key file.
+    pub fn read_private_key(&self) -> anyhow::Result<Pem> {
+        let private_key_path = self.get_private_key_path();
+        let pem = pem::parse(fs::read(private_key_path)?)?;
+        Ok(pem)
+    }
+
+    /// Read the contents of the public key file.
+    pub fn read_public_key(&self) -> anyhow::Result<Pem> {
+        let public_key_path = self.get_public_key_path();
+        let pem = pem::parse(fs::read(public_key_path)?)?;
+        Ok(pem)
+    }
+
     /// Materialize the [`NeonLocalInitConf`] to disk. Called during [`neon_local init`].
     pub fn init(conf: NeonLocalInitConf, force: &InitForceMode) -> anyhow::Result<()> {
         let base_path = base_path();
@@ -956,6 +981,7 @@ fn generate_auth_keys(private_key_path: &Path, public_key_path: &Path) -> anyhow
             String::from_utf8_lossy(&keygen_output.stderr)
         );
     }
+
     // Extract the public key from the private key file
     //
     // openssl pkey -in auth_private_key.pem -pubout -out auth_public_key.pem
@@ -972,6 +998,7 @@ fn generate_auth_keys(private_key_path: &Path, public_key_path: &Path) -> anyhow
             String::from_utf8_lossy(&keygen_output.stderr)
         );
     }
+
     Ok(())
 }
 
@@ -980,7 +1007,7 @@ fn generate_ssl_ca_cert(cert_path: &Path, key_path: &Path) -> anyhow::Result<()>
     // -out rootCA.crt -keyout rootCA.key
     let keygen_output = Command::new("openssl")
         .args([
-            "req", "-x509", "-newkey", "rsa:2048", "-nodes", "-days", "36500",
+            "req", "-x509", "-newkey", "ed25519", "-nodes", "-days", "36500",
         ])
         .args(["-subj", "/CN=Neon Local CA"])
         .args(["-out", cert_path.to_str().unwrap()])
@@ -1010,7 +1037,7 @@ fn generate_ssl_cert(
     // -subj "/CN=localhost" -addext "subjectAltName=DNS:localhost,IP:127.0.0.1"
     let keygen_output = Command::new("openssl")
         .args(["req", "-new", "-nodes"])
-        .args(["-newkey", "rsa:2048"])
+        .args(["-newkey", "ed25519"])
         .args(["-subj", "/CN=localhost"])
         .args(["-addext", "subjectAltName=DNS:localhost,IP:127.0.0.1"])
         .args(["-keyout", key_path.to_str().unwrap()])

control_plane/src/pageserver.rs

@@ -413,6 +413,11 @@ impl PageServerNode {
                 .map(serde_json::from_str)
                 .transpose()
                 .context("Failed to parse 'compaction_algorithm' json")?,
+            compaction_shard_ancestor: settings
+                .remove("compaction_shard_ancestor")
+                .map(|x| x.parse::<bool>())
+                .transpose()
+                .context("Failed to parse 'compaction_shard_ancestor' as a bool")?,
             compaction_l0_first: settings
                 .remove("compaction_l0_first")
                 .map(|x| x.parse::<bool>())
@@ -535,6 +540,11 @@ impl PageServerNode {
                 .map(|x| x.parse::<bool>())
                 .transpose()
                 .context("Failed to parse 'gc_compaction_enabled' as bool")?,
+            gc_compaction_verification: settings
+                .remove("gc_compaction_verification")
+                .map(|x| x.parse::<bool>())
+                .transpose()
+                .context("Failed to parse 'gc_compaction_verification' as bool")?,
             gc_compaction_initial_threshold_kb: settings
                 .remove("gc_compaction_initial_threshold_kb")
                 .map(|x| x.parse::<u64>())

control_plane/src/storage_controller.rs

@@ -13,9 +13,12 @@ use pageserver_api::controller_api::{
     NodeConfigureRequest, NodeDescribeResponse, NodeRegisterRequest, TenantCreateRequest,
     TenantCreateResponse, TenantLocateResponse,
 };
-use pageserver_api::models::{TenantConfigRequest, TimelineCreateRequest, TimelineInfo};
+use pageserver_api::models::{
+    TenantConfig, TenantConfigRequest, TimelineCreateRequest, TimelineInfo,
+};
 use pageserver_api::shard::TenantShardId;
 use pageserver_client::mgmt_api::ResponseErrorMessageExt;
+use pem::Pem;
 use postgres_backend::AuthType;
 use reqwest::{Certificate, Method};
 use serde::de::DeserializeOwned;
@@ -32,8 +35,8 @@ use crate::local_env::{LocalEnv, NeonStorageControllerConf};
 
 pub struct StorageController {
     env: LocalEnv,
-    private_key: Option<Vec<u8>>,
-    public_key: Option<String>,
+    private_key: Option<Pem>,
+    public_key: Option<Pem>,
     client: reqwest::Client,
     config: NeonStorageControllerConf,
 
@@ -82,7 +85,8 @@ impl NeonStorageControllerStopArgs {
 pub struct AttachHookRequest {
     pub tenant_shard_id: TenantShardId,
     pub node_id: Option<NodeId>,
-    pub generation_override: Option<i32>,
+    pub generation_override: Option<i32>, // only new tenants
+    pub config: Option<TenantConfig>,     // only new tenants
 }
 
 #[derive(Serialize, Deserialize)]
@@ -113,7 +117,9 @@ impl StorageController {
             AuthType::Trust => (None, None),
             AuthType::NeonJWT => {
                 let private_key_path = env.get_private_key_path();
-                let private_key = fs::read(private_key_path).expect("failed to read private key");
+                let private_key =
+                    pem::parse(fs::read(private_key_path).expect("failed to read private key"))
+                        .expect("failed to parse PEM file");
 
                 // If pageserver auth is enabled, this implicitly enables auth for this service,
                 // using the same credentials.
@@ -135,9 +141,13 @@ impl StorageController {
                         .expect("Empty key dir")
                         .expect("Error reading key dir");
 
-                    std::fs::read_to_string(dent.path()).expect("Can't read public key")
+                    pem::parse(std::fs::read_to_string(dent.path()).expect("Can't read public key"))
+                        .expect("Failed to parse PEM file")
                 } else {
-                    std::fs::read_to_string(&public_key_path).expect("Can't read public key")
+                    pem::parse(
+                        std::fs::read_to_string(&public_key_path).expect("Can't read public key"),
+                    )
+                    .expect("Failed to parse PEM file")
                 };
                 (Some(private_key), Some(public_key))
             }
@@ -805,6 +815,7 @@ impl StorageController {
             tenant_shard_id,
             node_id: Some(pageserver_id),
             generation_override: None,
+            config: None,
         };
 
         let response = self

docker-compose/compute_wrapper/shell/compute.sh

@@ -11,8 +11,8 @@ generate_id() {
 
 PG_VERSION=${PG_VERSION:-14}
 
-SPEC_FILE_ORG=/var/db/postgres/specs/spec.json
-SPEC_FILE=/tmp/spec.json
+CONFIG_FILE_ORG=/var/db/postgres/configs/config.json
+CONFIG_FILE=/tmp/config.json
 
 echo "Waiting pageserver become ready."
 while ! nc -z pageserver 6400; do
@@ -20,7 +20,7 @@ while ! nc -z pageserver 6400; do
 done
 echo "Page server is ready."
 
-cp ${SPEC_FILE_ORG} ${SPEC_FILE}
+cp ${CONFIG_FILE_ORG} ${CONFIG_FILE}
 
  if [ -n "${TENANT_ID:-}" ] && [ -n "${TIMELINE_ID:-}" ]; then
    tenant_id=${TENANT_ID}
@@ -73,17 +73,27 @@ else
   ulid_extension=ulid
 fi
 echo "Adding pgx_ulid"
-shared_libraries=$(jq -r '.cluster.settings[] | select(.name=="shared_preload_libraries").value' ${SPEC_FILE})
-sed -i "s/${shared_libraries}/${shared_libraries},${ulid_extension}/" ${SPEC_FILE}
+shared_libraries=$(jq -r '.spec.cluster.settings[] | select(.name=="shared_preload_libraries").value' ${CONFIG_FILE})
+sed -i "s/${shared_libraries}/${shared_libraries},${ulid_extension}/" ${CONFIG_FILE}
 echo "Overwrite tenant id and timeline id in spec file"
-sed -i "s/TENANT_ID/${tenant_id}/" ${SPEC_FILE}
-sed -i "s/TIMELINE_ID/${timeline_id}/" ${SPEC_FILE}
+sed -i "s/TENANT_ID/${tenant_id}/" ${CONFIG_FILE}
+sed -i "s/TIMELINE_ID/${timeline_id}/" ${CONFIG_FILE}
 
-cat ${SPEC_FILE}
+cat ${CONFIG_FILE}
+
+# TODO(tristan957): Remove these workarounds for backwards compatibility after
+# the next compute release. That includes these next few lines and the
+# --spec-path in the compute_ctl invocation.
+if compute_ctl --help | grep --quiet -- '--config'; then
+  SPEC_PATH="$CONFIG_FILE"
+else
+  jq '.spec' < "$CONFIG_FILE" > /tmp/spec.json
+  SPEC_PATH=/tmp/spec.json
+fi
 
 echo "Start compute node"
 /usr/local/bin/compute_ctl --pgdata /var/db/postgres/compute \
      -C "postgresql://cloud_admin@localhost:55433/postgres"  \
      -b /usr/local/bin/postgres                              \
      --compute-id "compute-$RANDOM"                          \
-     -S ${SPEC_FILE}
+     --spec-path "$SPEC_PATH"

docker-compose/compute_wrapper/var/db/postgres/configs/config.json

@@ -0,0 +1,148 @@
+{
+    "spec": {
+        "format_version": 1.0,
+
+        "timestamp": "2022-10-12T18:00:00.000Z",
+        "operation_uuid": "0f657b36-4b0f-4a2d-9c2e-1dcd615e7d8c",
+
+        "cluster": {
+            "cluster_id": "docker_compose",
+            "name": "docker_compose_test",
+            "state": "restarted",
+            "roles": [
+                {
+                    "name": "cloud_admin",
+                    "encrypted_password": "b093c0d3b281ba6da1eacc608620abd8",
+                    "options": null
+                }
+            ],
+            "databases": [
+            ],
+            "settings": [
+                {
+                    "name": "fsync",
+                    "value": "off",
+                    "vartype": "bool"
+                },
+                {
+                    "name": "wal_level",
+                    "value": "logical",
+                    "vartype": "enum"
+                },
+                {
+                    "name": "wal_log_hints",
+                    "value": "on",
+                    "vartype": "bool"
+                },
+                {
+                    "name": "log_connections",
+                    "value": "on",
+                    "vartype": "bool"
+                },
+                {
+                    "name": "port",
+                    "value": "55433",
+                    "vartype": "integer"
+                },
+                {
+                    "name": "shared_buffers",
+                    "value": "1MB",
+                    "vartype": "string"
+                },
+                {
+                    "name": "max_connections",
+                    "value": "100",
+                    "vartype": "integer"
+                },
+                {
+                    "name": "listen_addresses",
+                    "value": "0.0.0.0",
+                    "vartype": "string"
+                },
+                {
+                    "name": "max_wal_senders",
+                    "value": "10",
+                    "vartype": "integer"
+                },
+                {
+                    "name": "max_replication_slots",
+                    "value": "10",
+                    "vartype": "integer"
+                },
+                {
+                    "name": "wal_sender_timeout",
+                    "value": "5s",
+                    "vartype": "string"
+                },
+                {
+                    "name": "wal_keep_size",
+                    "value": "0",
+                    "vartype": "integer"
+                },
+                {
+                    "name": "password_encryption",
+                    "value": "md5",
+                    "vartype": "enum"
+                },
+                {
+                    "name": "restart_after_crash",
+                    "value": "off",
+                    "vartype": "bool"
+                },
+                {
+                    "name": "synchronous_standby_names",
+                    "value": "walproposer",
+                    "vartype": "string"
+                },
+                {
+                    "name": "shared_preload_libraries",
+                    "value": "neon,pg_cron,timescaledb,pg_stat_statements",
+                    "vartype": "string"
+                },
+                {
+                    "name": "neon.safekeepers",
+                    "value": "safekeeper1:5454,safekeeper2:5454,safekeeper3:5454",
+                    "vartype": "string"
+                },
+                {
+                    "name": "neon.timeline_id",
+                    "value": "TIMELINE_ID",
+                    "vartype": "string"
+                },
+                {
+                    "name": "neon.tenant_id",
+                    "value": "TENANT_ID",
+                    "vartype": "string"
+                },
+                {
+                    "name": "neon.pageserver_connstring",
+                    "value": "host=pageserver port=6400",
+                    "vartype": "string"
+                },
+                {
+                    "name": "max_replication_write_lag",
+                    "value": "500MB",
+                    "vartype": "string"
+                },
+                {
+                    "name": "max_replication_flush_lag",
+                    "value": "10GB",
+                    "vartype": "string"
+                },
+                {
+                    "name": "cron.database",
+                    "value": "postgres",
+                    "vartype": "string"
+                }
+            ]
+        },
+
+        "delta_operations": [
+        ]
+    },
+    "compute_ctl_config": {
+        "jwks": {
+            "keys": []
+        }
+    }
+}

docker-compose/compute_wrapper/var/db/postgres/specs/spec.json

@@ -1,141 +0,0 @@
-{
-    "format_version": 1.0,
-
-    "timestamp": "2022-10-12T18:00:00.000Z",
-    "operation_uuid": "0f657b36-4b0f-4a2d-9c2e-1dcd615e7d8c",
-
-    "cluster": {
-        "cluster_id": "docker_compose",
-        "name": "docker_compose_test",
-        "state": "restarted",
-        "roles": [
-            {
-                "name": "cloud_admin",
-                "encrypted_password": "b093c0d3b281ba6da1eacc608620abd8",
-                "options": null
-            }
-        ],
-        "databases": [
-        ],
-        "settings": [
-            {
-                "name": "fsync",
-                "value": "off",
-                "vartype": "bool"
-            },
-            {
-                "name": "wal_level",
-                "value": "logical",
-                "vartype": "enum"
-            },
-            {
-                "name": "wal_log_hints",
-                "value": "on",
-                "vartype": "bool"
-            },
-            {
-                "name": "log_connections",
-                "value": "on",
-                "vartype": "bool"
-            },
-            {
-                "name": "port",
-                "value": "55433",
-                "vartype": "integer"
-            },
-            {
-                "name": "shared_buffers",
-                "value": "1MB",
-                "vartype": "string"
-            },
-            {
-                "name": "max_connections",
-                "value": "100",
-                "vartype": "integer"
-            },
-            {
-                "name": "listen_addresses",
-                "value": "0.0.0.0",
-                "vartype": "string"
-            },
-            {
-                "name": "max_wal_senders",
-                "value": "10",
-                "vartype": "integer"
-            },
-            {
-                "name": "max_replication_slots",
-                "value": "10",
-                "vartype": "integer"
-            },
-            {
-                "name": "wal_sender_timeout",
-                "value": "5s",
-                "vartype": "string"
-            },
-            {
-                "name": "wal_keep_size",
-                "value": "0",
-                "vartype": "integer"
-            },
-            {
-                "name": "password_encryption",
-                "value": "md5",
-                "vartype": "enum"
-            },
-            {
-                "name": "restart_after_crash",
-                "value": "off",
-                "vartype": "bool"
-            },
-            {
-                "name": "synchronous_standby_names",
-                "value": "walproposer",
-                "vartype": "string"
-            },
-            {
-                "name": "shared_preload_libraries",
-                "value": "neon,pg_cron,timescaledb,pg_stat_statements",
-                "vartype": "string"
-            },
-            {
-                "name": "neon.safekeepers",
-                "value": "safekeeper1:5454,safekeeper2:5454,safekeeper3:5454",
-                "vartype": "string"
-            },
-            {
-                "name": "neon.timeline_id",
-                "value": "TIMELINE_ID",
-                "vartype": "string"
-            },
-            {
-                "name": "neon.tenant_id",
-                "value": "TENANT_ID",
-                "vartype": "string"
-            },
-            {
-                "name": "neon.pageserver_connstring",
-                "value": "host=pageserver port=6400",
-                "vartype": "string"
-            },
-            {
-                "name": "max_replication_write_lag",
-                "value": "500MB",
-                "vartype": "string"
-            },
-            {
-                "name": "max_replication_flush_lag",
-                "value": "10GB",
-                "vartype": "string"
-            },
-            {
-                "name": "cron.database",
-                "value": "postgres",
-                "vartype": "string"
-            }
-        ]
-    },
-
-    "delta_operations": [
-    ]
-}

docker-compose/docker-compose.yml

@@ -159,7 +159,7 @@ services:
       #- RUST_BACKTRACE=1
     # Mount the test files directly, for faster editing cycle.
     volumes:
-      - ./compute_wrapper/var/db/postgres/specs/:/var/db/postgres/specs/
+      - ./compute_wrapper/var/db/postgres/configs/:/var/db/postgres/configs/
       - ./compute_wrapper/shell/:/shell/
     ports:
       - 55433:55433 # pg protocol handler

docker-compose/ext-src/pg_jsonschema-src/Makefile

@@ -0,0 +1,8 @@
+EXTENSION = pg_jsonschema
+DATA = pg_jsonschema--1.0.sql
+REGRESS = jsonschema_valid_api  jsonschema_edge_cases
+REGRESS_OPTS = --load-extension=pg_jsonschema
+
+PG_CONFIG ?= pg_config
+PGXS := $(shell $(PG_CONFIG) --pgxs)
+include $(PGXS)

docker-compose/ext-src/pg_jsonschema-src/expected/jsonschema_edge_cases.out

@@ -0,0 +1,87 @@
+-- Schema with enums, nulls, extra properties disallowed
+SELECT jsonschema_is_valid('{
+  "type": "object",
+  "properties": {
+    "status": { "type": "string", "enum": ["active", "inactive", "pending"] },
+    "email": { "type": ["string", "null"], "format": "email" }
+  },
+  "required": ["status"],
+  "additionalProperties": false
+}'::json);
+ jsonschema_is_valid 
+---------------------
+ t
+(1 row)
+
+-- Valid enum and null email
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "status": { "type": "string", "enum": ["active", "inactive", "pending"] },
+      "email": { "type": ["string", "null"], "format": "email" }
+    },
+    "required": ["status"],
+    "additionalProperties": false
+  }'::json,
+  '{"status": "active", "email": null}'::json
+);
+ jsonschema_validation_errors 
+------------------------------
+ {}
+(1 row)
+
+-- Invalid enum value
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "status": { "type": "string", "enum": ["active", "inactive", "pending"] },
+      "email": { "type": ["string", "null"], "format": "email" }
+    },
+    "required": ["status"],
+    "additionalProperties": false
+  }'::json,
+  '{"status": "disabled", "email": null}'::json
+);
+                     jsonschema_validation_errors                     
+----------------------------------------------------------------------
+ {"\"disabled\" is not one of [\"active\",\"inactive\",\"pending\"]"}
+(1 row)
+
+-- Invalid email format (assuming format is validated)
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "status": { "type": "string", "enum": ["active", "inactive", "pending"] },
+      "email": { "type": ["string", "null"], "format": "email" }
+    },
+    "required": ["status"],
+    "additionalProperties": false
+  }'::json,
+  '{"status": "active", "email": "not-an-email"}'::json
+);
+      jsonschema_validation_errors       
+-----------------------------------------
+ {"\"not-an-email\" is not a \"email\""}
+(1 row)
+
+-- Extra property not allowed
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "status": { "type": "string", "enum": ["active", "inactive", "pending"] },
+      "email": { "type": ["string", "null"], "format": "email" }
+    },
+    "required": ["status"],
+    "additionalProperties": false
+  }'::json,
+  '{"status": "active", "extra": "should not be here"}'::json
+);
+                    jsonschema_validation_errors                    
+--------------------------------------------------------------------
+ {"Additional properties are not allowed ('extra' was unexpected)"}
+(1 row)
+

docker-compose/ext-src/pg_jsonschema-src/expected/jsonschema_valid_api.out

@@ -0,0 +1,65 @@
+-- Define schema
+SELECT jsonschema_is_valid('{
+  "type": "object",
+  "properties": {
+    "username": { "type": "string" },
+    "age": { "type": "integer" }
+  },
+  "required": ["username"]
+}'::json);
+ jsonschema_is_valid 
+---------------------
+ t
+(1 row)
+
+-- Valid instance
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "username": { "type": "string" },
+      "age": { "type": "integer" }
+    },
+    "required": ["username"]
+  }'::json,
+  '{"username": "alice", "age": 25}'::json
+);
+ jsonschema_validation_errors 
+------------------------------
+ {}
+(1 row)
+
+-- Invalid instance: missing required "username"
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "username": { "type": "string" },
+      "age": { "type": "integer" }
+    },
+    "required": ["username"]
+  }'::json,
+  '{"age": 25}'::json
+);
+      jsonschema_validation_errors       
+-----------------------------------------
+ {"\"username\" is a required property"}
+(1 row)
+
+-- Invalid instance: wrong type for "age"
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "username": { "type": "string" },
+      "age": { "type": "integer" }
+    },
+    "required": ["username"]
+  }'::json,
+  '{"username": "bob", "age": "twenty"}'::json
+);
+       jsonschema_validation_errors        
+-------------------------------------------
+ {"\"twenty\" is not of type \"integer\""}
+(1 row)
+

docker-compose/ext-src/pg_jsonschema-src/sql/jsonschema_edge_cases.sql

@@ -0,0 +1,66 @@
+-- Schema with enums, nulls, extra properties disallowed
+SELECT jsonschema_is_valid('{
+  "type": "object",
+  "properties": {
+    "status": { "type": "string", "enum": ["active", "inactive", "pending"] },
+    "email": { "type": ["string", "null"], "format": "email" }
+  },
+  "required": ["status"],
+  "additionalProperties": false
+}'::json);
+
+-- Valid enum and null email
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "status": { "type": "string", "enum": ["active", "inactive", "pending"] },
+      "email": { "type": ["string", "null"], "format": "email" }
+    },
+    "required": ["status"],
+    "additionalProperties": false
+  }'::json,
+  '{"status": "active", "email": null}'::json
+);
+
+-- Invalid enum value
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "status": { "type": "string", "enum": ["active", "inactive", "pending"] },
+      "email": { "type": ["string", "null"], "format": "email" }
+    },
+    "required": ["status"],
+    "additionalProperties": false
+  }'::json,
+  '{"status": "disabled", "email": null}'::json
+);
+
+-- Invalid email format (assuming format is validated)
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "status": { "type": "string", "enum": ["active", "inactive", "pending"] },
+      "email": { "type": ["string", "null"], "format": "email" }
+    },
+    "required": ["status"],
+    "additionalProperties": false
+  }'::json,
+  '{"status": "active", "email": "not-an-email"}'::json
+);
+
+-- Extra property not allowed
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "status": { "type": "string", "enum": ["active", "inactive", "pending"] },
+      "email": { "type": ["string", "null"], "format": "email" }
+    },
+    "required": ["status"],
+    "additionalProperties": false
+  }'::json,
+  '{"status": "active", "extra": "should not be here"}'::json
+);

docker-compose/ext-src/pg_jsonschema-src/sql/jsonschema_valid_api.sql

@@ -0,0 +1,48 @@
+-- Define schema
+SELECT jsonschema_is_valid('{
+  "type": "object",
+  "properties": {
+    "username": { "type": "string" },
+    "age": { "type": "integer" }
+  },
+  "required": ["username"]
+}'::json);
+
+-- Valid instance
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "username": { "type": "string" },
+      "age": { "type": "integer" }
+    },
+    "required": ["username"]
+  }'::json,
+  '{"username": "alice", "age": 25}'::json
+);
+
+-- Invalid instance: missing required "username"
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "username": { "type": "string" },
+      "age": { "type": "integer" }
+    },
+    "required": ["username"]
+  }'::json,
+  '{"age": 25}'::json
+);
+
+-- Invalid instance: wrong type for "age"
+SELECT jsonschema_validation_errors(
+  '{
+    "type": "object",
+    "properties": {
+      "username": { "type": "string" },
+      "age": { "type": "integer" }
+    },
+    "required": ["username"]
+  }'::json,
+  '{"username": "bob", "age": "twenty"}'::json
+);

docker-compose/ext-src/pg_session_jwt-src/Makefile

@@ -0,0 +1,9 @@
+EXTENSION = pg_session_jwt
+
+REGRESS = basic_functions
+REGRESS_OPTS = --load-extension=$(EXTENSION)
+export PGOPTIONS = -c pg_session_jwt.jwk={"crv":"Ed25519","kty":"OKP","x":"R_Abz-63zJ00l-IraL5fQhwkhGVZCSooQFV5ntC3C7M"}
+
+PG_CONFIG ?= pg_config
+PGXS := $(shell $(PG_CONFIG) --pgxs)
+include $(PGXS)

docker-compose/ext-src/pg_session_jwt-src/expected/basic_functions.out

@@ -0,0 +1,35 @@
+-- Basic functionality tests for pg_session_jwt
+-- Test auth.init() function
+SELECT auth.init();
+ init 
+------
+ 
+(1 row)
+
+-- Test an invalid JWT
+SELECT auth.jwt_session_init('INVALID-JWT');
+ERROR:  invalid JWT encoding
+-- Test creating a session with an expired JWT
+SELECT auth.jwt_session_init('eyJhbGciOiJFZERTQSJ9.eyJleHAiOjE3NDI1NjQ0MzIsImlhdCI6MTc0MjU2NDI1MiwianRpIjo0MjQyNDIsInN1YiI6InVzZXIxMjMifQ.A6FwKuaSduHB9O7Gz37g0uoD_U9qVS0JNtT7YABGVgB7HUD1AMFc9DeyhNntWBqncg8k5brv-hrNTuUh5JYMAw');
+ERROR:  Token used after it has expired
+-- Test creating a session with a valid JWT
+SELECT auth.jwt_session_init('eyJhbGciOiJFZERTQSJ9.eyJleHAiOjQ4OTYxNjQyNTIsImlhdCI6MTc0MjU2NDI1MiwianRpIjo0MzQzNDMsInN1YiI6InVzZXIxMjMifQ.2TXVgjb6JSUq6_adlvp-m_SdOxZSyGS30RS9TLB0xu2N83dMSs2NybwE1NMU8Fb0tcAZR_ET7M2rSxbTrphfCg');
+ jwt_session_init 
+------------------
+ 
+(1 row)
+
+-- Test auth.session() function
+SELECT auth.session();
+                                 session                                 
+-------------------------------------------------------------------------
+ {"exp": 4896164252, "iat": 1742564252, "jti": 434343, "sub": "user123"}
+(1 row)
+
+-- Test auth.user_id() function
+SELECT auth.user_id() AS user_id;
+ user_id 
+---------
+ user123
+(1 row)
+

docker-compose/ext-src/pg_session_jwt-src/sql/basic_functions.sql

@@ -0,0 +1,19 @@
+-- Basic functionality tests for pg_session_jwt
+
+-- Test auth.init() function
+SELECT auth.init();
+
+-- Test an invalid JWT
+SELECT auth.jwt_session_init('INVALID-JWT');
+
+-- Test creating a session with an expired JWT
+SELECT auth.jwt_session_init('eyJhbGciOiJFZERTQSJ9.eyJleHAiOjE3NDI1NjQ0MzIsImlhdCI6MTc0MjU2NDI1MiwianRpIjo0MjQyNDIsInN1YiI6InVzZXIxMjMifQ.A6FwKuaSduHB9O7Gz37g0uoD_U9qVS0JNtT7YABGVgB7HUD1AMFc9DeyhNntWBqncg8k5brv-hrNTuUh5JYMAw');
+
+-- Test creating a session with a valid JWT
+SELECT auth.jwt_session_init('eyJhbGciOiJFZERTQSJ9.eyJleHAiOjQ4OTYxNjQyNTIsImlhdCI6MTc0MjU2NDI1MiwianRpIjo0MzQzNDMsInN1YiI6InVzZXIxMjMifQ.2TXVgjb6JSUq6_adlvp-m_SdOxZSyGS30RS9TLB0xu2N83dMSs2NybwE1NMU8Fb0tcAZR_ET7M2rSxbTrphfCg');
+
+-- Test auth.session() function
+SELECT auth.session();
+
+-- Test auth.user_id() function
+SELECT auth.user_id() AS user_id;

libs/compute_api/src/responses.rs

@@ -14,6 +14,32 @@ pub struct GenericAPIError {
     pub error: String,
 }
 
+/// All configuration parameters necessary for a compute. When
+/// [`ComputeConfig::spec`] is provided, it means that the compute is attached
+/// to a tenant. [`ComputeConfig::compute_ctl_config`] will always be provided
+/// and contains parameters necessary for operating `compute_ctl` independently
+/// of whether a tenant is attached to the compute or not.
+///
+/// This also happens to be the body of `compute_ctl`'s /configure request.
+#[derive(Debug, Deserialize, Serialize)]
+pub struct ComputeConfig {
+    /// The compute spec
+    pub spec: Option<ComputeSpec>,
+
+    /// The compute_ctl configuration
+    #[allow(dead_code)]
+    pub compute_ctl_config: ComputeCtlConfig,
+}
+
+impl From<ControlPlaneConfigResponse> for ComputeConfig {
+    fn from(value: ControlPlaneConfigResponse) -> Self {
+        Self {
+            spec: value.spec,
+            compute_ctl_config: value.compute_ctl_config,
+        }
+    }
+}
+
 #[derive(Debug, Clone, Serialize)]
 pub struct ExtensionInstallResponse {
     pub extension: PgIdent,
@@ -134,7 +160,7 @@ pub struct CatalogObjects {
     pub databases: Vec<Database>,
 }
 
-#[derive(Clone, Debug, Deserialize, Serialize)]
+#[derive(Clone, Debug, Deserialize, Eq, PartialEq, Serialize)]
 pub struct ComputeCtlConfig {
     /// Set of JSON web keys that the compute can use to authenticate
     /// communication from the control plane.
@@ -153,7 +179,7 @@ impl Default for ComputeCtlConfig {
     }
 }
 
-#[derive(Clone, Debug, Deserialize, Serialize)]
+#[derive(Clone, Debug, Deserialize, Eq, PartialEq, Serialize)]
 pub struct TlsConfig {
     pub key_path: String,
     pub cert_path: String,
@@ -161,7 +187,7 @@ pub struct TlsConfig {
 
 /// Response of the `/computes/{compute_id}/spec` control-plane API.
 #[derive(Deserialize, Debug)]
-pub struct ControlPlaneSpecResponse {
+pub struct ControlPlaneConfigResponse {
     pub spec: Option<ComputeSpec>,
     pub status: ControlPlaneComputeStatus,
     pub compute_ctl_config: ComputeCtlConfig,

libs/compute_api/src/spec.rs

@@ -1,8 +1,8 @@
-//! `ComputeSpec` represents the contents of the spec.json file.
-//!
-//! The spec.json file is used to pass information to 'compute_ctl'. It contains
-//! all the information needed to start up the right version of PostgreSQL,
-//! and connect it to the storage nodes.
+//! The ComputeSpec contains all the information needed to start up
+//! the right version of PostgreSQL, and connect it to the storage nodes.
+//! It can be passed as part of the `config.json`, or the control plane can
+//! provide it by calling the compute_ctl's `/compute_ctl` endpoint, or
+//! compute_ctl can fetch it by calling the control plane's API.
 use std::collections::HashMap;
 
 use indexmap::IndexMap;
@@ -104,6 +104,12 @@ pub struct ComputeSpec {
     pub timeline_id: Option<TimelineId>,
     pub pageserver_connstring: Option<String>,
 
+    // More neon ids that we expose to the compute_ctl
+    // and to postgres as neon extension GUCs.
+    pub project_id: Option<String>,
+    pub branch_id: Option<String>,
+    pub endpoint_id: Option<String>,
+
     /// Safekeeper membership config generation. It is put in
     /// neon.safekeepers GUC and serves two purposes:
     /// 1) Non zero value forces walproposer to use membership configurations.
@@ -159,13 +165,7 @@ pub struct ComputeSpec {
     #[serde(default)] // Default false
     pub drop_subscriptions_before_start: bool,
 
-    /// Log level for audit logging:
-    ///
-    /// Disabled - no audit logging. This is the default.
-    /// log - log masked statements to the postgres log using pgaudit extension
-    /// hipaa - log unmasked statements to the file using pgaudit and pgauditlogtofile extension
-    ///
-    /// Extensions should be present in shared_preload_libraries
+    /// Log level for compute audit logging
     #[serde(default)]
     pub audit_log_level: ComputeAudit,
 
@@ -289,14 +289,25 @@ impl ComputeMode {
 }
 
 /// Log level for audit logging
-/// Disabled, log, hipaa
-/// Default is Disabled
 #[derive(Clone, Debug, Default, Eq, PartialEq, Deserialize, Serialize)]
 pub enum ComputeAudit {
     #[default]
     Disabled,
+    // Deprecated, use Base instead
     Log,
+    // (pgaudit.log = 'ddl', pgaudit.log_parameter='off')
+    // logged to the standard postgresql log stream
+    Base,
+    // Deprecated, use Full or Extended instead
     Hipaa,
+    // (pgaudit.log = 'all, -misc', pgaudit.log_parameter='off')
+    // logged to separate files collected by rsyslog
+    // into dedicated log storage with strict access
+    Extended,
+    // (pgaudit.log='all', pgaudit.log_parameter='on'),
+    // logged to separate files collected by rsyslog
+    // into dedicated log storage with strict access.
+    Full,
 }
 
 #[derive(Clone, Debug, Default, Deserialize, Serialize, PartialEq, Eq)]

libs/http-utils/Cargo.toml

@@ -14,6 +14,7 @@ futures.workspace = true
 hyper0.workspace = true
 itertools.workspace = true
 jemalloc_pprof.workspace = true
+jsonwebtoken.workspace = true
 once_cell.workspace = true
 pprof.workspace = true
 regex.workspace = true
@@ -30,6 +31,7 @@ tokio.workspace = true
 tracing.workspace = true
 url.workspace = true
 uuid.workspace = true
+x509-cert.workspace = true
 
 # to use tokio channels as streams, this is faster to compile than async_stream
 # why is it only here? no other crate should use it, streams are rarely needed.

libs/http-utils/src/endpoint.rs

@@ -8,6 +8,7 @@ use bytes::{Bytes, BytesMut};
 use hyper::header::{AUTHORIZATION, CONTENT_DISPOSITION, CONTENT_TYPE, HeaderName};
 use hyper::http::HeaderValue;
 use hyper::{Body, Method, Request, Response};
+use jsonwebtoken::TokenData;
 use metrics::{Encoder, IntCounter, TextEncoder, register_int_counter};
 use once_cell::sync::Lazy;
 use pprof::ProfilerGuardBuilder;
@@ -618,7 +619,7 @@ pub fn auth_middleware<B: hyper::body::HttpBody + Send + Sync + 'static>(
                     })?;
                     let token = parse_token(header_value)?;
 
-                    let data = auth.decode(token).map_err(|err| {
+                    let data: TokenData<Claims> = auth.decode(token).map_err(|err| {
                         warn!("Authentication error: {err}");
                         // Rely on From<AuthError> for ApiError impl
                         err

libs/http-utils/src/server.rs

@@ -4,6 +4,8 @@ use futures::StreamExt;
 use futures::stream::FuturesUnordered;
 use hyper0::Body;
 use hyper0::server::conn::Http;
+use metrics::{IntCounterVec, register_int_counter_vec};
+use once_cell::sync::Lazy;
 use routerify::{RequestService, RequestServiceBuilder};
 use tokio::io::{AsyncRead, AsyncWrite};
 use tokio_rustls::TlsAcceptor;
@@ -26,6 +28,24 @@ pub struct Server {
     tls_acceptor: Option<TlsAcceptor>,
 }
 
+static CONNECTION_STARTED_COUNT: Lazy<IntCounterVec> = Lazy::new(|| {
+    register_int_counter_vec!(
+        "http_server_connection_started_total",
+        "Number of established http/https connections",
+        &["scheme"]
+    )
+    .expect("failed to define a metric")
+});
+
+static CONNECTION_ERROR_COUNT: Lazy<IntCounterVec> = Lazy::new(|| {
+    register_int_counter_vec!(
+        "http_server_connection_errors_total",
+        "Number of occured connection errors by type",
+        &["type"]
+    )
+    .expect("failed to define a metric")
+});
+
 impl Server {
     pub fn new(
         request_service: Arc<RequestServiceBuilder<Body, ApiError>>,
@@ -60,6 +80,15 @@ impl Server {
             false
         }
 
+        let tcp_error_cnt = CONNECTION_ERROR_COUNT.with_label_values(&["tcp"]);
+        let tls_error_cnt = CONNECTION_ERROR_COUNT.with_label_values(&["tls"]);
+        let http_error_cnt = CONNECTION_ERROR_COUNT.with_label_values(&["http"]);
+        let https_error_cnt = CONNECTION_ERROR_COUNT.with_label_values(&["https"]);
+        let panic_error_cnt = CONNECTION_ERROR_COUNT.with_label_values(&["panic"]);
+
+        let http_connection_cnt = CONNECTION_STARTED_COUNT.with_label_values(&["http"]);
+        let https_connection_cnt = CONNECTION_STARTED_COUNT.with_label_values(&["https"]);
+
         let mut connections = FuturesUnordered::new();
         loop {
             tokio::select! {
@@ -67,6 +96,7 @@ impl Server {
                     let (tcp_stream, remote_addr) = match stream {
                         Ok(stream) => stream,
                         Err(err) => {
+                            tcp_error_cnt.inc();
                             if !suppress_io_error(&err) {
                                 info!("Failed to accept TCP connection: {err:#}");
                             }
@@ -78,11 +108,18 @@ impl Server {
                     let tls_acceptor = self.tls_acceptor.clone();
                     let cancel = cancel.clone();
 
+                    let tls_error_cnt = tls_error_cnt.clone();
+                    let http_error_cnt = http_error_cnt.clone();
+                    let https_error_cnt = https_error_cnt.clone();
+                    let http_connection_cnt = http_connection_cnt.clone();
+                    let https_connection_cnt = https_connection_cnt.clone();
+
                     connections.push(tokio::spawn(
                         async move {
                             match tls_acceptor {
                                 Some(tls_acceptor) => {
                                     // Handle HTTPS connection.
+                                    https_connection_cnt.inc();
                                     let tls_stream = tokio::select! {
                                         tls_stream = tls_acceptor.accept(tcp_stream) => tls_stream,
                                         _ = cancel.cancelled() => return,
@@ -90,6 +127,7 @@ impl Server {
                                     let tls_stream = match tls_stream {
                                         Ok(tls_stream) => tls_stream,
                                         Err(err) => {
+                                            tls_error_cnt.inc();
                                             if !suppress_io_error(&err) {
                                                 info!(%remote_addr, "Failed to accept TLS connection: {err:#}");
                                             }
@@ -97,6 +135,7 @@ impl Server {
                                         }
                                     };
                                     if let Err(err) = Self::serve_connection(tls_stream, service, cancel).await {
+                                        https_error_cnt.inc();
                                         if !suppress_hyper_error(&err) {
                                             info!(%remote_addr, "Failed to serve HTTPS connection: {err:#}");
                                         }
@@ -104,7 +143,9 @@ impl Server {
                                 }
                                 None => {
                                     // Handle HTTP connection.
+                                    http_connection_cnt.inc();
                                     if let Err(err) = Self::serve_connection(tcp_stream, service, cancel).await {
+                                        http_error_cnt.inc();
                                         if !suppress_hyper_error(&err) {
                                             info!(%remote_addr, "Failed to serve HTTP connection: {err:#}");
                                         }
@@ -115,6 +156,7 @@ impl Server {
                  }
                 Some(conn) = connections.next() => {
                     if let Err(err) = conn {
+                        panic_error_cnt.inc();
                         error!("Connection panicked: {err:#}");
                     }
                 }
@@ -122,6 +164,7 @@ impl Server {
                     // Wait for graceful shutdown of all connections.
                     while let Some(conn) = connections.next().await {
                         if let Err(err) = conn {
+                            panic_error_cnt.inc();
                             error!("Connection panicked: {err:#}");
                         }
                     }

libs/http-utils/src/tls_certs.rs

@@ -3,11 +3,14 @@ use std::{sync::Arc, time::Duration};
 use anyhow::Context;
 use arc_swap::ArcSwap;
 use camino::Utf8Path;
+use metrics::{IntCounterVec, UIntGaugeVec, register_int_counter_vec, register_uint_gauge_vec};
+use once_cell::sync::Lazy;
 use rustls::{
-    pki_types::{CertificateDer, PrivateKeyDer},
+    pki_types::{CertificateDer, PrivateKeyDer, UnixTime},
     server::{ClientHello, ResolvesServerCert},
     sign::CertifiedKey,
 };
+use x509_cert::der::Reader;
 
 pub async fn load_cert_chain(filename: &Utf8Path) -> anyhow::Result<Vec<CertificateDer<'static>>> {
     let cert_data = tokio::fs::read(filename)
@@ -53,6 +56,76 @@ pub async fn load_certified_key(
     Ok(certified_key)
 }
 
+/// rustls's CertifiedKey with extra parsed fields used for metrics.
+struct ParsedCertifiedKey {
+    certified_key: CertifiedKey,
+    expiration_time: UnixTime,
+}
+
+/// Parse expiration time from an X509 certificate.
+fn parse_expiration_time(cert: &CertificateDer<'_>) -> anyhow::Result<UnixTime> {
+    let parsed_cert = x509_cert::der::SliceReader::new(cert)
+        .context("Failed to parse cerficiate")?
+        .decode::<x509_cert::Certificate>()
+        .context("Failed to parse cerficiate")?;
+
+    Ok(UnixTime::since_unix_epoch(
+        parsed_cert
+            .tbs_certificate
+            .validity
+            .not_after
+            .to_unix_duration(),
+    ))
+}
+
+async fn load_and_parse_certified_key(
+    key_filename: &Utf8Path,
+    cert_filename: &Utf8Path,
+) -> anyhow::Result<ParsedCertifiedKey> {
+    let certified_key = load_certified_key(key_filename, cert_filename).await?;
+    let expiration_time = parse_expiration_time(certified_key.end_entity_cert()?)?;
+    Ok(ParsedCertifiedKey {
+        certified_key,
+        expiration_time,
+    })
+}
+
+static CERT_EXPIRATION_TIME: Lazy<UIntGaugeVec> = Lazy::new(|| {
+    register_uint_gauge_vec!(
+        "tls_certs_expiration_time_seconds",
+        "Expiration time of the loaded certificate since unix epoch in seconds",
+        &["resolver_name"]
+    )
+    .expect("failed to define a metric")
+});
+
+static CERT_RELOAD_STARTED_COUNTER: Lazy<IntCounterVec> = Lazy::new(|| {
+    register_int_counter_vec!(
+        "tls_certs_reload_started_total",
+        "Number of certificate reload loop iterations started",
+        &["resolver_name"]
+    )
+    .expect("failed to define a metric")
+});
+
+static CERT_RELOAD_UPDATED_COUNTER: Lazy<IntCounterVec> = Lazy::new(|| {
+    register_int_counter_vec!(
+        "tls_certs_reload_updated_total",
+        "Number of times the certificate was updated to the new one",
+        &["resolver_name"]
+    )
+    .expect("failed to define a metric")
+});
+
+static CERT_RELOAD_FAILED_COUNTER: Lazy<IntCounterVec> = Lazy::new(|| {
+    register_int_counter_vec!(
+        "tls_certs_reload_failed_total",
+        "Number of times the certificate reload failed",
+        &["resolver_name"]
+    )
+    .expect("failed to define a metric")
+});
+
 /// Implementation of [`rustls::server::ResolvesServerCert`] which reloads certificates from
 /// the disk periodically.
 #[derive(Debug)]
@@ -63,16 +136,28 @@ pub struct ReloadingCertificateResolver {
 impl ReloadingCertificateResolver {
     /// Creates a new Resolver by loading certificate and private key from FS and
     /// creating tokio::task to reload them with provided reload_period.
+    /// resolver_name is used as metric's label.
     pub async fn new(
+        resolver_name: &str,
         key_filename: &Utf8Path,
         cert_filename: &Utf8Path,
         reload_period: Duration,
     ) -> anyhow::Result<Arc<Self>> {
+        // Create metrics for current resolver.
+        let cert_expiration_time = CERT_EXPIRATION_TIME.with_label_values(&[resolver_name]);
+        let cert_reload_started_counter =
+            CERT_RELOAD_STARTED_COUNTER.with_label_values(&[resolver_name]);
+        let cert_reload_updated_counter =
+            CERT_RELOAD_UPDATED_COUNTER.with_label_values(&[resolver_name]);
+        let cert_reload_failed_counter =
+            CERT_RELOAD_FAILED_COUNTER.with_label_values(&[resolver_name]);
+
+        let parsed_key = load_and_parse_certified_key(key_filename, cert_filename).await?;
+
         let this = Arc::new(Self {
-            certified_key: ArcSwap::from_pointee(
-                load_certified_key(key_filename, cert_filename).await?,
-            ),
+            certified_key: ArcSwap::from_pointee(parsed_key.certified_key),
         });
+        cert_expiration_time.set(parsed_key.expiration_time.as_secs());
 
         tokio::spawn({
             let weak_this = Arc::downgrade(&this);
@@ -88,17 +173,22 @@ impl ReloadingCertificateResolver {
                         Some(this) => this,
                         None => break, // Resolver has been destroyed, exit.
                     };
-                    match load_certified_key(&key_filename, &cert_filename).await {
-                        Ok(new_certified_key) => {
-                            if new_certified_key.cert == this.certified_key.load().cert {
+                    cert_reload_started_counter.inc();
+
+                    match load_and_parse_certified_key(&key_filename, &cert_filename).await {
+                        Ok(parsed_key) => {
+                            if parsed_key.certified_key.cert == this.certified_key.load().cert {
                                 tracing::debug!("Certificate has not changed since last reloading");
                             } else {
                                 tracing::info!("Certificate has been reloaded");
-                                this.certified_key.store(Arc::new(new_certified_key));
+                                this.certified_key.store(Arc::new(parsed_key.certified_key));
+                                cert_expiration_time.set(parsed_key.expiration_time.as_secs());
+                                cert_reload_updated_counter.inc();
                             }
                             last_reload_failed = false;
                         }
                         Err(err) => {
+                            cert_reload_failed_counter.inc();
                             // Note: Reloading certs may fail if it conflicts with the script updating
                             // the files at the same time. Warn only if the error is persistent.
                             if last_reload_failed {

libs/metrics/src/lib.rs

@@ -76,14 +76,7 @@ pub fn gather() -> Vec<prometheus::proto::MetricFamily> {
     mfs
 }
 
-static DISK_IO_BYTES: Lazy<IntGaugeVec> = Lazy::new(|| {
-    register_int_gauge_vec!(
-        "libmetrics_disk_io_bytes_total",
-        "Bytes written and read from disk, grouped by the operation (read|write)",
-        &["io_operation"]
-    )
-    .expect("Failed to register disk i/o bytes int gauge vec")
-});
+
 
 static MAXRSS_KB: Lazy<IntGauge> = Lazy::new(|| {
     register_int_gauge!(
@@ -261,12 +254,7 @@ const BYTES_IN_BLOCK: i64 = 512;
 fn update_rusage_metrics() {
     let rusage_stats = get_rusage_stats();
 
-    DISK_IO_BYTES
-        .with_label_values(&["read"])
-        .set(rusage_stats.ru_inblock * BYTES_IN_BLOCK);
-    DISK_IO_BYTES
-        .with_label_values(&["write"])
-        .set(rusage_stats.ru_oublock * BYTES_IN_BLOCK);
+    
 
     // On macOS, the unit of maxrss is bytes; on Linux, it's kilobytes. https://stackoverflow.com/a/59915669
     #[cfg(target_os = "macos")]
@@ -357,10 +345,7 @@ impl<P: Atomic> GenericCounterPairVec<P> {
         self.get_metric_with_label_values(vals).unwrap()
     }
 
-    pub fn remove_label_values(&self, res: &mut [prometheus::Result<()>; 2], vals: &[&str]) {
-        res[0] = self.inc.remove_label_values(vals);
-        res[1] = self.dec.remove_label_values(vals);
-    }
+    
 }
 
 impl<P: Atomic> GenericCounterPair<P> {

libs/pageserver_api/src/config.rs

@@ -180,6 +180,7 @@ pub struct ConfigToml {
     #[serde(skip_serializing_if = "Option::is_none")]
     pub generate_unarchival_heatmap: Option<bool>,
     pub tracing: Option<Tracing>,
+    pub enable_tls_page_service_api: bool,
 }
 
 #[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
@@ -206,6 +207,10 @@ pub struct PageServicePipeliningConfigPipelined {
     /// Causes runtime errors if larger than max get_vectored batch size.
     pub max_batch_size: NonZeroUsize,
     pub execution: PageServiceProtocolPipelinedExecutionStrategy,
+    // The default below is such that new versions of the software can start
+    // with the old configuration.
+    #[serde(default)]
+    pub batching: PageServiceProtocolPipelinedBatchingStrategy,
 }
 
 #[derive(Debug, Clone, Copy, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
@@ -215,6 +220,19 @@ pub enum PageServiceProtocolPipelinedExecutionStrategy {
     Tasks,
 }
 
+#[derive(Default, Debug, Clone, Copy, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
+#[serde(rename_all = "kebab-case")]
+pub enum PageServiceProtocolPipelinedBatchingStrategy {
+    /// All get page requests in a batch will be at the same LSN
+    #[default]
+    UniformLsn,
+    /// Get page requests in a batch may be at different LSN
+    ///
+    /// One key cannot be present more than once at different LSNs in
+    /// the same batch.
+    ScatteredLsn,
+}
+
 #[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
 #[serde(tag = "mode", rename_all = "kebab-case")]
 pub enum GetVectoredConcurrentIo {
@@ -361,6 +379,8 @@ pub struct TenantConfigToml {
     /// size exceeds `compaction_upper_limit * checkpoint_distance`.
     pub compaction_upper_limit: usize,
     pub compaction_algorithm: crate::models::CompactionAlgorithmSettings,
+    /// If true, enable shard ancestor compaction (enabled by default).
+    pub compaction_shard_ancestor: bool,
     /// If true, compact down L0 across all tenant timelines before doing regular compaction. L0
     /// compaction must be responsive to avoid read amp during heavy ingestion. Defaults to true.
     pub compaction_l0_first: bool,
@@ -451,6 +471,8 @@ pub struct TenantConfigToml {
     // gc-compaction related configs
     /// Enable automatic gc-compaction trigger on this tenant.
     pub gc_compaction_enabled: bool,
+    /// Enable verification of gc-compaction results.
+    pub gc_compaction_verification: bool,
     /// The initial threshold for gc-compaction in KB. Once the total size of layers below the gc-horizon is above this threshold,
     /// gc-compaction will be triggered.
     pub gc_compaction_initial_threshold_kb: u64,
@@ -612,9 +634,12 @@ impl Default for ConfigToml {
             page_service_pipelining: if !cfg!(test) {
                 PageServicePipeliningConfig::Serial
             } else {
+                // Do not turn this into the default until scattered reads have been
+                // validated and rolled-out fully.
                 PageServicePipeliningConfig::Pipelined(PageServicePipeliningConfigPipelined {
                     max_batch_size: NonZeroUsize::new(32).unwrap(),
                     execution: PageServiceProtocolPipelinedExecutionStrategy::ConcurrentFutures,
+                    batching: PageServiceProtocolPipelinedBatchingStrategy::ScatteredLsn,
                 })
             },
             get_vectored_concurrent_io: if !cfg!(test) {
@@ -631,6 +656,7 @@ impl Default for ConfigToml {
             load_previous_heatmap: None,
             generate_unarchival_heatmap: None,
             tracing: None,
+            enable_tls_page_service_api: false,
         }
     }
 }
@@ -653,6 +679,7 @@ pub mod tenant_conf_defaults {
 
     pub const DEFAULT_COMPACTION_PERIOD: &str = "20 s";
     pub const DEFAULT_COMPACTION_THRESHOLD: usize = 10;
+    pub const DEFAULT_COMPACTION_SHARD_ANCESTOR: bool = true;
 
     // This value needs to be tuned to avoid OOM. We have 3/4*CPUs threads for L0 compaction, that's
     // 3/4*16=9 on most of our pageservers. Compacting 20 layers requires about 1 GB memory (could
@@ -690,6 +717,7 @@ pub mod tenant_conf_defaults {
     // image layers should be created.
     pub const DEFAULT_IMAGE_LAYER_CREATION_CHECK_THRESHOLD: u8 = 2;
     pub const DEFAULT_GC_COMPACTION_ENABLED: bool = false;
+    pub const DEFAULT_GC_COMPACTION_VERIFICATION: bool = true;
     pub const DEFAULT_GC_COMPACTION_INITIAL_THRESHOLD_KB: u64 = 5 * 1024 * 1024; // 5GB
     pub const DEFAULT_GC_COMPACTION_RATIO_PERCENT: u64 = 100;
 }
@@ -709,6 +737,7 @@ impl Default for TenantConfigToml {
             compaction_algorithm: crate::models::CompactionAlgorithmSettings {
                 kind: DEFAULT_COMPACTION_ALGORITHM,
             },
+            compaction_shard_ancestor: DEFAULT_COMPACTION_SHARD_ANCESTOR,
             compaction_l0_first: DEFAULT_COMPACTION_L0_FIRST,
             compaction_l0_semaphore: DEFAULT_COMPACTION_L0_SEMAPHORE,
             l0_flush_delay_threshold: None,
@@ -744,6 +773,7 @@ impl Default for TenantConfigToml {
             wal_receiver_protocol_override: None,
             rel_size_v2_enabled: false,
             gc_compaction_enabled: DEFAULT_GC_COMPACTION_ENABLED,
+            gc_compaction_verification: DEFAULT_GC_COMPACTION_VERIFICATION,
             gc_compaction_initial_threshold_kb: DEFAULT_GC_COMPACTION_INITIAL_THRESHOLD_KB,
             gc_compaction_ratio_percent: DEFAULT_GC_COMPACTION_RATIO_PERCENT,
             sampling_ratio: None,

libs/pageserver_api/src/controller_api.rs

@@ -7,7 +7,8 @@ use std::time::{Duration, Instant};
 /// API (`/control/v1` prefix).  Implemented by the server
 /// in [`storage_controller::http`]
 use serde::{Deserialize, Serialize};
-use utils::id::{NodeId, TenantId};
+use utils::id::{NodeId, TenantId, TimelineId};
+use utils::lsn::Lsn;
 
 use crate::models::{PageserverUtilization, ShardParameters, TenantConfig};
 use crate::shard::{ShardStripeSize, TenantShardId};
@@ -499,6 +500,15 @@ pub struct SafekeeperSchedulingPolicyRequest {
     pub scheduling_policy: SkSchedulingPolicy,
 }
 
+/// Import request for safekeeper timelines.
+#[derive(Serialize, Deserialize, Clone)]
+pub struct TimelineImportRequest {
+    pub tenant_id: TenantId,
+    pub timeline_id: TimelineId,
+    pub start_lsn: Lsn,
+    pub sk_set: Vec<NodeId>,
+}
+
 #[cfg(test)]
 mod test {
     use serde_json;

libs/pageserver_api/src/key.rs

@@ -927,7 +927,7 @@ impl Key {
 
     /// Guaranteed to return `Ok()` if [`Self::is_rel_block_key`] returns `true` for `key`.
     #[inline(always)]
-    pub fn to_rel_block(self) -> anyhow::Result<(RelTag, BlockNumber)> {
+    pub fn to_rel_block(self) -> Result<(RelTag, BlockNumber), ToRelBlockError> {
         Ok(match self.field1 {
             0x00 => (
                 RelTag {
@@ -938,7 +938,7 @@ impl Key {
                 },
                 self.field6,
             ),
-            _ => anyhow::bail!("unexpected value kind 0x{:02x}", self.field1),
+            _ => return Err(ToRelBlockError(self.field1)),
         })
     }
 }
@@ -951,6 +951,17 @@ impl std::str::FromStr for Key {
     }
 }
 
+#[derive(Debug)]
+pub struct ToRelBlockError(u8);
+
+impl fmt::Display for ToRelBlockError {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        write!(f, "unexpected value kind 0x{:02x}", self.0)
+    }
+}
+
+impl std::error::Error for ToRelBlockError {}
+
 #[cfg(test)]
 mod tests {
     use std::str::FromStr;

libs/pageserver_api/src/models.rs

@@ -526,6 +526,8 @@ pub struct TenantConfigPatch {
     #[serde(skip_serializing_if = "FieldPatch::is_noop")]
     pub compaction_algorithm: FieldPatch<CompactionAlgorithmSettings>,
     #[serde(skip_serializing_if = "FieldPatch::is_noop")]
+    pub compaction_shard_ancestor: FieldPatch<bool>,
+    #[serde(skip_serializing_if = "FieldPatch::is_noop")]
     pub compaction_l0_first: FieldPatch<bool>,
     #[serde(skip_serializing_if = "FieldPatch::is_noop")]
     pub compaction_l0_semaphore: FieldPatch<bool>,
@@ -576,6 +578,8 @@ pub struct TenantConfigPatch {
     #[serde(skip_serializing_if = "FieldPatch::is_noop")]
     pub gc_compaction_enabled: FieldPatch<bool>,
     #[serde(skip_serializing_if = "FieldPatch::is_noop")]
+    pub gc_compaction_verification: FieldPatch<bool>,
+    #[serde(skip_serializing_if = "FieldPatch::is_noop")]
     pub gc_compaction_initial_threshold_kb: FieldPatch<u64>,
     #[serde(skip_serializing_if = "FieldPatch::is_noop")]
     pub gc_compaction_ratio_percent: FieldPatch<u64>,
@@ -613,6 +617,9 @@ pub struct TenantConfig {
     #[serde(skip_serializing_if = "Option::is_none")]
     pub compaction_algorithm: Option<CompactionAlgorithmSettings>,
 
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub compaction_shard_ancestor: Option<bool>,
+
     #[serde(skip_serializing_if = "Option::is_none")]
     pub compaction_l0_first: Option<bool>,
 
@@ -696,6 +703,9 @@ pub struct TenantConfig {
     #[serde(skip_serializing_if = "Option::is_none")]
     pub gc_compaction_enabled: Option<bool>,
 
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub gc_compaction_verification: Option<bool>,
+
     #[serde(skip_serializing_if = "Option::is_none")]
     pub gc_compaction_initial_threshold_kb: Option<u64>,
 
@@ -719,6 +729,7 @@ impl TenantConfig {
             mut compaction_threshold,
             mut compaction_upper_limit,
             mut compaction_algorithm,
+            mut compaction_shard_ancestor,
             mut compaction_l0_first,
             mut compaction_l0_semaphore,
             mut l0_flush_delay_threshold,
@@ -744,6 +755,7 @@ impl TenantConfig {
             mut wal_receiver_protocol_override,
             mut rel_size_v2_enabled,
             mut gc_compaction_enabled,
+            mut gc_compaction_verification,
             mut gc_compaction_initial_threshold_kb,
             mut gc_compaction_ratio_percent,
             mut sampling_ratio,
@@ -766,6 +778,9 @@ impl TenantConfig {
             .compaction_upper_limit
             .apply(&mut compaction_upper_limit);
         patch.compaction_algorithm.apply(&mut compaction_algorithm);
+        patch
+            .compaction_shard_ancestor
+            .apply(&mut compaction_shard_ancestor);
         patch.compaction_l0_first.apply(&mut compaction_l0_first);
         patch
             .compaction_l0_semaphore
@@ -835,6 +850,9 @@ impl TenantConfig {
         patch
             .gc_compaction_enabled
             .apply(&mut gc_compaction_enabled);
+        patch
+            .gc_compaction_verification
+            .apply(&mut gc_compaction_verification);
         patch
             .gc_compaction_initial_threshold_kb
             .apply(&mut gc_compaction_initial_threshold_kb);
@@ -851,6 +869,7 @@ impl TenantConfig {
             compaction_threshold,
             compaction_upper_limit,
             compaction_algorithm,
+            compaction_shard_ancestor,
             compaction_l0_first,
             compaction_l0_semaphore,
             l0_flush_delay_threshold,
@@ -876,6 +895,7 @@ impl TenantConfig {
             wal_receiver_protocol_override,
             rel_size_v2_enabled,
             gc_compaction_enabled,
+            gc_compaction_verification,
             gc_compaction_initial_threshold_kb,
             gc_compaction_ratio_percent,
             sampling_ratio,
@@ -910,6 +930,9 @@ impl TenantConfig {
                 .as_ref()
                 .unwrap_or(&global_conf.compaction_algorithm)
                 .clone(),
+            compaction_shard_ancestor: self
+                .compaction_shard_ancestor
+                .unwrap_or(global_conf.compaction_shard_ancestor),
             compaction_l0_first: self
                 .compaction_l0_first
                 .unwrap_or(global_conf.compaction_l0_first),
@@ -974,6 +997,9 @@ impl TenantConfig {
             gc_compaction_enabled: self
                 .gc_compaction_enabled
                 .unwrap_or(global_conf.gc_compaction_enabled),
+            gc_compaction_verification: self
+                .gc_compaction_verification
+                .unwrap_or(global_conf.gc_compaction_verification),
             gc_compaction_initial_threshold_kb: self
                 .gc_compaction_initial_threshold_kb
                 .unwrap_or(global_conf.gc_compaction_initial_threshold_kb),

libs/remote_storage/src/azure_blob.rs

@@ -23,7 +23,6 @@ use futures::future::Either;
 use futures::stream::Stream;
 use futures_util::{StreamExt, TryStreamExt};
 use http_types::{StatusCode, Url};
-use scopeguard::ScopeGuard;
 use tokio_util::sync::CancellationToken;
 use tracing::debug;
 use utils::backoff;
@@ -32,7 +31,7 @@ use utils::backoff::exponential_backoff_duration_seconds;
 use super::REMOTE_STORAGE_PREFIX_SEPARATOR;
 use crate::config::AzureConfig;
 use crate::error::Cancelled;
-use crate::metrics::{AttemptOutcome, RequestKind, start_measuring_requests};
+use crate::metrics::RequestKind;
 use crate::{
     ConcurrencyLimiter, Download, DownloadError, DownloadKind, DownloadOpts, Listing, ListingMode,
     ListingObject, RemotePath, RemoteStorage, StorageMetadata, TimeTravelError, TimeoutOrCancel,
@@ -165,7 +164,7 @@ impl AzureBlobStorage {
         let mut last_modified = None;
         let mut metadata = HashMap::new();
 
-        let started_at = start_measuring_requests(kind);
+      
 
         let download = async {
             let response = builder
@@ -237,19 +236,8 @@ impl AzureBlobStorage {
                 TimeoutOrCancel::Cancel => return Err(DownloadError::Cancelled),
             },
         };
-        let started_at = ScopeGuard::into_inner(started_at);
-        let outcome = match &download {
-            Ok(_) => AttemptOutcome::Ok,
-            // At this level in the stack 404 and 304 responses do not indicate an error.
-            // There's expected cases when a blob may not exist or hasn't been modified since
-            // the last get (e.g. probing for timeline indices and heatmap downloads).
-            // Callers should handle errors if they are unexpected.
-            Err(DownloadError::NotFound | DownloadError::Unmodified) => AttemptOutcome::Ok,
-            Err(_) => AttemptOutcome::Err,
-        };
-        crate::metrics::BUCKET_METRICS
-            .req_seconds
-            .observe_elapsed(kind, outcome, started_at);
+        
+      
         download
     }
 
@@ -431,7 +419,7 @@ impl RemoteStorage for AzureBlobStorage {
         let kind = RequestKind::Head;
         let _permit = self.permit(kind, cancel).await?;
 
-        let started_at = start_measuring_requests(kind);
+      
 
         let blob_client = self.client.blob_client(self.relative_path_to_name(key));
         let properties_future = blob_client.get_properties().into_future();
@@ -443,12 +431,9 @@ impl RemoteStorage for AzureBlobStorage {
             _ = cancel.cancelled() => return Err(TimeoutOrCancel::Cancel.into()),
         };
 
-        if let Ok(inner) = &res {
-            // do not incl. timeouts as errors in metrics but cancellations
-            let started_at = ScopeGuard::into_inner(started_at);
-            crate::metrics::BUCKET_METRICS
-                .req_seconds
-                .observe_elapsed(kind, inner, started_at);
+        if let Ok(_inner) = &res {
+            
+         
         }
 
         let data = match res {
@@ -476,7 +461,7 @@ impl RemoteStorage for AzureBlobStorage {
         let kind = RequestKind::Put;
         let _permit = self.permit(kind, cancel).await?;
 
-        let started_at = start_measuring_requests(kind);
+      
 
         let op = async {
             let blob_client = self.client.blob_client(self.relative_path_to_name(to));
@@ -509,14 +494,7 @@ impl RemoteStorage for AzureBlobStorage {
             _ = cancel.cancelled() => return Err(TimeoutOrCancel::Cancel.into()),
         };
 
-        let outcome = match res {
-            Ok(_) => AttemptOutcome::Ok,
-            Err(_) => AttemptOutcome::Err,
-        };
-        let started_at = ScopeGuard::into_inner(started_at);
-        crate::metrics::BUCKET_METRICS
-            .req_seconds
-            .observe_elapsed(kind, outcome, started_at);
+      
 
         res
     }
@@ -562,7 +540,7 @@ impl RemoteStorage for AzureBlobStorage {
     ) -> anyhow::Result<()> {
         let kind = RequestKind::Delete;
         let _permit = self.permit(kind, cancel).await?;
-        let started_at = start_measuring_requests(kind);
+    
 
         let op = async {
             // TODO batch requests are not supported by the SDK
@@ -628,10 +606,8 @@ impl RemoteStorage for AzureBlobStorage {
             _ = cancel.cancelled() => return Err(TimeoutOrCancel::Cancel.into()),
         };
 
-        let started_at = ScopeGuard::into_inner(started_at);
-        crate::metrics::BUCKET_METRICS
-            .req_seconds
-            .observe_elapsed(kind, &res, started_at);
+       
+        
         res
     }
 
@@ -647,7 +623,7 @@ impl RemoteStorage for AzureBlobStorage {
     ) -> anyhow::Result<()> {
         let kind = RequestKind::Copy;
         let _permit = self.permit(kind, cancel).await?;
-        let started_at = start_measuring_requests(kind);
+  
 
         let timeout = tokio::time::sleep(self.timeout);
 
@@ -701,10 +677,8 @@ impl RemoteStorage for AzureBlobStorage {
             },
         };
 
-        let started_at = ScopeGuard::into_inner(started_at);
-        crate::metrics::BUCKET_METRICS
-            .req_seconds
-            .observe_elapsed(kind, &res, started_at);
+        
+      
         res
     }
 

libs/remote_storage/src/metrics.rs

@@ -1,9 +1,7 @@
-use metrics::{
-    Histogram, IntCounter, register_histogram_vec, register_int_counter, register_int_counter_vec,
-};
-use once_cell::sync::Lazy;
 
-pub(super) static BUCKET_METRICS: Lazy<BucketMetrics> = Lazy::new(Default::default);
+
+
+
 
 #[derive(Clone, Copy, Debug)]
 pub(crate) enum RequestKind {
@@ -16,62 +14,9 @@ pub(crate) enum RequestKind {
     Head = 6,
 }
 
-use RequestKind::*;
+
 use scopeguard::ScopeGuard;
 
-impl RequestKind {
-    const fn as_str(&self) -> &'static str {
-        match self {
-            Get => "get_object",
-            Put => "put_object",
-            Delete => "delete_object",
-            List => "list_objects",
-            Copy => "copy_object",
-            TimeTravel => "time_travel_recover",
-            Head => "head_object",
-        }
-    }
-    const fn as_index(&self) -> usize {
-        *self as usize
-    }
-}
-
-const REQUEST_KIND_COUNT: usize = 7;
-pub(crate) struct RequestTyped<C>([C; REQUEST_KIND_COUNT]);
-
-impl<C> RequestTyped<C> {
-    pub(crate) fn get(&self, kind: RequestKind) -> &C {
-        &self.0[kind.as_index()]
-    }
-
-    fn build_with(mut f: impl FnMut(RequestKind) -> C) -> Self {
-        use RequestKind::*;
-        let mut it = [Get, Put, Delete, List, Copy, TimeTravel, Head].into_iter();
-        let arr = std::array::from_fn::<C, REQUEST_KIND_COUNT, _>(|index| {
-            let next = it.next().unwrap();
-            assert_eq!(index, next.as_index());
-            f(next)
-        });
-
-        if let Some(next) = it.next() {
-            panic!("unexpected {next:?}");
-        }
-
-        RequestTyped(arr)
-    }
-}
-
-impl RequestTyped<Histogram> {
-    pub(crate) fn observe_elapsed(&self, kind: RequestKind, started_at: std::time::Instant) {
-        self.get(kind).observe(started_at.elapsed().as_secs_f64())
-    }
-}
-
-pub(crate) struct PassFailCancelledRequestTyped<C> {
-    success: RequestTyped<C>,
-    fail: RequestTyped<C>,
-    cancelled: RequestTyped<C>,
-}
 
 #[derive(Debug, Clone, Copy)]
 pub(crate) enum AttemptOutcome {
@@ -89,138 +34,22 @@ impl<T, E> From<&Result<T, E>> for AttemptOutcome {
     }
 }
 
-impl AttemptOutcome {
-    pub(crate) fn as_str(&self) -> &'static str {
-        match self {
-            AttemptOutcome::Ok => "ok",
-            AttemptOutcome::Err => "err",
-            AttemptOutcome::Cancelled => "cancelled",
-        }
-    }
-}
 
-impl<C> PassFailCancelledRequestTyped<C> {
-    pub(crate) fn get(&self, kind: RequestKind, outcome: AttemptOutcome) -> &C {
-        let target = match outcome {
-            AttemptOutcome::Ok => &self.success,
-            AttemptOutcome::Err => &self.fail,
-            AttemptOutcome::Cancelled => &self.cancelled,
-        };
-        target.get(kind)
-    }
 
-    fn build_with(mut f: impl FnMut(RequestKind, AttemptOutcome) -> C) -> Self {
-        let success = RequestTyped::build_with(|kind| f(kind, AttemptOutcome::Ok));
-        let fail = RequestTyped::build_with(|kind| f(kind, AttemptOutcome::Err));
-        let cancelled = RequestTyped::build_with(|kind| f(kind, AttemptOutcome::Cancelled));
 
-        PassFailCancelledRequestTyped {
-            success,
-            fail,
-            cancelled,
-        }
-    }
-}
 
-impl PassFailCancelledRequestTyped<Histogram> {
-    pub(crate) fn observe_elapsed(
-        &self,
-        kind: RequestKind,
-        outcome: impl Into<AttemptOutcome>,
-        started_at: std::time::Instant,
-    ) {
-        self.get(kind, outcome.into())
-            .observe(started_at.elapsed().as_secs_f64())
-    }
-}
 
-/// On drop (cancellation) count towards [`BucketMetrics::cancelled_waits`].
-pub(crate) fn start_counting_cancelled_wait(
-    kind: RequestKind,
-) -> ScopeGuard<std::time::Instant, impl FnOnce(std::time::Instant), scopeguard::OnSuccess> {
-    scopeguard::guard_on_success(std::time::Instant::now(), move |_| {
-        crate::metrics::BUCKET_METRICS
-            .cancelled_waits
-            .get(kind)
-            .inc()
-    })
-}
+
+
 
 /// On drop (cancellation) add time to [`BucketMetrics::req_seconds`].
 pub(crate) fn start_measuring_requests(
-    kind: RequestKind,
+    _kind: RequestKind,
 ) -> ScopeGuard<std::time::Instant, impl FnOnce(std::time::Instant), scopeguard::OnSuccess> {
-    scopeguard::guard_on_success(std::time::Instant::now(), move |started_at| {
-        crate::metrics::BUCKET_METRICS.req_seconds.observe_elapsed(
-            kind,
-            AttemptOutcome::Cancelled,
-            started_at,
-        )
+    scopeguard::guard_on_success(std::time::Instant::now(), move |_started_at| {
+        
     })
 }
 
-pub(crate) struct BucketMetrics {
-    /// Full request duration until successful completion, error or cancellation.
-    pub(crate) req_seconds: PassFailCancelledRequestTyped<Histogram>,
-    /// Total amount of seconds waited on queue.
-    pub(crate) wait_seconds: RequestTyped<Histogram>,
 
-    /// Track how many semaphore awaits were cancelled per request type.
-    ///
-    /// This is in case cancellations are happening more than expected.
-    pub(crate) cancelled_waits: RequestTyped<IntCounter>,
 
-    /// Total amount of deleted objects in batches or single requests.
-    pub(crate) deleted_objects_total: IntCounter,
-}
-
-impl Default for BucketMetrics {
-    fn default() -> Self {
-        // first bucket 100 microseconds to count requests that do not need to wait at all
-        // and get a permit immediately
-        let buckets = [0.0001, 0.01, 0.10, 0.5, 1.0, 5.0, 10.0, 50.0, 100.0];
-
-        let req_seconds = register_histogram_vec!(
-            "remote_storage_s3_request_seconds",
-            "Seconds to complete a request",
-            &["request_type", "result"],
-            buckets.to_vec(),
-        )
-        .unwrap();
-        let req_seconds = PassFailCancelledRequestTyped::build_with(|kind, outcome| {
-            req_seconds.with_label_values(&[kind.as_str(), outcome.as_str()])
-        });
-
-        let wait_seconds = register_histogram_vec!(
-            "remote_storage_s3_wait_seconds",
-            "Seconds rate limited",
-            &["request_type"],
-            buckets.to_vec(),
-        )
-        .unwrap();
-        let wait_seconds =
-            RequestTyped::build_with(|kind| wait_seconds.with_label_values(&[kind.as_str()]));
-
-        let cancelled_waits = register_int_counter_vec!(
-            "remote_storage_s3_cancelled_waits_total",
-            "Times a semaphore wait has been cancelled per request type",
-            &["request_type"],
-        )
-        .unwrap();
-        let cancelled_waits =
-            RequestTyped::build_with(|kind| cancelled_waits.with_label_values(&[kind.as_str()]));
-
-        let deleted_objects_total = register_int_counter!(
-            "remote_storage_s3_deleted_objects_total",
-            "Amount of deleted objects in total",
-        )
-        .unwrap();
-
-        Self {
-            req_seconds,
-            wait_seconds,
-            cancelled_waits,
-            deleted_objects_total,
-        }
-    }
-}

libs/remote_storage/src/s3_bucket.rs

@@ -41,7 +41,7 @@ use super::StorageMetadata;
 use crate::config::S3Config;
 use crate::error::Cancelled;
 pub(super) use crate::metrics::RequestKind;
-use crate::metrics::{AttemptOutcome, start_counting_cancelled_wait, start_measuring_requests};
+use crate::metrics::{AttemptOutcome, start_measuring_requests};
 use crate::support::PermitCarrying;
 use crate::{
     ConcurrencyLimiter, Download, DownloadError, DownloadOpts, Listing, ListingMode, ListingObject,
@@ -199,7 +199,7 @@ impl S3Bucket {
         kind: RequestKind,
         cancel: &CancellationToken,
     ) -> Result<tokio::sync::SemaphorePermit<'_>, Cancelled> {
-        let started_at = start_counting_cancelled_wait(kind);
+       
         let acquire = self.concurrency_limiter.acquire(kind);
 
         let permit = tokio::select! {
@@ -207,10 +207,8 @@ impl S3Bucket {
             _ = cancel.cancelled() => return Err(Cancelled),
         };
 
-        let started_at = ScopeGuard::into_inner(started_at);
-        crate::metrics::BUCKET_METRICS
-            .wait_seconds
-            .observe_elapsed(kind, started_at);
+       
+        
 
         Ok(permit)
     }
@@ -220,7 +218,7 @@ impl S3Bucket {
         kind: RequestKind,
         cancel: &CancellationToken,
     ) -> Result<tokio::sync::OwnedSemaphorePermit, Cancelled> {
-        let started_at = start_counting_cancelled_wait(kind);
+       
         let acquire = self.concurrency_limiter.acquire_owned(kind);
 
         let permit = tokio::select! {
@@ -228,10 +226,8 @@ impl S3Bucket {
             _ = cancel.cancelled() => return Err(Cancelled),
         };
 
-        let started_at = ScopeGuard::into_inner(started_at);
-        crate::metrics::BUCKET_METRICS
-            .wait_seconds
-            .observe_elapsed(kind, started_at);
+       
+        
         Ok(permit)
     }
 
@@ -273,11 +269,7 @@ impl S3Bucket {
                 // Count this in the AttemptOutcome::Ok bucket, because 404 is not
                 // an error: we expect to sometimes fetch an object and find it missing,
                 // e.g. when probing for timeline indices.
-                crate::metrics::BUCKET_METRICS.req_seconds.observe_elapsed(
-                    kind,
-                    AttemptOutcome::Ok,
-                    started_at,
-                );
+                
                 return Err(DownloadError::NotFound);
             }
             Err(SdkError::ServiceError(e))
@@ -287,19 +279,11 @@ impl S3Bucket {
                 if e.raw().status().as_u16() == StatusCode::NotModified =>
             {
                 // Count an unmodified file as a success.
-                crate::metrics::BUCKET_METRICS.req_seconds.observe_elapsed(
-                    kind,
-                    AttemptOutcome::Ok,
-                    started_at,
-                );
+               
                 return Err(DownloadError::Unmodified);
             }
             Err(e) => {
-                crate::metrics::BUCKET_METRICS.req_seconds.observe_elapsed(
-                    kind,
-                    AttemptOutcome::Err,
-                    started_at,
-                );
+                
 
                 return Err(DownloadError::Other(
                     anyhow::Error::new(e).context("download s3 object"),
@@ -346,11 +330,11 @@ impl S3Bucket {
         delete_objects: &[ObjectIdentifier],
         cancel: &CancellationToken,
     ) -> anyhow::Result<()> {
-        let kind = RequestKind::Delete;
+   
         let mut cancel = std::pin::pin!(cancel.cancelled());
 
         for chunk in delete_objects.chunks(MAX_KEYS_PER_DELETE_S3) {
-            let started_at = start_measuring_requests(kind);
+           
 
             let req = self
                 .client
@@ -370,15 +354,10 @@ impl S3Bucket {
                 _ = &mut cancel => return Err(TimeoutOrCancel::Cancel.into()),
             };
 
-            let started_at = ScopeGuard::into_inner(started_at);
-            crate::metrics::BUCKET_METRICS
-                .req_seconds
-                .observe_elapsed(kind, &resp, started_at);
-
+          
+            
             let resp = resp.context("request deletion")?;
-            crate::metrics::BUCKET_METRICS
-                .deleted_objects_total
-                .inc_by(chunk.len() as u64);
+            
 
             if let Some(errors) = resp.errors {
                 // Log a bounded number of the errors within the response:
@@ -445,8 +424,8 @@ pin_project_lite::pin_project! {
     }
 
     impl<S> PinnedDrop for TimedDownload<S> {
-        fn drop(mut this: Pin<&mut Self>) {
-            crate::metrics::BUCKET_METRICS.req_seconds.observe_elapsed(RequestKind::Get, this.outcome, this.started_at);
+        fn drop(mut _this: Pin<&mut Self>) {
+           
         }
     }
 }
@@ -511,7 +490,7 @@ impl RemoteStorage for S3Bucket {
 
             let mut continuation_token = None;
             'outer: loop {
-                let started_at = start_measuring_requests(kind);
+           
 
                 // min of two Options, returning Some if one is value and another is
                 // None (None is smaller than anything, so plain min doesn't work).
@@ -544,11 +523,9 @@ impl RemoteStorage for S3Bucket {
                     .context("Failed to list S3 prefixes")
                     .map_err(DownloadError::Other);
 
-                let started_at = ScopeGuard::into_inner(started_at);
+               
 
-                crate::metrics::BUCKET_METRICS
-                    .req_seconds
-                    .observe_elapsed(kind, &response, started_at);
+                
 
                 let response = match response {
                     Ok(response) => response,
@@ -629,7 +606,7 @@ impl RemoteStorage for S3Bucket {
         let kind = RequestKind::Head;
         let _permit = self.permit(kind, cancel).await?;
 
-        let started_at = start_measuring_requests(kind);
+     
 
         let head_future = self
             .client
@@ -648,30 +625,18 @@ impl RemoteStorage for S3Bucket {
         let res = res.map_err(|_e| DownloadError::Timeout)?;
 
         // do not incl. timeouts as errors in metrics but cancellations
-        let started_at = ScopeGuard::into_inner(started_at);
-        crate::metrics::BUCKET_METRICS
-            .req_seconds
-            .observe_elapsed(kind, &res, started_at);
-
+  
+        
         let data = match res {
             Ok(object_output) => object_output,
             Err(SdkError::ServiceError(e)) if matches!(e.err(), HeadObjectError::NotFound(_)) => {
                 // Count this in the AttemptOutcome::Ok bucket, because 404 is not
                 // an error: we expect to sometimes fetch an object and find it missing,
                 // e.g. when probing for timeline indices.
-                crate::metrics::BUCKET_METRICS.req_seconds.observe_elapsed(
-                    kind,
-                    AttemptOutcome::Ok,
-                    started_at,
-                );
                 return Err(DownloadError::NotFound);
             }
             Err(e) => {
-                crate::metrics::BUCKET_METRICS.req_seconds.observe_elapsed(
-                    kind,
-                    AttemptOutcome::Err,
-                    started_at,
-                );
+                
 
                 return Err(DownloadError::Other(
                     anyhow::Error::new(e).context("s3 head object"),
@@ -704,7 +669,7 @@ impl RemoteStorage for S3Bucket {
         let kind = RequestKind::Put;
         let _permit = self.permit(kind, cancel).await?;
 
-        let started_at = start_measuring_requests(kind);
+      
 
         let body = StreamBody::new(from.map(|x| x.map(Frame::data)));
         let bytes_stream = ByteStream::new(SdkBody::from_body_1_x(body));
@@ -727,12 +692,10 @@ impl RemoteStorage for S3Bucket {
             _ = cancel.cancelled() => return Err(TimeoutOrCancel::Cancel.into()),
         };
 
-        if let Ok(inner) = &res {
+        if let Ok(_inner) = &res {
             // do not incl. timeouts as errors in metrics but cancellations
-            let started_at = ScopeGuard::into_inner(started_at);
-            crate::metrics::BUCKET_METRICS
-                .req_seconds
-                .observe_elapsed(kind, inner, started_at);
+       
+            
         }
 
         match res {
@@ -753,7 +716,7 @@ impl RemoteStorage for S3Bucket {
 
         let timeout = tokio::time::sleep(self.timeout);
 
-        let started_at = start_measuring_requests(kind);
+       
 
         // we need to specify bucket_name as a prefix
         let copy_source = format!(
@@ -777,10 +740,8 @@ impl RemoteStorage for S3Bucket {
             _ = cancel.cancelled() => return Err(TimeoutOrCancel::Cancel.into()),
         };
 
-        let started_at = ScopeGuard::into_inner(started_at);
-        crate::metrics::BUCKET_METRICS
-            .req_seconds
-            .observe_elapsed(kind, &res, started_at);
+       
+        
 
         res?;
 

libs/utils/Cargo.toml

@@ -29,6 +29,7 @@ futures = { workspace = true }
 jsonwebtoken.workspace = true
 nix = { workspace = true, features = ["ioctl"] }
 once_cell.workspace = true
+pem.workspace = true
 pin-project-lite.workspace = true
 regex.workspace = true
 serde.workspace = true

libs/utils/src/auth.rs

@@ -11,7 +11,8 @@ use camino::Utf8Path;
 use jsonwebtoken::{
     Algorithm, DecodingKey, EncodingKey, Header, TokenData, Validation, decode, encode,
 };
-use serde::{Deserialize, Serialize};
+use pem::Pem;
+use serde::{Deserialize, Serialize, de::DeserializeOwned};
 
 use crate::id::TenantId;
 
@@ -73,7 +74,10 @@ impl SwappableJwtAuth {
     pub fn swap(&self, jwt_auth: JwtAuth) {
         self.0.swap(Arc::new(jwt_auth));
     }
-    pub fn decode(&self, token: &str) -> std::result::Result<TokenData<Claims>, AuthError> {
+    pub fn decode<D: DeserializeOwned>(
+        &self,
+        token: &str,
+    ) -> std::result::Result<TokenData<D>, AuthError> {
         self.0.load().decode(token)
     }
 }
@@ -148,7 +152,10 @@ impl JwtAuth {
     /// The function tries the stored decoding keys in succession,
     /// and returns the first yielding a successful result.
     /// If there is no working decoding key, it returns the last error.
-    pub fn decode(&self, token: &str) -> std::result::Result<TokenData<Claims>, AuthError> {
+    pub fn decode<D: DeserializeOwned>(
+        &self,
+        token: &str,
+    ) -> std::result::Result<TokenData<D>, AuthError> {
         let mut res = None;
         for decoding_key in &self.decoding_keys {
             res = Some(decode(token, decoding_key, &self.validation));
@@ -173,8 +180,8 @@ impl std::fmt::Debug for JwtAuth {
 }
 
 // this function is used only for testing purposes in CLI e g generate tokens during init
-pub fn encode_from_key_file<S: Serialize>(claims: &S, key_data: &[u8]) -> Result<String> {
-    let key = EncodingKey::from_ed_pem(key_data)?;
+pub fn encode_from_key_file<S: Serialize>(claims: &S, pem: &Pem) -> Result<String> {
+    let key = EncodingKey::from_ed_der(pem.contents());
     Ok(encode(&Header::new(STORAGE_TOKEN_ALGORITHM), claims, &key)?)
 }
 
@@ -188,13 +195,13 @@ mod tests {
     //
     // openssl genpkey -algorithm ed25519 -out ed25519-priv.pem
     // openssl pkey -in ed25519-priv.pem -pubout -out ed25519-pub.pem
-    const TEST_PUB_KEY_ED25519: &[u8] = br#"
+    const TEST_PUB_KEY_ED25519: &str = r#"
 -----BEGIN PUBLIC KEY-----
 MCowBQYDK2VwAyEARYwaNBayR+eGI0iXB4s3QxE3Nl2g1iWbr6KtLWeVD/w=
 -----END PUBLIC KEY-----
 "#;
 
-    const TEST_PRIV_KEY_ED25519: &[u8] = br#"
+    const TEST_PRIV_KEY_ED25519: &str = r#"
 -----BEGIN PRIVATE KEY-----
 MC4CAQAwBQYDK2VwBCIEID/Drmc1AA6U/znNRWpF3zEGegOATQxfkdWxitcOMsIH
 -----END PRIVATE KEY-----
@@ -222,9 +229,9 @@ MC4CAQAwBQYDK2VwBCIEID/Drmc1AA6U/znNRWpF3zEGegOATQxfkdWxitcOMsIH
 
         // Check it can be validated with the public key
         let auth = JwtAuth::new(vec![
-            DecodingKey::from_ed_pem(TEST_PUB_KEY_ED25519).unwrap(),
+            DecodingKey::from_ed_pem(TEST_PUB_KEY_ED25519.as_bytes()).unwrap(),
         ]);
-        let claims_from_token = auth.decode(encoded_eddsa).unwrap().claims;
+        let claims_from_token: Claims = auth.decode(encoded_eddsa).unwrap().claims;
         assert_eq!(claims_from_token, expected_claims);
     }
 
@@ -235,13 +242,14 @@ MC4CAQAwBQYDK2VwBCIEID/Drmc1AA6U/znNRWpF3zEGegOATQxfkdWxitcOMsIH
             scope: Scope::Tenant,
         };
 
-        let encoded = encode_from_key_file(&claims, TEST_PRIV_KEY_ED25519).unwrap();
+        let pem = pem::parse(TEST_PRIV_KEY_ED25519).unwrap();
+        let encoded = encode_from_key_file(&claims, &pem).unwrap();
 
         // decode it back
         let auth = JwtAuth::new(vec![
-            DecodingKey::from_ed_pem(TEST_PUB_KEY_ED25519).unwrap(),
+            DecodingKey::from_ed_pem(TEST_PUB_KEY_ED25519.as_bytes()).unwrap(),
         ]);
-        let decoded = auth.decode(&encoded).unwrap();
+        let decoded: TokenData<Claims> = auth.decode(&encoded).unwrap();
 
         assert_eq!(decoded.claims, claims);
     }

libs/utils/src/circuit_breaker.rs

@@ -1,7 +1,6 @@
 use std::fmt::Display;
 use std::time::{Duration, Instant};
 
-use metrics::IntCounter;
 
 /// Circuit breakers are for operations that are expensive and fallible.
 ///
@@ -54,7 +53,7 @@ impl CircuitBreaker {
         }
     }
 
-    pub fn fail<E>(&mut self, metric: &IntCounter, error: E)
+    pub fn fail<E>(&mut self,  error: E)
     where
         E: Display,
     {
@@ -64,18 +63,18 @@ impl CircuitBreaker {
 
         self.fail_count += 1;
         if self.broken_at.is_none() && self.fail_count >= self.fail_threshold {
-            self.break_circuit(metric, error);
+            self.break_circuit( error);
         }
     }
 
     /// Call this after successfully executing an operation
-    pub fn success(&mut self, metric: &IntCounter) {
+    pub fn success(&mut self) {
         self.fail_count = 0;
         if let Some(broken_at) = &self.broken_at {
             tracing::info!(breaker=%self.name, "Circuit breaker failure ended (was broken for {})",
                 humantime::format_duration(broken_at.elapsed()));
             self.broken_at = None;
-            metric.inc();
+          
         }
     }
 
@@ -98,13 +97,13 @@ impl CircuitBreaker {
         }
     }
 
-    fn break_circuit<E>(&mut self, metric: &IntCounter, error: E)
+    fn break_circuit<E>(&mut self,  error: E)
     where
         E: Display,
     {
         self.broken_at = Some(Instant::now());
         tracing::error!(breaker=%self.name, "Circuit breaker broken!  Last error: {error}");
-        metric.inc();
+        
     }
 
     fn reset_circuit(&mut self) {

pageserver/Cargo.toml

@@ -10,6 +10,8 @@ default = []
 # which adds some runtime cost to run tests on outage conditions
 testing = ["fail/failpoints", "pageserver_api/testing", "wal_decoder/testing", "pageserver_client/testing"]
 
+fuzz-read-path = ["testing"]
+
 [dependencies]
 anyhow.workspace = true
 arc-swap.workspace = true
@@ -33,6 +35,7 @@ humantime.workspace = true
 humantime-serde.workspace = true
 hyper0.workspace = true
 itertools.workspace = true
+jsonwebtoken.workspace = true
 md5.workspace = true
 nix.workspace = true
 # hack to get the number of worker threads tokio uses

pageserver/benches/bench_ingest.rs

@@ -126,7 +126,7 @@ async fn ingest(
             max_concurrency: NonZeroUsize::new(1).unwrap(),
         });
         let (_desc, path) = layer
-            .write_to_disk(&ctx, None, l0_flush_state.inner())
+            .write_to_disk(&ctx, None, l0_flush_state.inner(), &gate, cancel.clone())
             .await?
             .unwrap();
         tokio::fs::remove_file(path).await?;

pageserver/src/basebackup.rs

@@ -34,7 +34,7 @@ use utils::lsn::Lsn;
 use crate::context::RequestContext;
 use crate::pgdatadir_mapping::Version;
 use crate::tenant::storage_layer::IoConcurrency;
-use crate::tenant::timeline::GetVectoredError;
+use crate::tenant::timeline::{GetVectoredError, VersionedKeySpaceQuery};
 use crate::tenant::{PageReconstructError, Timeline};
 
 #[derive(Debug, thiserror::Error)]
@@ -353,9 +353,10 @@ where
             let mut slru_builder = SlruSegmentsBuilder::new(&mut self.ar);
 
             for part in slru_partitions.parts {
+                let query = VersionedKeySpaceQuery::uniform(part, self.lsn);
                 let blocks = self
                     .timeline
-                    .get_vectored(part, self.lsn, self.io_concurrency.clone(), self.ctx)
+                    .get_vectored(query, self.io_concurrency.clone(), self.ctx)
                     .await?;
 
                 for (key, block) in blocks {

pageserver/src/bin/pageserver.rs

@@ -20,7 +20,6 @@ use pageserver::config::{PageServerConf, PageserverIdentity, ignored_fields};
 use pageserver::controller_upcall_client::StorageControllerUpcallClient;
 use pageserver::deletion_queue::DeletionQueue;
 use pageserver::disk_usage_eviction_task::{self, launch_disk_usage_global_eviction_task};
-use pageserver::metrics::{STARTUP_DURATION, STARTUP_IS_LOADING};
 use pageserver::task_mgr::{
     BACKGROUND_RUNTIME, COMPUTE_REQUEST_RUNTIME, MGMT_REQUEST_RUNTIME, WALRECEIVER_RUNTIME,
 };
@@ -321,10 +320,9 @@ where
     }
 }
 
-fn startup_checkpoint(started_at: Instant, phase: &str, human_phase: &str) {
+fn startup_checkpoint(started_at: Instant, _phase: &str, human_phase: &str) {
     let elapsed = started_at.elapsed();
     let secs = elapsed.as_secs_f64();
-    STARTUP_DURATION.with_label_values(&[phase]).set(secs);
 
     info!(
         elapsed_ms = elapsed.as_millis(),
@@ -355,10 +353,7 @@ fn start_pageserver(
     set_launch_timestamp_metric(launch_ts);
     #[cfg(target_os = "linux")]
     metrics::register_internal(Box::new(metrics::more_process_metrics::Collector::new())).unwrap();
-    metrics::register_internal(Box::new(
-        pageserver::metrics::tokio_epoll_uring::Collector::new(),
-    ))
-    .unwrap();
+    
     pageserver::preinitialize_metrics(conf, ignored);
 
     // If any failpoints were set from FAILPOINTS environment variable,
@@ -452,6 +447,24 @@ fn start_pageserver(
     info!("Using auth for http API: {:#?}", conf.http_auth_type);
     info!("Using auth for pg connections: {:#?}", conf.pg_auth_type);
 
+    let tls_server_config = if conf.listen_https_addr.is_some() || conf.enable_tls_page_service_api
+    {
+        let resolver = BACKGROUND_RUNTIME.block_on(ReloadingCertificateResolver::new(
+            "main",
+            &conf.ssl_key_file,
+            &conf.ssl_cert_file,
+            conf.ssl_cert_reload_period,
+        ))?;
+
+        let server_config = rustls::ServerConfig::builder()
+            .with_no_client_auth()
+            .with_cert_resolver(resolver);
+
+        Some(Arc::new(server_config))
+    } else {
+        None
+    };
+
     match var("NEON_AUTH_TOKEN") {
         Ok(v) => {
             info!("Loaded JWT token for authentication with Safekeeper");
@@ -484,7 +497,6 @@ fn start_pageserver(
     // Up to this point no significant I/O has been done: this should have been fast.  Record
     // duration prior to starting I/O intensive phase of startup.
     startup_checkpoint(started_startup_at, "initial", "Starting loading tenants");
-    STARTUP_IS_LOADING.set(1);
 
     // Startup staging or optimizing:
     //
@@ -560,7 +572,6 @@ fn start_pageserver(
                     "initial_tenant_load",
                     "Initial load completed",
                 );
-                STARTUP_IS_LOADING.set(0);
             });
 
             let WaitForPhaseResult {
@@ -670,17 +681,11 @@ fn start_pageserver(
 
         let https_task = match https_listener {
             Some(https_listener) => {
-                let resolver = MGMT_REQUEST_RUNTIME.block_on(ReloadingCertificateResolver::new(
-                    &conf.ssl_key_file,
-                    &conf.ssl_cert_file,
-                    conf.ssl_cert_reload_period,
-                ))?;
+                let tls_server_config = tls_server_config
+                    .clone()
+                    .expect("tls_server_config is set earlier if https is enabled");
 
-                let server_config = rustls::ServerConfig::builder()
-                    .with_no_client_auth()
-                    .with_cert_resolver(resolver);
-
-                let tls_acceptor = tokio_rustls::TlsAcceptor::from(Arc::new(server_config));
+                let tls_acceptor = tokio_rustls::TlsAcceptor::from(tls_server_config);
 
                 let server =
                     http_utils::server::Server::new(service, https_listener, Some(tls_acceptor))?;
@@ -736,6 +741,11 @@ fn start_pageserver(
             tokio::net::TcpListener::from_std(pageserver_listener)
                 .context("create tokio listener")?
         },
+        if conf.enable_tls_page_service_api {
+            tls_server_config
+        } else {
+            None
+        },
     );
 
     // All started up! Now just sit and wait for shutdown signal.

pageserver/src/config.rs

@@ -219,6 +219,11 @@ pub struct PageServerConf {
     pub generate_unarchival_heatmap: bool,
 
     pub tracing: Option<pageserver_api::config::Tracing>,
+
+    /// Enable TLS in page service API.
+    /// Does not force TLS: the client negotiates TLS usage during the handshake.
+    /// Uses key and certificate from ssl_key_file/ssl_cert_file.
+    pub enable_tls_page_service_api: bool,
 }
 
 /// Token for authentication to safekeepers
@@ -391,6 +396,7 @@ impl PageServerConf {
             load_previous_heatmap,
             generate_unarchival_heatmap,
             tracing,
+            enable_tls_page_service_api,
         } = config_toml;
 
         let mut conf = PageServerConf {
@@ -441,6 +447,7 @@ impl PageServerConf {
             page_service_pipelining,
             get_vectored_concurrent_io,
             tracing,
+            enable_tls_page_service_api,
 
             // ------------------------------------------------------------
             // fields that require additional validation or custom handling

pageserver/src/consumption_metrics/metrics.rs

@@ -261,7 +261,7 @@ where
     let mut tenants = std::pin::pin!(tenants);
 
     while let Some((tenant_id, tenant)) = tenants.next().await {
-        let mut tenant_resident_size = 0;
+        let tenant_resident_size = 0;
 
         for timeline in tenant.list_timelines() {
             let timeline_id = timeline.timeline_id;
@@ -286,7 +286,6 @@ where
                 }
             }
 
-            tenant_resident_size += timeline.resident_physical_size();
         }
 
         let snap = TenantSnapshot::collect(&tenant, tenant_resident_size);

pageserver/src/context.rs

@@ -91,12 +91,12 @@
 
 use std::{sync::Arc, time::Duration};
 
-use once_cell::sync::Lazy;
+
 use tracing::warn;
 use utils::{id::TimelineId, shard::TenantShardId};
 
 use crate::{
-    metrics::{StorageIoSizeMetrics, TimelineMetrics},
+    metrics::TimelineMetrics,
     task_mgr::TaskKind,
     tenant::Timeline,
 };
@@ -122,38 +122,35 @@ pub struct RequestContext {
 #[derive(Clone)]
 pub(crate) enum Scope {
     Global {
-        io_size_metrics: &'static crate::metrics::StorageIoSizeMetrics,
+        
     },
     SecondaryTenant {
-        io_size_metrics: &'static crate::metrics::StorageIoSizeMetrics,
+       
     },
     SecondaryTimeline {
-        io_size_metrics: crate::metrics::StorageIoSizeMetrics,
+       
     },
     Timeline {
-        // We wrap the `Arc<TimelineMetrics>`s inside another Arc to avoid child
+       // We wrap the `Arc<TimelineMetrics>`s inside another Arc to avoid child
         // context creation contending for the ref counters of the Arc<TimelineMetrics>,
         // which are shared among all tasks that operate on the timeline, especially
         // concurrent page_service connections.
         #[allow(clippy::redundant_allocation)]
-        arc_arc: Arc<Arc<TimelineMetrics>>,
-    },
+        #[allow(dead_code)]
+        arc_arc: Arc<Arc<TimelineMetrics>>,    },
     #[cfg(test)]
     UnitTest {
-        io_size_metrics: &'static crate::metrics::StorageIoSizeMetrics,
+       
     },
     DebugTools {
-        io_size_metrics: &'static crate::metrics::StorageIoSizeMetrics,
+        
     },
 }
 
-static GLOBAL_IO_SIZE_METRICS: Lazy<crate::metrics::StorageIoSizeMetrics> =
-    Lazy::new(|| crate::metrics::StorageIoSizeMetrics::new("*", "*", "*"));
 
 impl Scope {
     pub(crate) fn new_global() -> Self {
         Scope::Global {
-            io_size_metrics: &GLOBAL_IO_SIZE_METRICS,
         }
     }
     /// NB: this allocates, so, use only at relatively long-lived roots, e.g., at start
@@ -173,18 +170,13 @@ impl Scope {
         }
     }
     pub(crate) fn new_secondary_timeline(
-        tenant_shard_id: &TenantShardId,
-        timeline_id: &TimelineId,
+        _tenant_shard_id: &TenantShardId,
+        _timeline_id: &TimelineId,
     ) -> Self {
         // TODO(https://github.com/neondatabase/neon/issues/11156): secondary timelines have no infrastructure for metrics lifecycle.
 
-        let tenant_id = tenant_shard_id.tenant_id.to_string();
-        let shard_id = tenant_shard_id.shard_slug().to_string();
-        let timeline_id = timeline_id.to_string();
 
-        let io_size_metrics =
-            crate::metrics::StorageIoSizeMetrics::new(&tenant_id, &shard_id, &timeline_id);
-        Scope::SecondaryTimeline { io_size_metrics }
+        Scope::SecondaryTimeline {  }
     }
     pub(crate) fn new_secondary_tenant(_tenant_shard_id: &TenantShardId) -> Self {
         // Before propagating metrics via RequestContext, the labels were inferred from file path.
@@ -197,19 +189,19 @@ impl Scope {
         // like we do for attached timelines. (We don't have attached-tenant-scoped usage of VirtualFile
         // at this point, so, we were able to completely side-step tenant-scoped stuff there).
         Scope::SecondaryTenant {
-            io_size_metrics: &GLOBAL_IO_SIZE_METRICS,
+           
         }
     }
     #[cfg(test)]
     pub(crate) fn new_unit_test() -> Self {
         Scope::UnitTest {
-            io_size_metrics: &GLOBAL_IO_SIZE_METRICS,
+          
         }
     }
 
     pub(crate) fn new_debug_tools() -> Self {
         Scope::DebugTools {
-            io_size_metrics: &GLOBAL_IO_SIZE_METRICS,
+          
         }
     }
 }
@@ -523,58 +515,18 @@ impl RequestContext {
         self.access_stats_behavior
     }
 
-    pub(crate) fn page_content_kind(&self) -> PageContentKind {
-        self.page_content_kind
-    }
 
     pub(crate) fn read_path_debug(&self) -> bool {
         self.read_path_debug
     }
 
-    pub(crate) fn io_size_metrics(&self) -> &StorageIoSizeMetrics {
-        match &self.scope {
-            Scope::Global { io_size_metrics } => {
-                let is_unit_test = cfg!(test);
-                let is_regress_test_build = cfg!(feature = "testing");
-                if is_unit_test || is_regress_test_build {
-                    panic!("all VirtualFile instances are timeline-scoped");
-                } else {
-                    use once_cell::sync::Lazy;
-                    use std::sync::Mutex;
-                    use std::time::Duration;
-                    use utils::rate_limit::RateLimit;
-                    static LIMIT: Lazy<Mutex<RateLimit>> =
-                        Lazy::new(|| Mutex::new(RateLimit::new(Duration::from_secs(1))));
-                    let mut guard = LIMIT.lock().unwrap();
-                    guard.call2(|rate_limit_stats| {
-                        warn!(
-                            %rate_limit_stats,
-                            backtrace=%std::backtrace::Backtrace::force_capture(),
-                            "all VirtualFile instances are timeline-scoped",
-                        );
-                    });
-
-                    io_size_metrics
-                }
-            }
-            Scope::Timeline { arc_arc } => &arc_arc.storage_io_size,
-            Scope::SecondaryTimeline { io_size_metrics } => io_size_metrics,
-            Scope::SecondaryTenant { io_size_metrics } => io_size_metrics,
-            #[cfg(test)]
-            Scope::UnitTest { io_size_metrics } => io_size_metrics,
-            Scope::DebugTools { io_size_metrics } => io_size_metrics,
-        }
-    }
-
     pub(crate) fn ondemand_download_wait_observe(&self, duration: Duration) {
         if duration == Duration::ZERO {
             return;
         }
 
         match &self.scope {
-            Scope::Timeline { arc_arc } => arc_arc
-                .wait_ondemand_download_time
-                .observe(self.task_kind, duration),
+            Scope::Timeline { arc_arc: _ } => {},
             _ => {
                 use once_cell::sync::Lazy;
                 use std::sync::Mutex;

pageserver/src/deletion_queue.rs

@@ -27,7 +27,6 @@ use self::list_writer::{DeletionOp, ListWriter, RecoverOp};
 use self::validator::Validator;
 use crate::config::PageServerConf;
 use crate::controller_upcall_client::StorageControllerUpcallApi;
-use crate::metrics;
 use crate::tenant::remote_timeline_client::{LayerFileMetadata, remote_timeline_path};
 use crate::tenant::storage_layer::LayerName;
 use crate::virtual_file::{MaybeFatalIo, VirtualFile};
@@ -163,11 +162,6 @@ struct TenantDeletionList {
     generation: Generation,
 }
 
-impl TenantDeletionList {
-    pub(crate) fn len(&self) -> usize {
-        self.timelines.values().map(|v| v.len()).sum()
-    }
-}
 
 /// Files ending with this suffix will be ignored and erased
 /// during recovery as startup.
@@ -467,9 +461,6 @@ impl DeletionQueueClient {
         // they may be historical.
         assert!(!current_generation.is_none());
 
-        metrics::DELETION_QUEUE
-            .keys_submitted
-            .inc_by(layers.len() as u64);
         self.do_push(
             &self.tx,
             ListWriterQueueMessage::Delete(DeletionOp {
@@ -553,9 +544,6 @@ impl DeletionQueueClient {
         &self,
         objects: Vec<RemotePath>,
     ) -> Result<(), DeletionQueueError> {
-        metrics::DELETION_QUEUE
-            .keys_submitted
-            .inc_by(objects.len() as u64);
         self.executor_tx
             .send(DeleterMessage::Delete(objects))
             .await

pageserver/src/deletion_queue/deleter.rs

@@ -14,7 +14,6 @@ use tracing::{info, warn};
 use utils::{backoff, pausable_failpoint};
 
 use super::{DeletionQueueError, FlushOp};
-use crate::metrics;
 
 const AUTOFLUSH_INTERVAL: Duration = Duration::from_secs(10);
 
@@ -60,10 +59,6 @@ impl Deleter {
                 fail::fail_point!("deletion-queue-before-execute", |_| {
                     info!("Skipping execution, failpoint set");
 
-                    metrics::DELETION_QUEUE
-                        .remote_errors
-                        .with_label_values(&["failpoint"])
-                        .inc();
                     Err(anyhow::anyhow!("failpoint: deletion-queue-before-execute"))
                 });
 
@@ -90,9 +85,6 @@ impl Deleter {
                 Ok(()) => {
                     // Note: we assume that the remote storage layer returns Ok(()) if some
                     // or all of the deleted objects were already gone.
-                    metrics::DELETION_QUEUE
-                        .keys_executed
-                        .inc_by(self.accumulator.len() as u64);
                     info!(
                         "Executed deletion batch {}..{}",
                         self.accumulator
@@ -109,10 +101,6 @@ impl Deleter {
                         return Err(DeletionQueueError::ShuttingDown);
                     }
                     warn!("DeleteObjects request failed: {e:#}, will continue trying");
-                    metrics::DELETION_QUEUE
-                        .remote_errors
-                        .with_label_values(&["execute"])
-                        .inc();
                 }
             };
         }

pageserver/src/deletion_queue/list_writer.rs

@@ -25,7 +25,6 @@ use utils::id::TimelineId;
 use super::{DeletionHeader, DeletionList, FlushOp, ValidatorQueueMessage};
 use crate::config::PageServerConf;
 use crate::deletion_queue::TEMP_SUFFIX;
-use crate::metrics;
 use crate::tenant::remote_timeline_client::{LayerFileMetadata, remote_layer_path};
 use crate::tenant::storage_layer::LayerName;
 use crate::virtual_file::{MaybeFatalIo, on_fatal_io_error};
@@ -152,7 +151,7 @@ impl ListWriter {
                 }
             }
             Err(e) => {
-                metrics::DELETION_QUEUE.unexpected_errors.inc();
+                
                 warn!(
                     sequence = self.pending.sequence,
                     "Failed to write deletion list, will retry later ({e:#})"
@@ -180,7 +179,6 @@ impl ListWriter {
                         // This should never happen unless we make a mistake with our serialization.
                         // Ignoring a deletion header is not consequential for correctnes because all deletions
                         // are ultimately allowed to fail: worst case we leak some objects for the scrubber to clean up.
-                        metrics::DELETION_QUEUE.unexpected_errors.inc();
                         Ok(None)
                     }
                 }
@@ -249,7 +247,6 @@ impl ListWriter {
                     .as_str()
             } else {
                 warn!("Unexpected key in deletion queue: {basename}");
-                metrics::DELETION_QUEUE.unexpected_errors.inc();
                 continue;
             };
 
@@ -257,7 +254,6 @@ impl ListWriter {
                 Ok(s) => s,
                 Err(e) => {
                     warn!("Malformed key '{basename}': {e}");
-                    metrics::DELETION_QUEUE.unexpected_errors.inc();
                     continue;
                 }
             };
@@ -286,7 +282,6 @@ impl ListWriter {
                     // Drop the list on the floor: any objects it referenced will be left behind
                     // for scrubbing to clean up.  This should never happen unless we have a serialization bug.
                     warn!(sequence = s, "Failed to deserialize deletion list: {e}");
-                    metrics::DELETION_QUEUE.unexpected_errors.inc();
                     continue;
                 }
             };
@@ -329,9 +324,6 @@ impl ListWriter {
 
             // We will drop out of recovery if this fails: it indicates that we are shutting down
             // or the backend has panicked
-            metrics::DELETION_QUEUE
-                .keys_submitted
-                .inc_by(deletion_list.len() as u64);
             self.tx
                 .send(ValidatorQueueMessage::Delete(deletion_list))
                 .await?;
@@ -353,7 +345,6 @@ impl ListWriter {
                 "Failed to create deletion list directory {}, deletions will not be executed ({e})",
                 self.conf.deletion_prefix(),
             );
-            metrics::DELETION_QUEUE.unexpected_errors.inc();
             return;
         }
 
@@ -422,7 +413,6 @@ impl ListWriter {
                             tracing::error!(
                                 "Failed to enqueue deletions, leaking objects.  This is a bug."
                             );
-                            metrics::DELETION_QUEUE.unexpected_errors.inc();
                         }
                     }
                 }
@@ -450,7 +440,6 @@ impl ListWriter {
                         tracing::error!(
                             "Deletion queue recovery called more than once.  This is a bug."
                         );
-                        metrics::DELETION_QUEUE.unexpected_errors.inc();
                         // Non-fatal: although this is a bug, since we did recovery at least once we may proceed.
                         continue;
                     }
@@ -462,7 +451,6 @@ impl ListWriter {
                         info!(
                             "Deletion queue recover aborted, deletion queue will not proceed ({e})"
                         );
-                        metrics::DELETION_QUEUE.unexpected_errors.inc();
                         return;
                     } else {
                         self.recovered = true;

pageserver/src/deletion_queue/validator.rs

@@ -26,7 +26,6 @@ use super::deleter::DeleterMessage;
 use super::{DeletionHeader, DeletionList, DeletionQueueError, FlushOp, VisibleLsnUpdates};
 use crate::config::PageServerConf;
 use crate::controller_upcall_client::{RetryForeverError, StorageControllerUpcallApi};
-use crate::metrics;
 use crate::virtual_file::MaybeFatalIo;
 
 // After this length of time, do any validation work that is pending,
@@ -186,7 +185,6 @@ where
                     "Dropped remote consistent LSN updates for tenant {tenant_id} in stale generation {:?}",
                     tenant_lsn_state.generation
                 );
-                metrics::DELETION_QUEUE.dropped_lsn_updates.inc();
             }
         }
 
@@ -221,11 +219,8 @@ where
 
                 if !this_list_valid {
                     info!("Dropping stale deletions for tenant {tenant_id} in generation {:?}, objects may be leaked", tenant.generation);
-                    metrics::DELETION_QUEUE.keys_dropped.inc_by(tenant.len() as u64);
                     mutated = true;
-                } else {
-                    metrics::DELETION_QUEUE.keys_validated.inc_by(tenant.len() as u64);
-                }
+                } 
                 this_list_valid
             });
             list.validated = true;
@@ -237,7 +232,7 @@ where
                     // Highly unexpected.  Could happen if e.g. disk full.
                     // If we didn't save the trimmed list, it is _not_ valid to execute.
                     warn!("Failed to save modified deletion list {list}: {e:#}");
-                    metrics::DELETION_QUEUE.unexpected_errors.inc();
+                    
 
                     // Rather than have a complex retry process, just drop it and leak the objects,
                     // scrubber will clean up eventually.
@@ -276,7 +271,7 @@ where
                 // The save() function logs a warning on error.
                 if let Err(e) = header.save(self.conf).await {
                     warn!("Failed to write deletion queue header: {e:#}");
-                    metrics::DELETION_QUEUE.unexpected_errors.inc();
+                    
                 }
             }
         }

pageserver/src/disk_usage_eviction_task.rs

@@ -56,7 +56,6 @@ use utils::completion;
 use utils::id::TimelineId;
 
 use crate::config::PageServerConf;
-use crate::metrics::disk_usage_based_eviction::METRICS;
 use crate::task_mgr::{self, BACKGROUND_RUNTIME};
 use crate::tenant::mgr::TenantManager;
 use crate::tenant::remote_timeline_client::LayerFileMetadata;
@@ -388,7 +387,7 @@ pub(crate) async fn disk_usage_eviction_task_iteration_impl<U: Usage>(
         }
     };
 
-    METRICS.layers_collected.inc_by(candidates.len() as u64);
+
 
     tracing::info!(
         elapsed_ms = collection_time.as_millis(),
@@ -428,7 +427,7 @@ pub(crate) async fn disk_usage_eviction_task_iteration_impl<U: Usage>(
     let (evicted_amount, usage_planned) =
         select_victims(&candidates, usage_pre).into_amount_and_planned();
 
-    METRICS.layers_selected.inc_by(evicted_amount as u64);
+ 
 
     // phase2: evict layers
 
@@ -457,7 +456,6 @@ pub(crate) async fn disk_usage_eviction_task_iteration_impl<U: Usage>(
             if let Some(next) = next {
                 match next {
                     Ok(Ok(file_size)) => {
-                        METRICS.layers_evicted.inc();
                         usage_assumed.add_available_bytes(file_size);
                     }
                     Ok(Err((
@@ -788,7 +786,6 @@ async fn collect_eviction_candidates(
     eviction_order: EvictionOrder,
     cancel: &CancellationToken,
 ) -> anyhow::Result<EvictionCandidates> {
-    const LOG_DURATION_THRESHOLD: std::time::Duration = std::time::Duration::from_secs(10);
 
     // get a snapshot of the list of tenants
     let tenants = tenant_manager
@@ -822,7 +819,7 @@ async fn collect_eviction_candidates(
             continue;
         }
 
-        let started_at = std::time::Instant::now();
+        
 
         // collect layers from all timelines in this tenant
         //
@@ -917,25 +914,11 @@ async fn collect_eviction_candidates(
                     (partition, candidate)
                 });
 
-        METRICS
-            .tenant_layer_count
-            .observe(tenant_candidates.len() as f64);
 
         candidates.extend(tenant_candidates);
 
-        let elapsed = started_at.elapsed();
-        METRICS
-            .tenant_collection_time
-            .observe(elapsed.as_secs_f64());
+       
 
-        if elapsed > LOG_DURATION_THRESHOLD {
-            tracing::info!(
-                tenant_id=%tenant.tenant_shard_id().tenant_id,
-                shard_id=%tenant.tenant_shard_id().shard_slug(),
-                elapsed_ms = elapsed.as_millis(),
-                "collection took longer than threshold"
-            );
-        }
     }
 
     // Note: the same tenant ID might be hit twice, if it transitions from attached to
@@ -962,7 +945,7 @@ async fn collect_eviction_candidates(
             layer_info.resident_layers.len()
         );
 
-        let started_at = std::time::Instant::now();
+       
 
         layer_info
             .resident_layers
@@ -984,28 +967,13 @@ async fn collect_eviction_candidates(
                         candidate,
                     )
                 });
-
-        METRICS
-            .tenant_layer_count
-            .observe(tenant_candidates.len() as f64);
         candidates.extend(tenant_candidates);
 
         tokio::task::yield_now().await;
 
-        let elapsed = started_at.elapsed();
+    
 
-        METRICS
-            .tenant_collection_time
-            .observe(elapsed.as_secs_f64());
-
-        if elapsed > LOG_DURATION_THRESHOLD {
-            tracing::info!(
-                tenant_id=%tenant.tenant_shard_id().tenant_id,
-                shard_id=%tenant.tenant_shard_id().shard_slug(),
-                elapsed_ms = elapsed.as_millis(),
-                "collection took longer than threshold"
-            );
-        }
+        
     }
 
     debug_assert!(

pageserver/src/http/routes.rs

@@ -3253,7 +3253,7 @@ async fn ingest_aux_files(
         modification
             .put_file(&fname, content.as_bytes(), &ctx)
             .await
-            .map_err(ApiError::InternalServerError)?;
+            .map_err(|e| ApiError::InternalServerError(e.into()))?;
     }
     modification
         .commit(&ctx)

pageserver/src/import_datadir.rs

@@ -24,10 +24,9 @@ use wal_decoder::models::InterpretedWalRecord;
 use walkdir::WalkDir;
 
 use crate::context::RequestContext;
-use crate::metrics::WAL_INGEST;
 use crate::pgdatadir_mapping::*;
 use crate::tenant::Timeline;
-use crate::walingest::WalIngest;
+use crate::walingest::{WalIngest, WalIngestErrorKind};
 
 // Returns checkpoint LSN from controlfile
 pub fn get_lsn_from_controlfile(path: &Utf8Path) -> Result<Lsn> {
@@ -157,9 +156,9 @@ async fn import_rel(
         .put_rel_creation(rel, nblocks as u32, ctx)
         .await
     {
-        match e {
-            RelationError::AlreadyExists => {
-                debug!("Relation {} already exist. We must be extending it.", rel)
+        match e.kind {
+            WalIngestErrorKind::RelationAlreadyExists(rel) => {
+                debug!("Relation {rel} already exists. We must be extending it.")
             }
             _ => return Err(e.into()),
         }
@@ -324,7 +323,6 @@ async fn import_wal(
                 walingest
                     .ingest_record(interpreted, &mut modification, ctx)
                     .await?;
-                WAL_INGEST.records_committed.inc();
 
                 modification.commit(ctx).await?;
                 last_lsn = lsn;

pageserver/src/page_cache.rs

@@ -77,7 +77,6 @@ use anyhow::Context;
 use once_cell::sync::OnceCell;
 
 use crate::context::RequestContext;
-use crate::metrics::{PageCacheSizeMetrics, page_cache_eviction_metrics};
 use crate::virtual_file::{IoBufferMut, IoPageSlice};
 
 static PAGE_CACHE: OnceCell<PageCache> = OnceCell::new();
@@ -195,7 +194,7 @@ impl SlotInner {
 }
 
 pub struct PageCache {
-    immutable_page_map: std::sync::RwLock<HashMap<(FileId, u32), usize>>,
+    immutable_page_maps: [std::sync::RwLock<HashMap<(FileId, u32), usize>>; 16],
 
     /// The actual buffers with their metadata.
     slots: Box<[Slot]>,
@@ -205,8 +204,103 @@ pub struct PageCache {
     /// Index of the next candidate to evict, for the Clock replacement algorithm.
     /// This is interpreted modulo the page cache size.
     next_evict_slot: AtomicUsize,
+}
 
-    size_metrics: &'static PageCacheSizeMetrics,
+impl PageCache {
+    /// Helper function to determine the shard index based on the low 4 bits of the u32 in the key tuple.
+    fn shard_index(_file_id: &FileId, blkno: u32) -> usize {
+        (blkno & 0xF) as usize
+    }
+
+    /// Search for a page in the cache using the given search key.
+    ///
+    /// Returns the slot index, if any.
+    ///
+    /// NOTE: We don't hold any lock on the mapping on return, so the slot might
+    /// get recycled for an unrelated page immediately after this function
+    /// returns. The caller is responsible for re-checking that the slot still
+    /// contains the page with the same key before using it.
+    ///
+    fn search_mapping(&self, cache_key: &CacheKey) -> Option<usize> {
+        match cache_key {
+            CacheKey::ImmutableFilePage { file_id, blkno } => {
+                let shard_idx = Self::shard_index(file_id, *blkno);
+                let map = self.immutable_page_maps[shard_idx].read().unwrap();
+                Some(*map.get(&(*file_id, *blkno))?)
+            }
+        }
+    }
+
+    ///
+    /// Remove mapping for given key.
+    ///
+    fn remove_mapping(&self, old_key: &CacheKey) {
+        match old_key {
+            CacheKey::ImmutableFilePage { file_id, blkno } => {
+                let shard_idx = Self::shard_index(file_id, *blkno);
+                let mut map = self.immutable_page_maps[shard_idx].write().unwrap();
+                map.remove(&(*file_id, *blkno))
+                    .expect("could not find old key in mapping");
+            }
+        }
+    }
+
+    ///
+    /// Insert mapping for given key.
+    ///
+    /// If a mapping already existed for the given key, returns the slot index
+    /// of the existing mapping and leaves it untouched.
+    fn try_insert_mapping(&self, new_key: &CacheKey, slot_idx: usize) -> Option<usize> {
+        match new_key {
+            CacheKey::ImmutableFilePage { file_id, blkno } => {
+                let shard_idx = Self::shard_index(file_id, *blkno);
+                let mut map = self.immutable_page_maps[shard_idx].write().unwrap();
+                match map.entry((*file_id, *blkno)) {
+                    Entry::Occupied(entry) => Some(*entry.get()),
+                    Entry::Vacant(entry) => {
+                        entry.insert(slot_idx);
+                        None
+                    }
+                }
+            }
+        }
+    }
+
+    /// Initialize a new page cache
+    ///
+    /// This should be called only once at page server startup.
+    fn new(num_pages: usize) -> Self {
+        assert!(num_pages > 0, "page cache size must be > 0");
+
+        // We could use Vec::leak here, but that potentially also leaks
+        // uninitialized reserved capacity. With into_boxed_slice and Box::leak
+        // this is avoided.
+        let page_buffer = IoBufferMut::with_capacity_zeroed(num_pages * PAGE_SZ).leak();
+
+        let slots = page_buffer
+            .chunks_exact_mut(PAGE_SZ)
+            .map(|chunk| {
+                // SAFETY: Each chunk has `PAGE_SZ` (8192) bytes, greater than 512, still aligned.
+                let buf = unsafe { IoPageSlice::new_unchecked(chunk.try_into().unwrap()) };
+
+                Slot {
+                    inner: tokio::sync::RwLock::new(SlotInner {
+                        key: None,
+                        buf,
+                        permit: std::sync::Mutex::new(Weak::new()),
+                    }),
+                    usage_count: AtomicU8::new(0),
+                }
+            })
+            .collect();
+
+        Self {
+            immutable_page_maps: Default::default(),
+            slots,
+            next_evict_slot: AtomicUsize::new(0),
+            pinned_slots: Arc::new(tokio::sync::Semaphore::new(num_pages)),
+        }
+    }
 }
 
 struct PinnedSlotsPermit {
@@ -414,32 +508,17 @@ impl PageCache {
     async fn lock_for_read(
         &self,
         cache_key: &CacheKey,
-        ctx: &RequestContext,
+        _ctx: &RequestContext,
     ) -> anyhow::Result<ReadBufResult> {
         let mut permit = Some(self.try_get_pinned_slot_permit().await?);
 
-        let (read_access, hit) = match cache_key {
-            CacheKey::ImmutableFilePage { .. } => (
-                &crate::metrics::PAGE_CACHE
-                    .for_ctx(ctx)
-                    .read_accesses_immutable,
-                &crate::metrics::PAGE_CACHE.for_ctx(ctx).read_hits_immutable,
-            ),
-        };
-        read_access.inc();
-
-        let mut is_first_iteration = true;
         loop {
             // First check if the key already exists in the cache.
             if let Some(read_guard) = self.try_lock_for_read(cache_key, &mut permit).await {
                 debug_assert!(permit.is_none());
-                if is_first_iteration {
-                    hit.inc();
-                }
                 return Ok(ReadBufResult::Found(read_guard));
             }
             debug_assert!(permit.is_some());
-            is_first_iteration = false;
 
             // Not found. Find a victim buffer
             let (slot_idx, mut inner) = self
@@ -484,63 +563,6 @@ impl PageCache {
         }
     }
 
-    //
-    // Section 3: Mapping functions
-    //
-
-    /// Search for a page in the cache using the given search key.
-    ///
-    /// Returns the slot index, if any.
-    ///
-    /// NOTE: We don't hold any lock on the mapping on return, so the slot might
-    /// get recycled for an unrelated page immediately after this function
-    /// returns.  The caller is responsible for re-checking that the slot still
-    /// contains the page with the same key before using it.
-    ///
-    fn search_mapping(&self, cache_key: &CacheKey) -> Option<usize> {
-        match cache_key {
-            CacheKey::ImmutableFilePage { file_id, blkno } => {
-                let map = self.immutable_page_map.read().unwrap();
-                Some(*map.get(&(*file_id, *blkno))?)
-            }
-        }
-    }
-
-    ///
-    /// Remove mapping for given key.
-    ///
-    fn remove_mapping(&self, old_key: &CacheKey) {
-        match old_key {
-            CacheKey::ImmutableFilePage { file_id, blkno } => {
-                let mut map = self.immutable_page_map.write().unwrap();
-                map.remove(&(*file_id, *blkno))
-                    .expect("could not find old key in mapping");
-                self.size_metrics.current_bytes_immutable.sub_page_sz(1);
-            }
-        }
-    }
-
-    ///
-    /// Insert mapping for given key.
-    ///
-    /// If a mapping already existed for the given key, returns the slot index
-    /// of the existing mapping and leaves it untouched.
-    fn try_insert_mapping(&self, new_key: &CacheKey, slot_idx: usize) -> Option<usize> {
-        match new_key {
-            CacheKey::ImmutableFilePage { file_id, blkno } => {
-                let mut map = self.immutable_page_map.write().unwrap();
-                match map.entry((*file_id, *blkno)) {
-                    Entry::Occupied(entry) => Some(*entry.get()),
-                    Entry::Vacant(entry) => {
-                        entry.insert(slot_idx);
-                        self.size_metrics.current_bytes_immutable.add_page_sz(1);
-                        None
-                    }
-                }
-            }
-        }
-    }
-
     //
     // Section 4: Misc internal helpers
     //
@@ -595,11 +617,7 @@ impl PageCache {
                             // Note that just yielding to tokio during iteration without such
                             // priority boosting is likely counter-productive. We'd just give more opportunities
                             // for B to bump usage count, further starving A.
-                            page_cache_eviction_metrics::observe(
-                                page_cache_eviction_metrics::Outcome::ItersExceeded {
-                                    iters: iters.try_into().unwrap(),
-                                },
-                            );
+                            
                             anyhow::bail!("exceeded evict iter limit");
                         }
                         continue;
@@ -609,84 +627,12 @@ impl PageCache {
                     // remove mapping for old buffer
                     self.remove_mapping(old_key);
                     inner.key = None;
-                    page_cache_eviction_metrics::observe(
-                        page_cache_eviction_metrics::Outcome::FoundSlotEvicted {
-                            iters: iters.try_into().unwrap(),
-                        },
-                    );
-                } else {
-                    page_cache_eviction_metrics::observe(
-                        page_cache_eviction_metrics::Outcome::FoundSlotUnused {
-                            iters: iters.try_into().unwrap(),
-                        },
-                    );
-                }
+                    
+                } 
                 return Ok((slot_idx, inner));
             }
         }
     }
 
-    /// Initialize a new page cache
-    ///
-    /// This should be called only once at page server startup.
-    fn new(num_pages: usize) -> Self {
-        assert!(num_pages > 0, "page cache size must be > 0");
-
-        // We could use Vec::leak here, but that potentially also leaks
-        // uninitialized reserved capacity. With into_boxed_slice and Box::leak
-        // this is avoided.
-        let page_buffer = IoBufferMut::with_capacity_zeroed(num_pages * PAGE_SZ).leak();
-
-        let size_metrics = &crate::metrics::PAGE_CACHE_SIZE;
-        size_metrics.max_bytes.set_page_sz(num_pages);
-        size_metrics.current_bytes_immutable.set_page_sz(0);
-
-        let slots = page_buffer
-            .chunks_exact_mut(PAGE_SZ)
-            .map(|chunk| {
-                // SAFETY: Each chunk has `PAGE_SZ` (8192) bytes, greater than 512, still aligned.
-                let buf = unsafe { IoPageSlice::new_unchecked(chunk.try_into().unwrap()) };
-
-                Slot {
-                    inner: tokio::sync::RwLock::new(SlotInner {
-                        key: None,
-                        buf,
-                        permit: std::sync::Mutex::new(Weak::new()),
-                    }),
-                    usage_count: AtomicU8::new(0),
-                }
-            })
-            .collect();
-
-        Self {
-            immutable_page_map: Default::default(),
-            slots,
-            next_evict_slot: AtomicUsize::new(0),
-            size_metrics,
-            pinned_slots: Arc::new(tokio::sync::Semaphore::new(num_pages)),
-        }
-    }
 }
 
-trait PageSzBytesMetric {
-    fn set_page_sz(&self, count: usize);
-    fn add_page_sz(&self, count: usize);
-    fn sub_page_sz(&self, count: usize);
-}
-
-#[inline(always)]
-fn count_times_page_sz(count: usize) -> u64 {
-    u64::try_from(count).unwrap() * u64::try_from(PAGE_SZ).unwrap()
-}
-
-impl PageSzBytesMetric for metrics::UIntGauge {
-    fn set_page_sz(&self, count: usize) {
-        self.set(count_times_page_sz(count));
-    }
-    fn add_page_sz(&self, count: usize) {
-        self.add(count_times_page_sz(count));
-    }
-    fn sub_page_sz(&self, count: usize) {
-        self.sub(count_times_page_sz(count));
-    }
-}

pageserver/src/page_service.rs

@@ -15,10 +15,11 @@ use async_compression::tokio::write::GzipEncoder;
 use bytes::Buf;
 use futures::FutureExt;
 use itertools::Itertools;
+use jsonwebtoken::TokenData;
 use once_cell::sync::OnceCell;
 use pageserver_api::config::{
     PageServicePipeliningConfig, PageServicePipeliningConfigPipelined,
-    PageServiceProtocolPipelinedExecutionStrategy,
+    PageServiceProtocolPipelinedBatchingStrategy, PageServiceProtocolPipelinedExecutionStrategy,
 };
 use pageserver_api::key::rel_block_to_key;
 use pageserver_api::models::{
@@ -58,8 +59,7 @@ use crate::context::{
     DownloadBehavior, PerfInstrumentFutureExt, RequestContext, RequestContextBuilder,
 };
 use crate::metrics::{
-    self, COMPUTE_COMMANDS_COUNTERS, ComputeCommandKind, LIVE_CONNECTIONS, SmgrOpTimer,
-    TimelineMetrics,
+    self, GetPageBatchBreakReason, SmgrOpTimer, TimelineMetrics,
 };
 use crate::pgdatadir_mapping::Version;
 use crate::span::{
@@ -105,6 +105,7 @@ pub fn spawn(
     pg_auth: Option<Arc<SwappableJwtAuth>>,
     perf_trace_dispatch: Option<Dispatch>,
     tcp_listener: tokio::net::TcpListener,
+    tls_config: Option<Arc<rustls::ServerConfig>>,
 ) -> Listener {
     let cancel = CancellationToken::new();
     let libpq_ctx = RequestContext::todo_child(
@@ -124,6 +125,7 @@ pub fn spawn(
             perf_trace_dispatch,
             tcp_listener,
             conf.pg_auth_type,
+            tls_config,
             conf.page_service_pipelining.clone(),
             libpq_ctx,
             cancel.clone(),
@@ -181,6 +183,7 @@ pub async fn libpq_listener_main(
     perf_trace_dispatch: Option<Dispatch>,
     listener: tokio::net::TcpListener,
     auth_type: AuthType,
+    tls_config: Option<Arc<rustls::ServerConfig>>,
     pipelining_config: PageServicePipeliningConfig,
     listener_ctx: RequestContext,
     listener_cancel: CancellationToken,
@@ -223,6 +226,7 @@ pub async fn libpq_listener_main(
                     local_auth,
                     socket,
                     auth_type,
+                    tls_config.clone(),
                     pipelining_config.clone(),
                     connection_ctx,
                     connections_cancel.child_token(),
@@ -264,14 +268,12 @@ async fn page_service_conn_main(
     auth: Option<Arc<SwappableJwtAuth>>,
     socket: tokio::net::TcpStream,
     auth_type: AuthType,
+    tls_config: Option<Arc<rustls::ServerConfig>>,
     pipelining_config: PageServicePipeliningConfig,
     connection_ctx: RequestContext,
     cancel: CancellationToken,
     gate_guard: GateGuard,
 ) -> ConnectionHandlerResult {
-    let _guard = LIVE_CONNECTIONS
-        .with_label_values(&["page_service"])
-        .guard();
 
     socket
         .set_nodelay(true)
@@ -334,7 +336,8 @@ async fn page_service_conn_main(
         cancel.clone(),
         gate_guard,
     );
-    let pgbackend = PostgresBackend::new_from_io(socket_fd, socket, peer_addr, auth_type, None)?;
+    let pgbackend =
+        PostgresBackend::new_from_io(socket_fd, socket, peer_addr, auth_type, tls_config)?;
 
     match pgbackend.run(&mut conn_handler, &cancel).await {
         Ok(()) => {
@@ -634,14 +637,13 @@ impl std::fmt::Display for BatchedPageStreamError {
 
 struct BatchedGetPageRequest {
     req: PagestreamGetPageRequest,
-    timer: SmgrOpTimer,
+    effective_request_lsn: Lsn,
     ctx: RequestContext,
 }
 
 #[cfg(feature = "testing")]
 struct BatchedTestRequest {
     req: models::PagestreamTestRequest,
-    timer: SmgrOpTimer,
 }
 
 /// NB: we only hold [`timeline::handle::WeakHandle`] inside this enum,
@@ -651,31 +653,31 @@ struct BatchedTestRequest {
 enum BatchedFeMessage {
     Exists {
         span: Span,
-        timer: SmgrOpTimer,
+       
         shard: timeline::handle::WeakHandle<TenantManagerTypes>,
         req: models::PagestreamExistsRequest,
     },
     Nblocks {
         span: Span,
-        timer: SmgrOpTimer,
+     
         shard: timeline::handle::WeakHandle<TenantManagerTypes>,
         req: models::PagestreamNblocksRequest,
     },
     GetPage {
         span: Span,
         shard: timeline::handle::WeakHandle<TenantManagerTypes>,
-        effective_request_lsn: Lsn,
         pages: smallvec::SmallVec<[BatchedGetPageRequest; 1]>,
+        batch_break_reason: GetPageBatchBreakReason,
     },
     DbSize {
         span: Span,
-        timer: SmgrOpTimer,
+  
         shard: timeline::handle::WeakHandle<TenantManagerTypes>,
         req: models::PagestreamDbSizeRequest,
     },
     GetSlruSegment {
         span: Span,
-        timer: SmgrOpTimer,
+   
         shard: timeline::handle::WeakHandle<TenantManagerTypes>,
         req: models::PagestreamGetSlruSegmentRequest,
     },
@@ -696,26 +698,119 @@ impl BatchedFeMessage {
         self.into()
     }
 
-    fn observe_execution_start(&mut self, at: Instant) {
-        match self {
-            BatchedFeMessage::Exists { timer, .. }
-            | BatchedFeMessage::Nblocks { timer, .. }
-            | BatchedFeMessage::DbSize { timer, .. }
-            | BatchedFeMessage::GetSlruSegment { timer, .. } => {
-                timer.observe_execution_start(at);
-            }
-            BatchedFeMessage::GetPage { pages, .. } => {
-                for page in pages {
-                    page.timer.observe_execution_start(at);
+    fn observe_execution_start(&mut self, _at: Instant) {
+    }
+
+    fn should_break_batch(
+        &self,
+        other: &BatchedFeMessage,
+        max_batch_size: NonZeroUsize,
+        batching_strategy: PageServiceProtocolPipelinedBatchingStrategy,
+    ) -> Option<GetPageBatchBreakReason> {
+        match (self, other) {
+            (
+                BatchedFeMessage::GetPage {
+                    shard: accum_shard,
+                    pages: accum_pages,
+                    ..
+                },
+                BatchedFeMessage::GetPage {
+                    shard: this_shard,
+                    pages: this_pages,
+                    ..
+                },
+            ) => {
+                assert_eq!(this_pages.len(), 1);
+                if accum_pages.len() >= max_batch_size.get() {
+                    trace!(%max_batch_size, "stopping batching because of batch size");
+                    assert_eq!(accum_pages.len(), max_batch_size.get());
+
+                    return Some(GetPageBatchBreakReason::BatchFull);
                 }
+                if !accum_shard.is_same_handle_as(this_shard) {
+                    trace!("stopping batching because timeline object mismatch");
+                    // TODO: we _could_ batch & execute each shard seperately (and in parallel).
+                    // But the current logic for keeping responses in order does not support that.
+
+                    return Some(GetPageBatchBreakReason::NonUniformTimeline);
+                }
+
+                match batching_strategy {
+                    PageServiceProtocolPipelinedBatchingStrategy::UniformLsn => {
+                        if let Some(last_in_batch) = accum_pages.last() {
+                            if last_in_batch.effective_request_lsn
+                                != this_pages[0].effective_request_lsn
+                            {
+                                trace!(
+                                    accum_lsn = %last_in_batch.effective_request_lsn,
+                                    this_lsn = %this_pages[0].effective_request_lsn,
+                                    "stopping batching because LSN changed"
+                                );
+
+                                return Some(GetPageBatchBreakReason::NonUniformLsn);
+                            }
+                        }
+                    }
+                    PageServiceProtocolPipelinedBatchingStrategy::ScatteredLsn => {
+                        // The read path doesn't curently support serving the same page at different LSNs.
+                        // While technically possible, it's uncertain if the complexity is worth it.
+                        // Break the batch if such a case is encountered.
+                        let same_page_different_lsn = accum_pages.iter().any(|batched| {
+                            batched.req.rel == this_pages[0].req.rel
+                                && batched.req.blkno == this_pages[0].req.blkno
+                                && batched.effective_request_lsn
+                                    != this_pages[0].effective_request_lsn
+                        });
+
+                        if same_page_different_lsn {
+                            trace!(
+                                rel=%this_pages[0].req.rel,
+                                blkno=%this_pages[0].req.blkno,
+                                lsn=%this_pages[0].effective_request_lsn,
+                                "stopping batching because same page was requested at different LSNs"
+                            );
+
+                            return Some(GetPageBatchBreakReason::SamePageAtDifferentLsn);
+                        }
+                    }
+                }
+
+                None
             }
             #[cfg(feature = "testing")]
-            BatchedFeMessage::Test { requests, .. } => {
-                for req in requests {
-                    req.timer.observe_execution_start(at);
+            (
+                BatchedFeMessage::Test {
+                    shard: accum_shard,
+                    requests: accum_requests,
+                    ..
+                },
+                BatchedFeMessage::Test {
+                    shard: this_shard,
+                    requests: this_requests,
+                    ..
+                },
+            ) => {
+                assert!(this_requests.len() == 1);
+                if accum_requests.len() >= max_batch_size.get() {
+                    trace!(%max_batch_size, "stopping batching because of batch size");
+                    assert_eq!(accum_requests.len(), max_batch_size.get());
+                    return Some(GetPageBatchBreakReason::BatchFull);
                 }
+                if !accum_shard.is_same_handle_as(this_shard) {
+                    trace!("stopping batching because timeline object mismatch");
+                    // TODO: we _could_ batch & execute each shard seperately (and in parallel).
+                    // But the current logic for keeping responses in order does not support that.
+                    return Some(GetPageBatchBreakReason::NonUniformTimeline);
+                }
+                let this_batch_key = this_requests[0].req.batch_key;
+                let accum_batch_key = accum_requests[0].req.batch_key;
+                if this_requests[0].req.batch_key != accum_requests[0].req.batch_key {
+                    trace!(%accum_batch_key, %this_batch_key, "stopping batching because batch key changed");
+                    return Some(GetPageBatchBreakReason::NonUniformKey);
+                }
+                None
             }
-            BatchedFeMessage::RespondError { .. } => {}
+            (_, _) => Some(GetPageBatchBreakReason::NonBatchableRequest),
         }
     }
 }
@@ -843,7 +938,7 @@ impl PageServerHandler {
                     .await?;
                 debug_assert_current_span_has_tenant_and_timeline_id_no_shard_id();
                 let span = tracing::info_span!(parent: &parent_span, "handle_get_rel_exists_request", rel = %req.rel, req_lsn = %req.hdr.request_lsn, shard_id = %shard.tenant_shard_id.shard_slug());
-                let timer = record_op_start_and_throttle(
+                record_op_start_and_throttle(
                     &shard,
                     metrics::SmgrQueryType::GetRelExists,
                     received_at,
@@ -851,7 +946,7 @@ impl PageServerHandler {
                 .await?;
                 BatchedFeMessage::Exists {
                     span,
-                    timer,
+                
                     shard: shard.downgrade(),
                     req,
                 }
@@ -861,7 +956,7 @@ impl PageServerHandler {
                     .get(tenant_id, timeline_id, ShardSelector::Zero)
                     .await?;
                 let span = tracing::info_span!(parent: &parent_span, "handle_get_nblocks_request", rel = %req.rel, req_lsn = %req.hdr.request_lsn, shard_id = %shard.tenant_shard_id.shard_slug());
-                let timer = record_op_start_and_throttle(
+                record_op_start_and_throttle(
                     &shard,
                     metrics::SmgrQueryType::GetRelSize,
                     received_at,
@@ -869,7 +964,7 @@ impl PageServerHandler {
                 .await?;
                 BatchedFeMessage::Nblocks {
                     span,
-                    timer,
+                   
                     shard: shard.downgrade(),
                     req,
                 }
@@ -879,7 +974,7 @@ impl PageServerHandler {
                     .get(tenant_id, timeline_id, ShardSelector::Zero)
                     .await?;
                 let span = tracing::info_span!(parent: &parent_span, "handle_db_size_request", dbnode = %req.dbnode, req_lsn = %req.hdr.request_lsn, shard_id = %shard.tenant_shard_id.shard_slug());
-                let timer = record_op_start_and_throttle(
+                record_op_start_and_throttle(
                     &shard,
                     metrics::SmgrQueryType::GetDbSize,
                     received_at,
@@ -887,7 +982,7 @@ impl PageServerHandler {
                 .await?;
                 BatchedFeMessage::DbSize {
                     span,
-                    timer,
+              
                     shard: shard.downgrade(),
                     req,
                 }
@@ -897,7 +992,7 @@ impl PageServerHandler {
                     .get(tenant_id, timeline_id, ShardSelector::Zero)
                     .await?;
                 let span = tracing::info_span!(parent: &parent_span, "handle_get_slru_segment_request", kind = %req.kind, segno = %req.segno, req_lsn = %req.hdr.request_lsn, shard_id = %shard.tenant_shard_id.shard_slug());
-                let timer = record_op_start_and_throttle(
+                record_op_start_and_throttle(
                     &shard,
                     metrics::SmgrQueryType::GetSlruSegment,
                     received_at,
@@ -905,7 +1000,7 @@ impl PageServerHandler {
                 .await?;
                 BatchedFeMessage::GetSlruSegment {
                     span,
-                    timer,
+               
                     shard: shard.downgrade(),
                     req,
                 }
@@ -1004,7 +1099,7 @@ impl PageServerHandler {
                 // request handler log messages contain the request-specific fields.
                 let span = mkspan!(shard.tenant_shard_id.shard_slug());
 
-                let timer = record_op_start_and_throttle(
+                record_op_start_and_throttle(
                     &shard,
                     metrics::SmgrQueryType::GetPageAtLsn,
                     received_at,
@@ -1019,34 +1114,31 @@ impl PageServerHandler {
                 .await?;
 
                 // We're holding the Handle
-                // TODO: if we actually need to wait for lsn here, it delays the entire batch which doesn't need to wait
-                let res = Self::wait_or_get_last_lsn(
+                let effective_request_lsn = match Self::effective_request_lsn(
                     &shard,
+                    shard.get_last_record_lsn(),
                     req.hdr.request_lsn,
                     req.hdr.not_modified_since,
                     &shard.get_applied_gc_cutoff_lsn(),
-                    &ctx,
-                )
-                .maybe_perf_instrument(&ctx, |current_perf_span| {
-                    info_span!(
-                        target: PERF_TRACE_TARGET,
-                        parent: current_perf_span,
-                        "WAIT_LSN",
-                    )
-                })
-                .await;
-
-                let effective_request_lsn = match res {
+                ) {
                     Ok(lsn) => lsn,
                     Err(e) => {
                         return respond_error!(span, e);
                     }
                 };
+
                 BatchedFeMessage::GetPage {
                     span,
                     shard: shard.downgrade(),
-                    effective_request_lsn,
-                    pages: smallvec::smallvec![BatchedGetPageRequest { req, timer, ctx }],
+                    pages: smallvec::smallvec![BatchedGetPageRequest {
+                        req,
+                        effective_request_lsn,
+                        ctx,
+                    }],
+                    // The executor grabs the batch when it becomes idle.
+                    // Hence, [`GetPageBatchBreakReason::ExecutorSteal`] is the
+                    // default reason for breaking the batch.
+                    batch_break_reason: GetPageBatchBreakReason::ExecutorSteal,
                 }
             }
             #[cfg(feature = "testing")]
@@ -1055,13 +1147,12 @@ impl PageServerHandler {
                     .get(tenant_id, timeline_id, ShardSelector::Zero)
                     .await?;
                 let span = tracing::info_span!(parent: &parent_span, "handle_test_request", shard_id = %shard.tenant_shard_id.shard_slug());
-                let timer =
-                    record_op_start_and_throttle(&shard, metrics::SmgrQueryType::Test, received_at)
+                record_op_start_and_throttle(&shard, metrics::SmgrQueryType::Test, received_at)
                         .await?;
                 BatchedFeMessage::Test {
                     span,
                     shard: shard.downgrade(),
-                    requests: vec![BatchedTestRequest { req, timer }],
+                    requests: vec![BatchedTestRequest { req,  }],
                 }
             }
         };
@@ -1072,6 +1163,7 @@ impl PageServerHandler {
     #[instrument(skip_all, level = tracing::Level::TRACE)]
     #[allow(clippy::boxed_local)]
     fn pagestream_do_batch(
+        batching_strategy: PageServiceProtocolPipelinedBatchingStrategy,
         max_batch_size: NonZeroUsize,
         batch: &mut Result<BatchedFeMessage, QueryError>,
         this_msg: Result<BatchedFeMessage, QueryError>,
@@ -1083,90 +1175,59 @@ impl PageServerHandler {
             Err(e) => return Err(Err(e)),
         };
 
-        match (&mut *batch, this_msg) {
-            // something batched already, let's see if we can add this message to the batch
-            (
-                Ok(BatchedFeMessage::GetPage {
-                    span: _,
-                    shard: accum_shard,
-                    pages: accum_pages,
-                    effective_request_lsn: accum_lsn,
-                }),
-                BatchedFeMessage::GetPage {
-                    span: _,
-                    shard: this_shard,
-                    pages: this_pages,
-                    effective_request_lsn: this_lsn,
-                },
-            ) if (|| {
-                assert_eq!(this_pages.len(), 1);
-                if accum_pages.len() >= max_batch_size.get() {
-                    trace!(%accum_lsn, %this_lsn, %max_batch_size, "stopping batching because of batch size");
-                    assert_eq!(accum_pages.len(), max_batch_size.get());
-                    return false;
-                }
-                if !accum_shard.is_same_handle_as(&this_shard) {
-                    trace!(%accum_lsn, %this_lsn, "stopping batching because timeline object mismatch");
-                    // TODO: we _could_ batch & execute each shard seperately (and in parallel).
-                    // But the current logic for keeping responses in order does not support that.
-                    return false;
-                }
-                // the vectored get currently only supports a single LSN, so, bounce as soon
-                // as the effective request_lsn changes
-                if *accum_lsn != this_lsn {
-                    trace!(%accum_lsn, %this_lsn, "stopping batching because LSN changed");
-                    return false;
-                }
-                true
-            })() =>
-            {
-                // ok to batch
-                accum_pages.extend(this_pages);
-                Ok(())
+        let eligible_batch = match batch {
+            Ok(b) => b,
+            Err(_) => {
+                return Err(Ok(this_msg));
             }
-            #[cfg(feature = "testing")]
-            (
-                Ok(BatchedFeMessage::Test {
-                    shard: accum_shard,
-                    requests: accum_requests,
-                    ..
-                }),
-                BatchedFeMessage::Test {
-                    shard: this_shard,
-                    requests: this_requests,
-                    ..
-                },
-            ) if (|| {
-                assert!(this_requests.len() == 1);
-                if accum_requests.len() >= max_batch_size.get() {
-                    trace!(%max_batch_size, "stopping batching because of batch size");
-                    assert_eq!(accum_requests.len(), max_batch_size.get());
-                    return false;
+        };
+
+        let batch_break =
+            eligible_batch.should_break_batch(&this_msg, max_batch_size, batching_strategy);
+
+        match batch_break {
+            Some(reason) => {
+                if let BatchedFeMessage::GetPage {
+                    batch_break_reason, ..
+                } = eligible_batch
+                {
+                    *batch_break_reason = reason;
                 }
-                if !accum_shard.is_same_handle_as(&this_shard) {
-                    trace!("stopping batching because timeline object mismatch");
-                    // TODO: we _could_ batch & execute each shard seperately (and in parallel).
-                    // But the current logic for keeping responses in order does not support that.
-                    return false;
-                }
-                let this_batch_key = this_requests[0].req.batch_key;
-                let accum_batch_key = accum_requests[0].req.batch_key;
-                if this_requests[0].req.batch_key != accum_requests[0].req.batch_key {
-                    trace!(%accum_batch_key, %this_batch_key, "stopping batching because batch key changed");
-                    return false;
-                }
-                true
-            })() =>
-            {
-                // ok to batch
-                accum_requests.extend(this_requests);
-                Ok(())
-            }
-            // something batched already but this message is unbatchable
-            (_, this_msg) => {
-                // by default, don't continue batching
+
                 Err(Ok(this_msg))
             }
+            None => {
+                // ok to batch
+                match (eligible_batch, this_msg) {
+                    (
+                        BatchedFeMessage::GetPage {
+                            pages: accum_pages, ..
+                        },
+                        BatchedFeMessage::GetPage {
+                            pages: this_pages, ..
+                        },
+                    ) => {
+                        accum_pages.extend(this_pages);
+                        Ok(())
+                    }
+                    #[cfg(feature = "testing")]
+                    (
+                        BatchedFeMessage::Test {
+                            requests: accum_requests,
+                            ..
+                        },
+                        BatchedFeMessage::Test {
+                            requests: this_requests,
+                            ..
+                        },
+                    ) => {
+                        accum_requests.extend(this_requests);
+                        Ok(())
+                    }
+                    // Shape guaranteed by [`BatchedFeMessage::should_break_batch`]
+                    _ => unreachable!(),
+                }
+            }
         }
     }
 
@@ -1192,7 +1253,7 @@ impl PageServerHandler {
 
         // Dispatch the batch to the appropriate request handler.
         let log_slow_name = batch.as_static_str();
-        let (mut handler_results, span) = {
+        let (handler_results, span) = {
             // TODO: we unfortunately have to pin the future on the heap, since GetPage futures are huge and
             // won't fit on the stack.
             let mut boxpinned =
@@ -1222,31 +1283,31 @@ impl PageServerHandler {
         // call, which (all unmeasured) adds syscall overhead but reduces time to first byte
         // and avoids building up a "giant" contiguous userspace buffer to hold the entire response.
         // TODO: vectored socket IO would be great, but pgb_writer doesn't support that.
-        let flush_timers = {
-            let flushing_start_time = Instant::now();
-            let mut flush_timers = Vec::with_capacity(handler_results.len());
-            for handler_result in &mut handler_results {
-                let flush_timer = match handler_result {
-                    Ok((_, timer)) => Some(
-                        timer
-                            .observe_execution_end(flushing_start_time)
-                            .expect("we are the first caller"),
-                    ),
-                    Err(_) => {
-                        // TODO: measure errors
-                        None
-                    }
-                };
-                flush_timers.push(flush_timer);
-            }
-            assert_eq!(flush_timers.len(), handler_results.len());
-            flush_timers
-        };
+        // let flush_timers = {
+        //     let flushing_start_time = Instant::now();
+        //     let mut flush_timers = Vec::with_capacity(handler_results.len());
+        //     for handler_result in &mut handler_results {
+        //         let flush_timer = match handler_result {
+        //             Ok((_, timer)) => Some(
+        //                 timer
+        //                     .observe_execution_end(flushing_start_time)
+        //                     .expect("we are the first caller"),
+        //             ),
+        //             Err(_) => {
+        //                 // TODO: measure errors
+        //                 None
+        //             }
+        //         };
+        //         flush_timers.push(flush_timer);
+        //     }
+        //     assert_eq!(flush_timers.len(), handler_results.len());
+        //     flush_timers
+        // };
 
         // Map handler result to protocol behavior.
         // Some handler errors cause exit from pagestream protocol.
         // Other handler errors are sent back as an error message and we stay in pagestream protocol.
-        for (handler_result, flushing_timer) in handler_results.into_iter().zip(flush_timers) {
+        for handler_result in handler_results.into_iter() {
             let response_msg = match handler_result {
                 Err(e) => match &e.err {
                     PageStreamError::Shutdown => {
@@ -1278,7 +1339,7 @@ impl PageServerHandler {
                         })
                     }
                 },
-                Ok((response_msg, _op_timer_already_observed)) => response_msg,
+                Ok((response_msg, )) => response_msg,
             };
 
             //
@@ -1292,17 +1353,17 @@ impl PageServerHandler {
             failpoint_support::sleep_millis_async!("before-pagestream-msg-flush", cancel);
 
             // what we want to do
-            let socket_fd = pgb_writer.socket_fd;
+           
             let flush_fut = pgb_writer.flush();
             // metric for how long flushing takes
-            let flush_fut = match flushing_timer {
-                Some(flushing_timer) => futures::future::Either::Left(flushing_timer.measure(
-                    Instant::now(),
-                    flush_fut,
-                    socket_fd,
-                )),
-                None => futures::future::Either::Right(flush_fut),
-            };
+            // let flush_fut = match flushing_timer {
+            //     Some(flushing_timer) => futures::future::Either::Left(flushing_timer.measure(
+            //         Instant::now(),
+            //         flush_fut,
+            //         socket_fd,
+            //     )),
+            //     None => futures::future::Either::Right(flush_fut),
+            // };
             // do it while respecting cancellation
             let _: () = async move {
                 tokio::select! {
@@ -1332,7 +1393,7 @@ impl PageServerHandler {
         ctx: &RequestContext,
     ) -> Result<
         (
-            Vec<Result<(PagestreamBeMessage, SmgrOpTimer), BatchedPageStreamError>>,
+            Vec<Result<(PagestreamBeMessage, ), BatchedPageStreamError>>,
             Span,
         ),
         QueryError,
@@ -1348,7 +1409,7 @@ impl PageServerHandler {
         Ok(match batch {
             BatchedFeMessage::Exists {
                 span,
-                timer,
+               
                 shard,
                 req,
             } => {
@@ -1359,7 +1420,7 @@ impl PageServerHandler {
                         self.handle_get_rel_exists_request(&shard, &req, &ctx)
                             .instrument(span.clone())
                             .await
-                            .map(|msg| (msg, timer))
+                            .map(|msg| (msg, ))
                             .map_err(|err| BatchedPageStreamError { err, req: req.hdr }),
                     ],
                     span,
@@ -1367,7 +1428,7 @@ impl PageServerHandler {
             }
             BatchedFeMessage::Nblocks {
                 span,
-                timer,
+           
                 shard,
                 req,
             } => {
@@ -1378,7 +1439,7 @@ impl PageServerHandler {
                         self.handle_get_nblocks_request(&shard, &req, &ctx)
                             .instrument(span.clone())
                             .await
-                            .map(|msg| (msg, timer))
+                            .map(|msg| (msg, ))
                             .map_err(|err| BatchedPageStreamError { err, req: req.hdr }),
                     ],
                     span,
@@ -1387,8 +1448,8 @@ impl PageServerHandler {
             BatchedFeMessage::GetPage {
                 span,
                 shard,
-                effective_request_lsn,
                 pages,
+                batch_break_reason,
             } => {
                 fail::fail_point!("ps::handle-pagerequest-message::getpage");
                 let (shard, ctx) = upgrade_handle_and_set_context!(shard);
@@ -1399,9 +1460,9 @@ impl PageServerHandler {
                         let res = self
                             .handle_get_page_at_lsn_request_batched(
                                 &shard,
-                                effective_request_lsn,
                                 pages,
                                 io_concurrency,
+                                batch_break_reason,
                                 &ctx,
                             )
                             .instrument(span.clone())
@@ -1414,7 +1475,6 @@ impl PageServerHandler {
             }
             BatchedFeMessage::DbSize {
                 span,
-                timer,
                 shard,
                 req,
             } => {
@@ -1425,7 +1485,7 @@ impl PageServerHandler {
                         self.handle_db_size_request(&shard, &req, &ctx)
                             .instrument(span.clone())
                             .await
-                            .map(|msg| (msg, timer))
+                            .map(|msg| (msg, ))
                             .map_err(|err| BatchedPageStreamError { err, req: req.hdr }),
                     ],
                     span,
@@ -1433,7 +1493,6 @@ impl PageServerHandler {
             }
             BatchedFeMessage::GetSlruSegment {
                 span,
-                timer,
                 shard,
                 req,
             } => {
@@ -1444,7 +1503,7 @@ impl PageServerHandler {
                         self.handle_get_slru_segment_request(&shard, &req, &ctx)
                             .instrument(span.clone())
                             .await
-                            .map(|msg| (msg, timer))
+                            .map(|msg| (msg, ))
                             .map_err(|err| BatchedPageStreamError { err, req: req.hdr }),
                     ],
                     span,
@@ -1718,6 +1777,7 @@ impl PageServerHandler {
         let PageServicePipeliningConfigPipelined {
             max_batch_size,
             execution,
+            batching: batching_strategy,
         } = pipelining_config;
 
         // Macro to _define_ a pipeline stage.
@@ -1769,7 +1829,7 @@ impl PageServerHandler {
                     exit |= read_res.is_err();
                     let could_send = batch_tx
                         .send(read_res, |batch, res| {
-                            Self::pagestream_do_batch(max_batch_size, batch, res)
+                            Self::pagestream_do_batch(batching_strategy, max_batch_size, batch, res)
                         })
                         .await;
                     exit |= could_send.is_err();
@@ -1865,7 +1925,39 @@ impl PageServerHandler {
         ctx: &RequestContext,
     ) -> Result<Lsn, PageStreamError> {
         let last_record_lsn = timeline.get_last_record_lsn();
+        let effective_request_lsn = Self::effective_request_lsn(
+            timeline,
+            last_record_lsn,
+            request_lsn,
+            not_modified_since,
+            latest_gc_cutoff_lsn,
+        )?;
 
+        if effective_request_lsn > last_record_lsn {
+            timeline
+                .wait_lsn(
+                    not_modified_since,
+                    crate::tenant::timeline::WaitLsnWaiter::PageService,
+                    timeline::WaitLsnTimeout::Default,
+                    ctx,
+                )
+                .await?;
+
+            // Since we waited for 'effective_request_lsn' to arrive, that is now the last
+            // record LSN. (Or close enough for our purposes; the last-record LSN can
+            // advance immediately after we return anyway)
+        }
+
+        Ok(effective_request_lsn)
+    }
+
+    fn effective_request_lsn(
+        timeline: &Timeline,
+        last_record_lsn: Lsn,
+        request_lsn: Lsn,
+        not_modified_since: Lsn,
+        latest_gc_cutoff_lsn: &RcuReadGuard<Lsn>,
+    ) -> Result<Lsn, PageStreamError> {
         // Sanity check the request
         if request_lsn < not_modified_since {
             return Err(PageStreamError::BadRequest(
@@ -1900,19 +1992,7 @@ impl PageServerHandler {
             }
         }
 
-        // Wait for WAL up to 'not_modified_since' to arrive, if necessary
         if not_modified_since > last_record_lsn {
-            timeline
-                .wait_lsn(
-                    not_modified_since,
-                    crate::tenant::timeline::WaitLsnWaiter::PageService,
-                    timeline::WaitLsnTimeout::Default,
-                    ctx,
-                )
-                .await?;
-            // Since we waited for 'not_modified_since' to arrive, that is now the last
-            // record LSN. (Or close enough for our purposes; the last-record LSN can
-            // advance immediately after we return anyway)
             Ok(not_modified_since)
         } else {
             // It might be better to use max(not_modified_since, latest_gc_cutoff_lsn)
@@ -2067,17 +2147,13 @@ impl PageServerHandler {
     async fn handle_get_page_at_lsn_request_batched(
         &mut self,
         timeline: &Timeline,
-        effective_lsn: Lsn,
         requests: smallvec::SmallVec<[BatchedGetPageRequest; 1]>,
         io_concurrency: IoConcurrency,
+        _batch_break_reason: GetPageBatchBreakReason,
         ctx: &RequestContext,
-    ) -> Vec<Result<(PagestreamBeMessage, SmgrOpTimer), BatchedPageStreamError>> {
+    ) -> Vec<Result<(PagestreamBeMessage, ), BatchedPageStreamError>> {
         debug_assert_current_span_has_tenant_and_timeline_id();
 
-        timeline
-            .query_metrics
-            .observe_getpage_batch_start(requests.len());
-
         // If a page trace is running, submit an event for this request.
         if let Some(page_trace) = timeline.page_trace.load().as_ref() {
             let time = SystemTime::now();
@@ -2086,20 +2162,81 @@ impl PageServerHandler {
                 // Ignore error (trace buffer may be full or tracer may have disconnected).
                 _ = page_trace.try_send(PageTraceEvent {
                     key,
-                    effective_lsn,
+                    effective_lsn: batch.effective_request_lsn,
                     time,
                 });
             }
         }
 
+        // If any request in the batch needs to wait for LSN, then do so now.
+        let mut perf_instrument = false;
+        let max_effective_lsn = requests
+            .iter()
+            .map(|req| {
+                if req.ctx.has_perf_span() {
+                    perf_instrument = true;
+                }
+
+                req.effective_request_lsn
+            })
+            .max()
+            .expect("batch is never empty");
+
+        let ctx = match perf_instrument {
+            true => RequestContextBuilder::from(ctx)
+                .root_perf_span(|| {
+                    info_span!(
+                        target: PERF_TRACE_TARGET,
+                        "GET_VECTORED",
+                        tenant_id = %timeline.tenant_shard_id.tenant_id,
+                        timeline_id = %timeline.timeline_id,
+                        shard = %timeline.tenant_shard_id.shard_slug(),
+                        %max_effective_lsn
+                    )
+                })
+                .attached_child(),
+            false => ctx.attached_child(),
+        };
+
+        let last_record_lsn = timeline.get_last_record_lsn();
+        if max_effective_lsn > last_record_lsn {
+            if let Err(e) = timeline
+                .wait_lsn(
+                    max_effective_lsn,
+                    crate::tenant::timeline::WaitLsnWaiter::PageService,
+                    timeline::WaitLsnTimeout::Default,
+                    &ctx,
+                )
+                .maybe_perf_instrument(&ctx, |current_perf_span| {
+                    info_span!(
+                        target: PERF_TRACE_TARGET,
+                        parent: current_perf_span,
+                        "WAIT_LSN",
+                    )
+                })
+                .await
+            {
+                return Vec::from_iter(requests.into_iter().map(|req| {
+                    Err(BatchedPageStreamError {
+                        err: PageStreamError::from(e.clone()),
+                        req: req.req.hdr,
+                    })
+                }));
+            }
+        }
+
         let results = timeline
             .get_rel_page_at_lsn_batched(
-                requests
-                    .iter()
-                    .map(|p| (&p.req.rel, &p.req.blkno, p.ctx.attached_child())),
-                effective_lsn,
+                requests.iter().map(|p| {
+                    (
+                        &p.req.rel,
+                        &p.req.blkno,
+                        p.effective_request_lsn,
+                        p.ctx.attached_child(),
+                    )
+                }),
                 io_concurrency,
-                ctx,
+                &ctx,
             )
             .await;
         assert_eq!(results.len(), requests.len());
@@ -2116,7 +2253,7 @@ impl PageServerHandler {
                                 req: req.req,
                                 page,
                             }),
-                            req.timer,
+                            
                         )
                     })
                     .map_err(|e| BatchedPageStreamError {
@@ -2161,7 +2298,7 @@ impl PageServerHandler {
         timeline: &Timeline,
         requests: Vec<BatchedTestRequest>,
         _ctx: &RequestContext,
-    ) -> Vec<Result<(PagestreamBeMessage, SmgrOpTimer), BatchedPageStreamError>> {
+    ) -> Vec<Result<(PagestreamBeMessage,), BatchedPageStreamError>> {
         // real requests would do something with the timeline
         let mut results = Vec::with_capacity(requests.len());
         for _req in requests.iter() {
@@ -2187,7 +2324,6 @@ impl PageServerHandler {
                             PagestreamBeMessage::Test(models::PagestreamTestResponse {
                                 req: req.req.clone(),
                             }),
-                            req.timer,
                         )
                     })
                     .map_err(|e| BatchedPageStreamError {
@@ -2667,7 +2803,7 @@ where
     ) -> Result<(), QueryError> {
         // this unwrap is never triggered, because check_auth_jwt only called when auth_type is NeonJWT
         // which requires auth to be present
-        let data = self
+        let data: TokenData<Claims> = self
             .auth
             .as_ref()
             .unwrap()
@@ -2742,12 +2878,7 @@ where
                     .record("timeline_id", field::display(timeline_id));
 
                 self.check_permission(Some(tenant_id))?;
-                let command_kind = match protocol_version {
-                    PagestreamProtocolVersion::V2 => ComputeCommandKind::PageStreamV2,
-                    PagestreamProtocolVersion::V3 => ComputeCommandKind::PageStreamV3,
-                };
-                COMPUTE_COMMANDS_COUNTERS.for_command(command_kind).inc();
-
+                
                 self.handle_pagerequests(pgb, tenant_id, timeline_id, protocol_version, ctx)
                     .await?;
             }
@@ -2764,10 +2895,7 @@ where
 
                 self.check_permission(Some(tenant_id))?;
 
-                COMPUTE_COMMANDS_COUNTERS
-                    .for_command(ComputeCommandKind::Basebackup)
-                    .inc();
-                let metric_recording = metrics::BASEBACKUP_QUERY_TIME.start_recording();
+                
                 let res = async {
                     self.handle_basebackup_request(
                         pgb,
@@ -2785,7 +2913,7 @@ where
                     Result::<(), QueryError>::Ok(())
                 }
                 .await;
-                metric_recording.observe(&res);
+
                 res?;
             }
             // same as basebackup, but result includes relational data as well
@@ -2801,9 +2929,7 @@ where
 
                 self.check_permission(Some(tenant_id))?;
 
-                COMPUTE_COMMANDS_COUNTERS
-                    .for_command(ComputeCommandKind::Fullbackup)
-                    .inc();
+                
 
                 // Check that the timeline exists
                 self.handle_basebackup_request(
@@ -2837,9 +2963,7 @@ where
 
                 self.check_permission(Some(tenant_shard_id.tenant_id))?;
 
-                COMPUTE_COMMANDS_COUNTERS
-                    .for_command(ComputeCommandKind::LeaseLsn)
-                    .inc();
+                
 
                 match self
                     .handle_make_lsn_lease(pgb, tenant_shard_id, timeline_id, lsn, &ctx)

pageserver/src/pgdatadir_mapping.rs

@@ -6,14 +6,14 @@
 //! walingest.rs handles a few things like implicit relation creation and extension.
 //! Clarify that)
 //!
-use std::collections::{BTreeMap, HashMap, HashSet, hash_map};
+use std::collections::{HashMap, HashSet, hash_map};
 use std::ops::{ControlFlow, Range};
 
-use crate::PERF_TRACE_TARGET;
-use anyhow::{Context, ensure};
+use crate::walingest::{WalIngestError, WalIngestErrorKind};
+use crate::{PERF_TRACE_TARGET, ensure_walingest};
+use anyhow::Context;
 use bytes::{Buf, Bytes, BytesMut};
 use enum_map::Enum;
-use itertools::Itertools;
 use pageserver_api::key::{
     AUX_FILES_KEY, CHECKPOINT_KEY, CONTROLFILE_KEY, CompactKey, DBDIR_KEY, Key, RelDirExists,
     TWOPHASEDIR_KEY, dbdir_key_range, rel_block_to_key, rel_dir_to_key, rel_key_range,
@@ -21,7 +21,7 @@ use pageserver_api::key::{
     repl_origin_key, repl_origin_key_range, slru_block_to_key, slru_dir_to_key,
     slru_segment_key_range, slru_segment_size_to_key, twophase_file_key, twophase_key_range,
 };
-use pageserver_api::keyspace::SparseKeySpace;
+use pageserver_api::keyspace::{KeySpaceRandomAccum, SparseKeySpace};
 use pageserver_api::models::RelSizeMigration;
 use pageserver_api::record::NeonWalRecord;
 use pageserver_api::reltag::{BlockNumber, RelTag, SlruKind};
@@ -36,21 +36,19 @@ use tracing::{debug, info, info_span, trace, warn};
 use utils::bin_ser::{BeSer, DeserializeError};
 use utils::lsn::Lsn;
 use utils::pausable_failpoint;
-use wal_decoder::serialized_batch::{SerializedValueBatch, ValueMeta};
+use wal_decoder::serialized_batch::SerializedValueBatch ;
 
 use super::tenant::{PageReconstructError, Timeline};
 use crate::aux_file;
-use crate::context::{PerfInstrumentFutureExt, RequestContext, RequestContextBuilder};
+use crate::context::{PerfInstrumentFutureExt, RequestContext};
 use crate::keyspace::{KeySpace, KeySpaceAccum};
-use crate::metrics::{
-    RELSIZE_CACHE_ENTRIES, RELSIZE_CACHE_HITS, RELSIZE_CACHE_MISSES, RELSIZE_CACHE_MISSES_OLD,
-};
+
 use crate::span::{
     debug_assert_current_span_has_tenant_and_timeline_id,
     debug_assert_current_span_has_tenant_and_timeline_id_no_shard_id,
 };
 use crate::tenant::storage_layer::IoConcurrency;
-use crate::tenant::timeline::GetVectoredError;
+use crate::tenant::timeline::{GetVectoredError, VersionedKeySpaceQuery};
 
 /// Max delta records appended to the AUX_FILES_KEY (for aux v1). The write path will write a full image once this threshold is reached.
 pub const MAX_AUX_FILE_DELTAS: usize = 1024;
@@ -136,12 +134,8 @@ impl From<PageReconstructError> for CalculateLogicalSizeError {
 
 #[derive(Debug, thiserror::Error)]
 pub enum RelationError {
-    #[error("Relation Already Exists")]
-    AlreadyExists,
     #[error("invalid relnode")]
     InvalidRelnode,
-    #[error(transparent)]
-    Other(#[from] anyhow::Error),
 }
 
 ///
@@ -210,10 +204,9 @@ impl Timeline {
                 let pages: smallvec::SmallVec<[_; 1]> = smallvec::smallvec![(tag, blknum)];
                 let res = self
                     .get_rel_page_at_lsn_batched(
-                        pages
-                            .iter()
-                            .map(|(tag, blknum)| (tag, blknum, ctx.attached_child())),
-                        effective_lsn,
+                        pages.iter().map(|(tag, blknum)| {
+                            (tag, blknum, effective_lsn, ctx.attached_child())
+                        }),
                         io_concurrency.clone(),
                         ctx,
                     )
@@ -251,8 +244,7 @@ impl Timeline {
     /// The ordering of the returned vec corresponds to the ordering of `pages`.
     pub(crate) async fn get_rel_page_at_lsn_batched(
         &self,
-        pages: impl ExactSizeIterator<Item = (&RelTag, &BlockNumber, RequestContext)>,
-        effective_lsn: Lsn,
+        pages: impl ExactSizeIterator<Item = (&RelTag, &BlockNumber, Lsn, RequestContext)>,
         io_concurrency: IoConcurrency,
         ctx: &RequestContext,
     ) -> Vec<Result<Bytes, PageReconstructError>> {
@@ -265,11 +257,13 @@ impl Timeline {
         let mut result = Vec::with_capacity(pages.len());
         let result_slots = result.spare_capacity_mut();
 
-        let mut keys_slots: BTreeMap<Key, smallvec::SmallVec<[(usize, RequestContext); 1]>> =
-            BTreeMap::default();
+        let mut keys_slots: HashMap<Key, smallvec::SmallVec<[(usize, RequestContext); 1]>> =
+            HashMap::with_capacity(pages.len());
 
-        let mut perf_instrument = false;
-        for (response_slot_idx, (tag, blknum, ctx)) in pages.enumerate() {
+        let mut req_keyspaces: HashMap<Lsn, KeySpaceRandomAccum> =
+            HashMap::with_capacity(pages.len());
+
+        for (response_slot_idx, (tag, blknum, lsn, ctx)) in pages.enumerate() {
             if tag.relnode == 0 {
                 result_slots[response_slot_idx].write(Err(PageReconstructError::Other(
                     RelationError::InvalidRelnode.into(),
@@ -280,14 +274,14 @@ impl Timeline {
             }
 
             let nblocks = match self
-                .get_rel_size(*tag, Version::Lsn(effective_lsn), &ctx)
+                .get_rel_size(*tag, Version::Lsn(lsn), &ctx)
                 .maybe_perf_instrument(&ctx, |crnt_perf_span| {
                     info_span!(
                         target: PERF_TRACE_TARGET,
                         parent: crnt_perf_span,
                         "GET_REL_SIZE",
                         reltag=%tag,
-                        lsn=%effective_lsn,
+                        lsn=%lsn,
                     )
                 })
                 .await
@@ -303,7 +297,7 @@ impl Timeline {
             if *blknum >= nblocks {
                 debug!(
                     "read beyond EOF at {} blk {} at {}, size is {}: returning all-zeros page",
-                    tag, blknum, effective_lsn, nblocks
+                    tag, blknum, lsn, nblocks
                 );
                 result_slots[response_slot_idx].write(Ok(ZERO_PAGE.clone()));
                 slots_filled += 1;
@@ -312,46 +306,29 @@ impl Timeline {
 
             let key = rel_block_to_key(*tag, *blknum);
 
-            if ctx.has_perf_span() {
-                perf_instrument = true;
-            }
-
             let key_slots = keys_slots.entry(key).or_default();
             key_slots.push((response_slot_idx, ctx));
+
+            let acc = req_keyspaces.entry(lsn).or_default();
+            acc.add_key(key);
         }
 
-        let keyspace = {
-            // add_key requires monotonicity
-            let mut acc = KeySpaceAccum::new();
-            for key in keys_slots
-                .keys()
-                // in fact it requires strong monotonicity
-                .dedup()
-            {
-                acc.add_key(*key);
-            }
-            acc.to_keyspace()
-        };
-
-        let ctx = match perf_instrument {
-            true => RequestContextBuilder::from(ctx)
-                .root_perf_span(|| {
-                    info_span!(
-                        target: PERF_TRACE_TARGET,
-                        "GET_VECTORED",
-                        tenant_id = %self.tenant_shard_id.tenant_id,
-                        timeline_id = %self.timeline_id,
-                        lsn = %effective_lsn,
-                        shard = %self.tenant_shard_id.shard_slug(),
-                    )
-                })
-                .attached_child(),
-            false => ctx.attached_child(),
-        };
+        let query: Vec<(Lsn, KeySpace)> = req_keyspaces
+            .into_iter()
+            .map(|(lsn, acc)| (lsn, acc.to_keyspace()))
+            .collect();
 
+        let query = VersionedKeySpaceQuery::scattered(query);
         let res = self
-            .get_vectored(keyspace, effective_lsn, io_concurrency, &ctx)
-            .maybe_perf_instrument(&ctx, |current_perf_span| current_perf_span.clone())
+            .get_vectored(query, io_concurrency, ctx)
+            .maybe_perf_instrument(ctx, |current_perf_span| {
+                info_span!(
+                    target: PERF_TRACE_TARGET,
+                    parent: current_perf_span,
+                    "GET_BATCH",
+                    batch_size = %page_count,
+                )
+            })
             .await;
 
         match res {
@@ -381,12 +358,12 @@ impl Timeline {
                         // There is no standardized way to express that the batched span followed from N request spans.
                         // So, abuse the system and mark the request contexts as follows_from the batch span, so we get
                         // some linkage in our trace viewer. It allows us to answer: which GET_VECTORED did this GET_PAGE wait for.
-                        req_ctx.perf_follows_from(&ctx);
+                        req_ctx.perf_follows_from(ctx);
                         slots_filled += 1;
                     }
 
                     result_slots[first_slot].write(res);
-                    first_req_ctx.perf_follows_from(&ctx);
+                    first_req_ctx.perf_follows_from(ctx);
                     slots_filled += 1;
                 }
             }
@@ -425,7 +402,7 @@ impl Timeline {
                         }
                     };
 
-                    req_ctx.perf_follows_from(&ctx);
+                    req_ctx.perf_follows_from(ctx);
                     result_slots[*slot].write(err);
                 }
 
@@ -664,8 +641,9 @@ impl Timeline {
 
         let mut segment = BytesMut::with_capacity(n_blocks as usize * BLCKSZ as usize);
         for batch in batches.parts {
+            let query = VersionedKeySpaceQuery::uniform(batch, lsn);
             let blocks = self
-                .get_vectored(batch, lsn, io_concurrency.clone(), ctx)
+                .get_vectored(query, io_concurrency.clone(), ctx)
                 .await?;
 
             for (_key, block) in blocks {
@@ -902,8 +880,9 @@ impl Timeline {
             );
 
             for batch in batches.parts.into_iter().rev() {
+                let query = VersionedKeySpaceQuery::uniform(batch, probe_lsn);
                 let blocks = self
-                    .get_vectored(batch, probe_lsn, io_concurrency.clone(), ctx)
+                    .get_vectored(query, io_concurrency.clone(), ctx)
                     .await?;
 
                 for (_key, clog_page) in blocks.into_iter().rev() {
@@ -1051,19 +1030,16 @@ impl Timeline {
             )
             .await?;
         let mut result = HashMap::new();
-        let mut sz = 0;
+
         for (_, v) in kv {
             let v = v?;
             let v = aux_file::decode_file_value_bytes(&v)
                 .context("value decode")
                 .map_err(PageReconstructError::Other)?;
             for (fname, content) in v {
-                sz += fname.len();
-                sz += content.len();
                 result.insert(fname, content);
             }
         }
-        self.aux_file_size_estimator.on_initial(sz);
         Ok(result)
     }
 
@@ -1334,12 +1310,12 @@ impl Timeline {
         let rel_size_cache = self.rel_size_cache.read().unwrap();
         if let Some((cached_lsn, nblocks)) = rel_size_cache.map.get(tag) {
             if lsn >= *cached_lsn {
-                RELSIZE_CACHE_HITS.inc();
+                
                 return Some(*nblocks);
             }
-            RELSIZE_CACHE_MISSES_OLD.inc();
+            
         }
-        RELSIZE_CACHE_MISSES.inc();
+       
         None
     }
 
@@ -1364,25 +1340,21 @@ impl Timeline {
             }
             hash_map::Entry::Vacant(entry) => {
                 entry.insert((lsn, nblocks));
-                RELSIZE_CACHE_ENTRIES.inc();
+               
             }
         }
     }
 
     /// Store cached relation size
-    pub fn set_cached_rel_size(&self, tag: RelTag, lsn: Lsn, nblocks: BlockNumber) {
-        let mut rel_size_cache = self.rel_size_cache.write().unwrap();
-        if rel_size_cache.map.insert(tag, (lsn, nblocks)).is_none() {
-            RELSIZE_CACHE_ENTRIES.inc();
-        }
+    pub fn set_cached_rel_size(&self, _tag: RelTag, _lsn: Lsn, _nblocks: BlockNumber) {
+       
+        
     }
 
     /// Remove cached relation size
-    pub fn remove_cached_rel_size(&self, tag: &RelTag) {
-        let mut rel_size_cache = self.rel_size_cache.write().unwrap();
-        if rel_size_cache.map.remove(tag).is_some() {
-            RELSIZE_CACHE_ENTRIES.dec();
-        }
+    pub fn remove_cached_rel_size(&self, _tag: &RelTag) {
+        
+        
     }
 }
 
@@ -1457,29 +1429,11 @@ impl DatadirModification<'_> {
             .is_some_and(|b| b.has_data())
     }
 
-    /// Returns statistics about the currently pending modifications.
-    pub(crate) fn stats(&self) -> DatadirModificationStats {
-        let mut stats = DatadirModificationStats::default();
-        for (_, _, value) in self.pending_metadata_pages.values().flatten() {
-            match value {
-                Value::Image(_) => stats.metadata_images += 1,
-                Value::WalRecord(r) if r.will_init() => stats.metadata_images += 1,
-                Value::WalRecord(_) => stats.metadata_deltas += 1,
-            }
-        }
-        for valuemeta in self.pending_data_batch.iter().flat_map(|b| &b.metadata) {
-            match valuemeta {
-                ValueMeta::Serialized(s) if s.will_init => stats.data_images += 1,
-                ValueMeta::Serialized(_) => stats.data_deltas += 1,
-                ValueMeta::Observed(_) => {}
-            }
-        }
-        stats
-    }
+    
 
     /// Set the current lsn
-    pub(crate) fn set_lsn(&mut self, lsn: Lsn) -> anyhow::Result<()> {
-        ensure!(
+    pub(crate) fn set_lsn(&mut self, lsn: Lsn) -> Result<(), WalIngestError> {
+        ensure_walingest!(
             lsn >= self.lsn,
             "setting an older lsn {} than {} is not allowed",
             lsn,
@@ -1578,7 +1532,7 @@ impl DatadirModification<'_> {
         &mut self,
         rel: RelTag,
         ctx: &RequestContext,
-    ) -> Result<u32, PageReconstructError> {
+    ) -> Result<u32, WalIngestError> {
         // Get current size and put rel creation if rel doesn't exist
         //
         // NOTE: we check the cache first even though get_rel_exists and get_rel_size would
@@ -1593,14 +1547,13 @@ impl DatadirModification<'_> {
             .await?
         {
             // create it with 0 size initially, the logic below will extend it
-            self.put_rel_creation(rel, 0, ctx)
-                .await
-                .context("Relation Error")?;
+            self.put_rel_creation(rel, 0, ctx).await?;
             Ok(0)
         } else {
-            self.tline
+            Ok(self
+                .tline
                 .get_rel_size(rel, Version::Modified(self), ctx)
-                .await
+                .await?)
         }
     }
 
@@ -1637,11 +1590,14 @@ impl DatadirModification<'_> {
         // TODO(vlad): remove this argument and replace the shard check with is_key_local
         shard: &ShardIdentity,
         ctx: &RequestContext,
-    ) -> anyhow::Result<()> {
+    ) -> Result<(), WalIngestError> {
         let mut gaps_at_lsns = Vec::default();
 
         for meta in batch.metadata.iter() {
-            let (rel, blkno) = Key::from_compact(meta.key()).to_rel_block()?;
+            let key = Key::from_compact(meta.key());
+            let (rel, blkno) = key
+                .to_rel_block()
+                .map_err(|_| WalIngestErrorKind::InvalidKey(key, meta.lsn()))?;
             let new_nblocks = blkno + 1;
 
             let old_nblocks = self.create_relation_if_required(rel, ctx).await?;
@@ -1683,8 +1639,8 @@ impl DatadirModification<'_> {
         rel: RelTag,
         blknum: BlockNumber,
         rec: NeonWalRecord,
-    ) -> anyhow::Result<()> {
-        anyhow::ensure!(rel.relnode != 0, RelationError::InvalidRelnode);
+    ) -> Result<(), WalIngestError> {
+        ensure_walingest!(rel.relnode != 0, RelationError::InvalidRelnode);
         self.put(rel_block_to_key(rel, blknum), Value::WalRecord(rec));
         Ok(())
     }
@@ -1696,7 +1652,7 @@ impl DatadirModification<'_> {
         segno: u32,
         blknum: BlockNumber,
         rec: NeonWalRecord,
-    ) -> anyhow::Result<()> {
+    ) -> Result<(), WalIngestError> {
         if !self.tline.tenant_shard_id.is_shard_zero() {
             return Ok(());
         }
@@ -1714,14 +1670,11 @@ impl DatadirModification<'_> {
         rel: RelTag,
         blknum: BlockNumber,
         img: Bytes,
-    ) -> anyhow::Result<()> {
-        anyhow::ensure!(rel.relnode != 0, RelationError::InvalidRelnode);
+    ) -> Result<(), WalIngestError> {
+        ensure_walingest!(rel.relnode != 0, RelationError::InvalidRelnode);
         let key = rel_block_to_key(rel, blknum);
         if !key.is_valid_key_on_write_path() {
-            anyhow::bail!(
-                "the request contains data not supported by pageserver at {}",
-                key
-            );
+            Err(WalIngestErrorKind::InvalidKey(key, self.lsn))?;
         }
         self.put(rel_block_to_key(rel, blknum), Value::Image(img));
         Ok(())
@@ -1733,15 +1686,12 @@ impl DatadirModification<'_> {
         segno: u32,
         blknum: BlockNumber,
         img: Bytes,
-    ) -> anyhow::Result<()> {
+    ) -> Result<(), WalIngestError> {
         assert!(self.tline.tenant_shard_id.is_shard_zero());
 
         let key = slru_block_to_key(kind, segno, blknum);
         if !key.is_valid_key_on_write_path() {
-            anyhow::bail!(
-                "the request contains data not supported by pageserver at {}",
-                key
-            );
+            Err(WalIngestErrorKind::InvalidKey(key, self.lsn))?;
         }
         self.put(key, Value::Image(img));
         Ok(())
@@ -1751,15 +1701,11 @@ impl DatadirModification<'_> {
         &mut self,
         rel: RelTag,
         blknum: BlockNumber,
-    ) -> anyhow::Result<()> {
-        anyhow::ensure!(rel.relnode != 0, RelationError::InvalidRelnode);
+    ) -> Result<(), WalIngestError> {
+        ensure_walingest!(rel.relnode != 0, RelationError::InvalidRelnode);
         let key = rel_block_to_key(rel, blknum);
         if !key.is_valid_key_on_write_path() {
-            anyhow::bail!(
-                "the request contains data not supported by pageserver: {} @ {}",
-                key,
-                self.lsn
-            );
+            Err(WalIngestErrorKind::InvalidKey(key, self.lsn))?;
         }
 
         let batch = self
@@ -1776,15 +1722,11 @@ impl DatadirModification<'_> {
         kind: SlruKind,
         segno: u32,
         blknum: BlockNumber,
-    ) -> anyhow::Result<()> {
+    ) -> Result<(), WalIngestError> {
         assert!(self.tline.tenant_shard_id.is_shard_zero());
         let key = slru_block_to_key(kind, segno, blknum);
         if !key.is_valid_key_on_write_path() {
-            anyhow::bail!(
-                "the request contains data not supported by pageserver: {} @ {}",
-                key,
-                self.lsn
-            );
+            Err(WalIngestErrorKind::InvalidKey(key, self.lsn))?;
         }
 
         let batch = self
@@ -1832,8 +1774,10 @@ impl DatadirModification<'_> {
         dbnode: Oid,
         img: Bytes,
         ctx: &RequestContext,
-    ) -> anyhow::Result<()> {
-        let v2_enabled = self.maybe_enable_rel_size_v2()?;
+    ) -> Result<(), WalIngestError> {
+        let v2_enabled = self
+            .maybe_enable_rel_size_v2()
+            .map_err(WalIngestErrorKind::MaybeRelSizeV2Error)?;
 
         // Add it to the directory (if it doesn't exist already)
         let buf = self.get(DBDIR_KEY, ctx).await?;
@@ -1874,13 +1818,13 @@ impl DatadirModification<'_> {
         xid: u64,
         img: Bytes,
         ctx: &RequestContext,
-    ) -> anyhow::Result<()> {
+    ) -> Result<(), WalIngestError> {
         // Add it to the directory entry
         let dirbuf = self.get(TWOPHASEDIR_KEY, ctx).await?;
         let newdirbuf = if self.tline.pg_version >= 17 {
             let mut dir = TwoPhaseDirectoryV17::des(&dirbuf)?;
             if !dir.xids.insert(xid) {
-                anyhow::bail!("twophase file for xid {} already exists", xid);
+                Err(WalIngestErrorKind::FileAlreadyExists(xid))?;
             }
             self.pending_directory_entries.push((
                 DirectoryKind::TwoPhase,
@@ -1891,7 +1835,7 @@ impl DatadirModification<'_> {
             let xid = xid as u32;
             let mut dir = TwoPhaseDirectory::des(&dirbuf)?;
             if !dir.xids.insert(xid) {
-                anyhow::bail!("twophase file for xid {} already exists", xid);
+                Err(WalIngestErrorKind::FileAlreadyExists(xid.into()))?;
             }
             self.pending_directory_entries.push((
                 DirectoryKind::TwoPhase,
@@ -1909,22 +1853,22 @@ impl DatadirModification<'_> {
         &mut self,
         origin_id: RepOriginId,
         origin_lsn: Lsn,
-    ) -> anyhow::Result<()> {
+    ) -> Result<(), WalIngestError> {
         let key = repl_origin_key(origin_id);
         self.put(key, Value::Image(origin_lsn.ser().unwrap().into()));
         Ok(())
     }
 
-    pub async fn drop_replorigin(&mut self, origin_id: RepOriginId) -> anyhow::Result<()> {
+    pub async fn drop_replorigin(&mut self, origin_id: RepOriginId) -> Result<(), WalIngestError> {
         self.set_replorigin(origin_id, Lsn::INVALID).await
     }
 
-    pub fn put_control_file(&mut self, img: Bytes) -> anyhow::Result<()> {
+    pub fn put_control_file(&mut self, img: Bytes) -> Result<(), WalIngestError> {
         self.put(CONTROLFILE_KEY, Value::Image(img));
         Ok(())
     }
 
-    pub fn put_checkpoint(&mut self, img: Bytes) -> anyhow::Result<()> {
+    pub fn put_checkpoint(&mut self, img: Bytes) -> Result<(), WalIngestError> {
         self.put(CHECKPOINT_KEY, Value::Image(img));
         Ok(())
     }
@@ -1934,7 +1878,7 @@ impl DatadirModification<'_> {
         spcnode: Oid,
         dbnode: Oid,
         ctx: &RequestContext,
-    ) -> anyhow::Result<()> {
+    ) -> Result<(), WalIngestError> {
         let total_blocks = self
             .tline
             .get_db_size(spcnode, dbnode, Version::Modified(self), ctx)
@@ -1973,20 +1917,21 @@ impl DatadirModification<'_> {
         rel: RelTag,
         nblocks: BlockNumber,
         ctx: &RequestContext,
-    ) -> Result<(), RelationError> {
+    ) -> Result<(), WalIngestError> {
         if rel.relnode == 0 {
-            return Err(RelationError::InvalidRelnode);
+            Err(WalIngestErrorKind::LogicalError(anyhow::anyhow!(
+                "invalid relnode"
+            )))?;
         }
         // It's possible that this is the first rel for this db in this
         // tablespace.  Create the reldir entry for it if so.
-        let mut dbdir = DbDirectory::des(&self.get(DBDIR_KEY, ctx).await.context("read db")?)
-            .context("deserialize db")?;
+        let mut dbdir = DbDirectory::des(&self.get(DBDIR_KEY, ctx).await?)?;
 
         let dbdir_exists =
             if let hash_map::Entry::Vacant(e) = dbdir.dbdirs.entry((rel.spcnode, rel.dbnode)) {
                 // Didn't exist. Update dbdir
                 e.insert(false);
-                let buf = DbDirectory::ser(&dbdir).context("serialize db")?;
+                let buf = DbDirectory::ser(&dbdir)?;
                 self.pending_directory_entries.push((
                     DirectoryKind::Db,
                     MetricsUpdate::Set(dbdir.dbdirs.len() as u64),
@@ -2003,27 +1948,25 @@ impl DatadirModification<'_> {
             RelDirectory::default()
         } else {
             // reldir already exists, fetch it
-            RelDirectory::des(&self.get(rel_dir_key, ctx).await.context("read db")?)
-                .context("deserialize db")?
+            RelDirectory::des(&self.get(rel_dir_key, ctx).await?)?
         };
 
-        let v2_enabled = self.maybe_enable_rel_size_v2()?;
+        let v2_enabled = self
+            .maybe_enable_rel_size_v2()
+            .map_err(WalIngestErrorKind::MaybeRelSizeV2Error)?;
 
         if v2_enabled {
             if rel_dir.rels.contains(&(rel.relnode, rel.forknum)) {
-                return Err(RelationError::AlreadyExists);
+                Err(WalIngestErrorKind::RelationAlreadyExists(rel))?;
             }
             let sparse_rel_dir_key =
                 rel_tag_sparse_key(rel.spcnode, rel.dbnode, rel.relnode, rel.forknum);
             // check if the rel_dir_key exists in v2
-            let val = self
-                .sparse_get(sparse_rel_dir_key, ctx)
-                .await
-                .map_err(|e| RelationError::Other(e.into()))?;
+            let val = self.sparse_get(sparse_rel_dir_key, ctx).await?;
             let val = RelDirExists::decode_option(val)
-                .map_err(|_| RelationError::Other(anyhow::anyhow!("invalid reldir key")))?;
+                .map_err(|_| WalIngestErrorKind::InvalidRelDirKey(sparse_rel_dir_key))?;
             if val == RelDirExists::Exists {
-                return Err(RelationError::AlreadyExists);
+                Err(WalIngestErrorKind::RelationAlreadyExists(rel))?;
             }
             self.put(
                 sparse_rel_dir_key,
@@ -2039,9 +1982,7 @@ impl DatadirModification<'_> {
                 // will be key not found errors if we don't create an empty one for rel_size_v2.
                 self.put(
                     rel_dir_key,
-                    Value::Image(Bytes::from(
-                        RelDirectory::ser(&RelDirectory::default()).context("serialize")?,
-                    )),
+                    Value::Image(Bytes::from(RelDirectory::ser(&RelDirectory::default())?)),
                 );
             }
             self.pending_directory_entries
@@ -2049,7 +1990,7 @@ impl DatadirModification<'_> {
         } else {
             // Add the new relation to the rel directory entry, and write it back
             if !rel_dir.rels.insert((rel.relnode, rel.forknum)) {
-                return Err(RelationError::AlreadyExists);
+                Err(WalIngestErrorKind::RelationAlreadyExists(rel))?;
             }
             if !dbdir_exists {
                 self.pending_directory_entries
@@ -2059,9 +2000,7 @@ impl DatadirModification<'_> {
                 .push((DirectoryKind::Rel, MetricsUpdate::Add(1)));
             self.put(
                 rel_dir_key,
-                Value::Image(Bytes::from(
-                    RelDirectory::ser(&rel_dir).context("serialize")?,
-                )),
+                Value::Image(Bytes::from(RelDirectory::ser(&rel_dir)?)),
             );
         }
 
@@ -2086,8 +2025,8 @@ impl DatadirModification<'_> {
         rel: RelTag,
         nblocks: BlockNumber,
         ctx: &RequestContext,
-    ) -> anyhow::Result<()> {
-        anyhow::ensure!(rel.relnode != 0, RelationError::InvalidRelnode);
+    ) -> Result<(), WalIngestError> {
+        ensure_walingest!(rel.relnode != 0, RelationError::InvalidRelnode);
         if self
             .tline
             .get_rel_exists(rel, Version::Modified(self), ctx)
@@ -2117,8 +2056,8 @@ impl DatadirModification<'_> {
         rel: RelTag,
         nblocks: BlockNumber,
         ctx: &RequestContext,
-    ) -> anyhow::Result<()> {
-        anyhow::ensure!(rel.relnode != 0, RelationError::InvalidRelnode);
+    ) -> Result<(), WalIngestError> {
+        ensure_walingest!(rel.relnode != 0, RelationError::InvalidRelnode);
 
         // Put size
         let size_key = rel_size_to_key(rel);
@@ -2142,8 +2081,10 @@ impl DatadirModification<'_> {
         &mut self,
         drop_relations: HashMap<(u32, u32), Vec<RelTag>>,
         ctx: &RequestContext,
-    ) -> anyhow::Result<()> {
-        let v2_enabled = self.maybe_enable_rel_size_v2()?;
+    ) -> Result<(), WalIngestError> {
+        let v2_enabled = self
+            .maybe_enable_rel_size_v2()
+            .map_err(WalIngestErrorKind::MaybeRelSizeV2Error)?;
         for ((spc_node, db_node), rel_tags) in drop_relations {
             let dir_key = rel_dir_to_key(spc_node, db_node);
             let buf = self.get(dir_key, ctx).await?;
@@ -2163,7 +2104,7 @@ impl DatadirModification<'_> {
                     let key =
                         rel_tag_sparse_key(spc_node, db_node, rel_tag.relnode, rel_tag.forknum);
                     let val = RelDirExists::decode_option(self.sparse_get(key, ctx).await?)
-                        .map_err(|_| RelationError::Other(anyhow::anyhow!("invalid reldir key")))?;
+                        .map_err(|_| WalIngestErrorKind::InvalidKey(key, self.lsn))?;
                     if val == RelDirExists::Exists {
                         self.pending_directory_entries
                             .push((DirectoryKind::RelV2, MetricsUpdate::Sub(1)));
@@ -2206,7 +2147,7 @@ impl DatadirModification<'_> {
         segno: u32,
         nblocks: BlockNumber,
         ctx: &RequestContext,
-    ) -> anyhow::Result<()> {
+    ) -> Result<(), WalIngestError> {
         assert!(self.tline.tenant_shard_id.is_shard_zero());
 
         // Add it to the directory entry
@@ -2215,7 +2156,7 @@ impl DatadirModification<'_> {
         let mut dir = SlruSegmentDirectory::des(&buf)?;
 
         if !dir.segments.insert(segno) {
-            anyhow::bail!("slru segment {kind:?}/{segno} already exists");
+            Err(WalIngestErrorKind::SlruAlreadyExists(kind, segno))?;
         }
         self.pending_directory_entries.push((
             DirectoryKind::SlruSegment(kind),
@@ -2242,7 +2183,7 @@ impl DatadirModification<'_> {
         kind: SlruKind,
         segno: u32,
         nblocks: BlockNumber,
-    ) -> anyhow::Result<()> {
+    ) -> Result<(), WalIngestError> {
         assert!(self.tline.tenant_shard_id.is_shard_zero());
 
         // Put size
@@ -2258,7 +2199,7 @@ impl DatadirModification<'_> {
         kind: SlruKind,
         segno: u32,
         ctx: &RequestContext,
-    ) -> anyhow::Result<()> {
+    ) -> Result<(), WalIngestError> {
         // Remove it from the directory entry
         let dir_key = slru_dir_to_key(kind);
         let buf = self.get(dir_key, ctx).await?;
@@ -2283,7 +2224,7 @@ impl DatadirModification<'_> {
     }
 
     /// Drop a relmapper file (pg_filenode.map)
-    pub fn drop_relmap_file(&mut self, _spcnode: Oid, _dbnode: Oid) -> anyhow::Result<()> {
+    pub fn drop_relmap_file(&mut self, _spcnode: Oid, _dbnode: Oid) -> Result<(), WalIngestError> {
         // TODO
         Ok(())
     }
@@ -2293,7 +2234,7 @@ impl DatadirModification<'_> {
         &mut self,
         xid: u64,
         ctx: &RequestContext,
-    ) -> anyhow::Result<()> {
+    ) -> Result<(), WalIngestError> {
         // Remove it from the directory entry
         let buf = self.get(TWOPHASEDIR_KEY, ctx).await?;
         let newdirbuf = if self.tline.pg_version >= 17 {
@@ -2308,7 +2249,8 @@ impl DatadirModification<'_> {
             ));
             Bytes::from(TwoPhaseDirectoryV17::ser(&dir)?)
         } else {
-            let xid: u32 = u32::try_from(xid)?;
+            let xid: u32 = u32::try_from(xid)
+                .map_err(|e| WalIngestErrorKind::LogicalError(anyhow::Error::from(e)))?;
             let mut dir = TwoPhaseDirectory::des(&buf)?;
 
             if !dir.xids.remove(&xid) {
@@ -2333,7 +2275,7 @@ impl DatadirModification<'_> {
         path: &str,
         content: &[u8],
         ctx: &RequestContext,
-    ) -> anyhow::Result<()> {
+    ) -> Result<(), WalIngestError> {
         let key = aux_file::encode_aux_file_key(path);
         // retrieve the key from the engine
         let old_val = match self.get(key, ctx).await {
@@ -2342,7 +2284,7 @@ impl DatadirModification<'_> {
             Err(e) => return Err(e.into()),
         };
         let files: Vec<(&str, &[u8])> = if let Some(ref old_val) = old_val {
-            aux_file::decode_file_value(old_val)?
+            aux_file::decode_file_value(old_val).map_err(WalIngestErrorKind::EncodeAuxFileError)?
         } else {
             Vec::new()
         };
@@ -2362,20 +2304,15 @@ impl DatadirModification<'_> {
         }
         let mut new_files = other_files;
         match (modifying_file, content.is_empty()) {
-            (Some(old_content), false) => {
-                self.tline
-                    .aux_file_size_estimator
-                    .on_update(old_content.len(), content.len());
+            (Some(_old_content), false) => {
+                
                 new_files.push((path, content));
             }
-            (Some(old_content), true) => {
-                self.tline
-                    .aux_file_size_estimator
-                    .on_remove(old_content.len());
+            (Some(_old_content), true) => {
+               
                 // not adding the file key to the final `new_files` vec.
             }
             (None, false) => {
-                self.tline.aux_file_size_estimator.on_add(content.len());
                 new_files.push((path, content));
             }
             // Compute may request delete of old version of pgstat AUX file if new one exceeds size limit.
@@ -2387,7 +2324,8 @@ impl DatadirModification<'_> {
             }
             (None, true) => warn!("removing non-existing aux file: {}", path),
         }
-        let new_val = aux_file::encode_file_value(&new_files)?;
+        let new_val = aux_file::encode_file_value(&new_files)
+            .map_err(WalIngestErrorKind::EncodeAuxFileError)?;
         self.put(key, Value::Image(new_val.into()));
 
         Ok(())

pageserver/src/tenant/blob_io.rs

@@ -22,6 +22,7 @@ use bytes::{BufMut, BytesMut};
 use pageserver_api::models::ImageCompressionAlgorithm;
 use tokio::io::AsyncWriteExt;
 use tokio_epoll_uring::{BoundedBuf, IoBuf, Slice};
+use tokio_util::sync::CancellationToken;
 use tracing::warn;
 
 use crate::context::RequestContext;
@@ -36,6 +37,63 @@ pub struct CompressionInfo {
     pub compressed_size: Option<usize>,
 }
 
+/// A blob header, with header+data length and compression info.
+///
+/// TODO: use this more widely, and add an encode() method too.
+/// TODO: document the header format.
+#[derive(Clone, Copy, Default)]
+pub struct Header {
+    pub header_len: usize,
+    pub data_len: usize,
+    pub compression_bits: u8,
+}
+
+impl Header {
+    /// Decodes a header from a byte slice.
+    pub fn decode(bytes: &[u8]) -> Result<Self, std::io::Error> {
+        let Some(&first_header_byte) = bytes.first() else {
+            return Err(std::io::Error::new(
+                std::io::ErrorKind::InvalidData,
+                "zero-length blob header",
+            ));
+        };
+
+        // If the first bit is 0, this is just a 1-byte length prefix up to 128 bytes.
+        if first_header_byte < 0x80 {
+            return Ok(Self {
+                header_len: 1, // by definition
+                data_len: first_header_byte as usize,
+                compression_bits: BYTE_UNCOMPRESSED,
+            });
+        }
+
+        // Otherwise, this is a 4-byte header containing compression information and length.
+        const HEADER_LEN: usize = 4;
+        let mut header_buf: [u8; HEADER_LEN] = bytes[0..HEADER_LEN].try_into().map_err(|_| {
+            std::io::Error::new(
+                std::io::ErrorKind::InvalidData,
+                format!("blob header too short: {bytes:?}"),
+            )
+        })?;
+
+        // TODO: verify the compression bits and convert to an enum.
+        let compression_bits = header_buf[0] & LEN_COMPRESSION_BIT_MASK;
+        header_buf[0] &= !LEN_COMPRESSION_BIT_MASK;
+        let data_len = u32::from_be_bytes(header_buf) as usize;
+
+        Ok(Self {
+            header_len: HEADER_LEN,
+            data_len,
+            compression_bits,
+        })
+    }
+
+    /// Returns the total header+data length.
+    pub fn total_len(&self) -> usize {
+        self.header_len + self.data_len
+    }
+}
+
 impl BlockCursor<'_> {
     /// Read a blob into a new buffer.
     pub async fn read_blob(
@@ -169,7 +227,13 @@ pub struct BlobWriter<const BUFFERED: bool> {
 }
 
 impl<const BUFFERED: bool> BlobWriter<BUFFERED> {
-    pub fn new(inner: VirtualFile, start_offset: u64) -> Self {
+    pub fn new(
+        inner: VirtualFile,
+        start_offset: u64,
+        _gate: &utils::sync::gate::Gate,
+        _cancel: CancellationToken,
+        _ctx: &RequestContext,
+    ) -> Self {
         Self {
             inner,
             offset: start_offset,
@@ -432,12 +496,14 @@ pub(crate) mod tests {
     ) -> Result<(Utf8TempDir, Utf8PathBuf, Vec<u64>), Error> {
         let temp_dir = camino_tempfile::tempdir()?;
         let pathbuf = temp_dir.path().join("file");
+        let gate = utils::sync::gate::Gate::default();
+        let cancel = CancellationToken::new();
 
         // Write part (in block to drop the file)
         let mut offsets = Vec::new();
         {
             let file = VirtualFile::create(pathbuf.as_path(), ctx).await?;
-            let mut wtr = BlobWriter::<BUFFERED>::new(file, 0);
+            let mut wtr = BlobWriter::<BUFFERED>::new(file, 0, &gate, cancel.clone(), ctx);
             for blob in blobs.iter() {
                 let (_, res) = if compression {
                     let res = wtr

pageserver/src/tenant/layer_map.rs

@@ -714,7 +714,7 @@ impl LayerMap {
         true
     }
 
-    pub fn iter_historic_layers(&self) -> impl '_ + Iterator<Item = Arc<PersistentLayerDesc>> {
+    pub fn iter_historic_layers(&self) -> impl ExactSizeIterator<Item = Arc<PersistentLayerDesc>> {
         self.historic.iter()
     }
 

pageserver/src/tenant/layer_map/historic_layer_coverage.rs

@@ -504,7 +504,7 @@ impl<Value: Clone> BufferedHistoricLayerCoverage<Value> {
     }
 
     /// Iterate all the layers
-    pub fn iter(&self) -> impl '_ + Iterator<Item = Value> {
+    pub fn iter(&self) -> impl ExactSizeIterator<Item = Value> {
         // NOTE we can actually perform this without rebuilding,
         //      but it's not necessary for now.
         if !self.buffer.is_empty() {

pageserver/src/tenant/mgr.rs

@@ -44,7 +44,7 @@ use crate::controller_upcall_client::{
 };
 use crate::deletion_queue::DeletionQueueClient;
 use crate::http::routes::ACTIVE_TENANT_TIMEOUT;
-use crate::metrics::{TENANT, TENANT_MANAGER as METRICS};
+use crate::metrics::TENANT_MANAGER as METRICS;
 use crate::task_mgr::{BACKGROUND_RUNTIME, TaskKind};
 use crate::tenant::config::{
     AttachedLocationConfig, AttachmentMode, LocationConf, LocationMode, SecondaryLocationConfig,
@@ -519,7 +519,7 @@ pub async fn init_tenant_mgr(
         tenant_configs.len(),
         conf.concurrent_tenant_warmup.initial_permits()
     );
-    TENANT.startup_scheduled.inc_by(tenant_configs.len() as u64);
+
 
     // Accumulate futures for writing tenant configs, so that we can execute in parallel
     let mut config_write_futs = Vec::new();
@@ -2177,9 +2177,7 @@ impl TenantManager {
                         // we would use if not doing any eviction.
                         progress.bytes_total
                     } else {
-                        // In the absence of heatmap info, assume that the secondary location simply
-                        // needs as much space as it is currently using.
-                        secondary.resident_size_metric.get()
+                        42
                     }
                 }
             }
@@ -2530,7 +2528,7 @@ impl SlotGuard {
                 Ok(())
             }
             None => {
-                METRICS.unexpected_errors.inc();
+              
                 error!(
                     tenant_shard_id = %self.tenant_shard_id,
                     "Missing InProgress marker during tenant upsert, this is a bug."
@@ -2540,7 +2538,7 @@ impl SlotGuard {
                 ))
             }
             Some(slot) => {
-                METRICS.unexpected_errors.inc();
+               
                 error!(tenant_shard_id=%self.tenant_shard_id, "Unexpected contents of TenantSlot during upsert, this is a bug.  Contents: {:?}", slot);
                 Err(TenantSlotUpsertError::InternalError(
                     "Unexpected contents of TenantSlot".into(),
@@ -2621,7 +2619,7 @@ impl Drop for SlotGuard {
         match m.entry(self.tenant_shard_id) {
             Entry::Occupied(mut entry) => {
                 if !matches!(entry.get(), TenantSlot::InProgress(_)) {
-                    METRICS.unexpected_errors.inc();
+                    
                     error!(tenant_shard_id=%self.tenant_shard_id, "Unexpected contents of TenantSlot during drop, this is a bug.  Contents: {:?}", entry.get());
                 }
 
@@ -2636,7 +2634,7 @@ impl Drop for SlotGuard {
                 }
             }
             Entry::Vacant(_) => {
-                METRICS.unexpected_errors.inc();
+                
                 error!(
                     tenant_shard_id = %self.tenant_shard_id,
                     "Missing InProgress marker during SlotGuard drop, this is a bug."
@@ -2696,7 +2694,7 @@ fn tenant_map_acquire_slot_impl(
     mode: TenantSlotAcquireMode,
 ) -> Result<SlotGuard, TenantSlotError> {
     use TenantSlotAcquireMode::*;
-    METRICS.tenant_slot_writes.inc();
+  
 
     let mut locked = tenants.write().unwrap();
     let span = tracing::info_span!("acquire_slot", tenant_id=%tenant_shard_id.tenant_id, shard_id = %tenant_shard_id.shard_slug());

pageserver/src/tenant/remote_timeline_client.rs

@@ -223,9 +223,8 @@ use crate::config::PageServerConf;
 use crate::context::RequestContext;
 use crate::deletion_queue::{DeletionQueueClient, DeletionQueueError};
 use crate::metrics::{
-    MeasureRemoteOp, REMOTE_ONDEMAND_DOWNLOADED_BYTES, REMOTE_ONDEMAND_DOWNLOADED_LAYERS,
-    RemoteOpFileKind, RemoteOpKind, RemoteTimelineClientMetrics,
-    RemoteTimelineClientMetricsCallTrackSize,
+    MeasureRemoteOp, 
+    RemoteOpFileKind, RemoteOpKind, 
 };
 use crate::task_mgr::{BACKGROUND_RUNTIME, TaskKind, shutdown_token};
 use crate::tenant::metadata::TimelineMetadata;
@@ -357,8 +356,6 @@ pub(crate) struct RemoteTimelineClient {
 
     upload_queue: Mutex<UploadQueue>,
 
-    pub(crate) metrics: Arc<RemoteTimelineClientMetrics>,
-
     storage_impl: GenericRemoteStorage,
 
     deletion_queue_client: DeletionQueueClient,
@@ -405,10 +402,6 @@ impl RemoteTimelineClient {
             storage_impl: remote_storage,
             deletion_queue_client,
             upload_queue: Mutex::new(UploadQueue::Uninitialized),
-            metrics: Arc::new(RemoteTimelineClientMetrics::new(
-                &tenant_shard_id,
-                &timeline_id,
-            )),
             config: std::sync::RwLock::new(RemoteTimelineClientConfig::from(location_conf)),
             cancel: CancellationToken::new(),
         }
@@ -597,21 +590,13 @@ impl RemoteTimelineClient {
             .map_err(|_| UploadQueueNotReadyError)
     }
 
-    fn update_remote_physical_size_gauge(&self, current_remote_index_part: Option<&IndexPart>) {
-        let size: u64 = if let Some(current_remote_index_part) = current_remote_index_part {
-            current_remote_index_part
-                .layer_metadata
-                .values()
-                .map(|ilmd| ilmd.file_size)
-                .sum()
-        } else {
-            0
-        };
-        self.metrics.remote_physical_size_gauge.set(size);
+    fn update_remote_physical_size_gauge(&self, _current_remote_index_part: Option<&IndexPart>) {
+
+        
     }
 
     pub fn get_remote_physical_size(&self) -> u64 {
-        self.metrics.remote_physical_size_gauge.get()
+  0
     }
 
     //
@@ -626,13 +611,6 @@ impl RemoteTimelineClient {
         &self,
         cancel: &CancellationToken,
     ) -> Result<MaybeDeletedIndexPart, DownloadError> {
-        let _unfinished_gauge_guard = self.metrics.call_begin(
-            &RemoteOpFileKind::Index,
-            &RemoteOpKind::Download,
-            crate::metrics::RemoteTimelineClientMetricsCallTrackSize::DontTrackSize {
-                reason: "no need for a downloads gauge",
-            },
-        );
 
         let (index_part, index_generation, index_last_modified) = download::download_index_part(
             &self.storage_impl,
@@ -645,7 +623,7 @@ impl RemoteTimelineClient {
             Option::<TaskKind>::None,
             RemoteOpFileKind::Index,
             RemoteOpKind::Download,
-            Arc::clone(&self.metrics),
+        
         )
         .await?;
 
@@ -720,13 +698,7 @@ impl RemoteTimelineClient {
         ctx: &RequestContext,
     ) -> Result<u64, DownloadError> {
         let downloaded_size = {
-            let _unfinished_gauge_guard = self.metrics.call_begin(
-                &RemoteOpFileKind::Layer,
-                &RemoteOpKind::Download,
-                crate::metrics::RemoteTimelineClientMetricsCallTrackSize::DontTrackSize {
-                    reason: "no need for a downloads gauge",
-                },
-            );
+        
             download::download_layer_file(
                 self.conf,
                 &self.storage_impl,
@@ -743,13 +715,11 @@ impl RemoteTimelineClient {
                 Some(ctx.task_kind()),
                 RemoteOpFileKind::Layer,
                 RemoteOpKind::Download,
-                Arc::clone(&self.metrics),
+          
             )
             .await?
         };
 
-        REMOTE_ONDEMAND_DOWNLOADED_LAYERS.inc();
-        REMOTE_ONDEMAND_DOWNLOADED_BYTES.inc_by(downloaded_size);
 
         Ok(downloaded_size)
     }
@@ -1027,7 +997,6 @@ impl RemoteTimelineClient {
         let op = UploadOp::UploadMetadata {
             uploaded: Box::new(index_part.clone()),
         };
-        self.metric_begin(&op);
         upload_queue.queued_operations.push_back(op);
         upload_queue.latest_files_changes_since_metadata_upload_scheduled = 0;
 
@@ -1265,7 +1234,6 @@ impl RemoteTimelineClient {
         );
 
         let op = UploadOp::UploadLayer(layer, metadata, None);
-        self.metric_begin(&op);
         upload_queue.queued_operations.push_back(op);
     }
 
@@ -1442,7 +1410,6 @@ impl RemoteTimelineClient {
         let op = UploadOp::Delete(Delete {
             layers: with_metadata,
         });
-        self.metric_begin(&op);
         upload_queue.queued_operations.push_back(op);
     }
 
@@ -2180,7 +2147,7 @@ impl RemoteTimelineClient {
                         Some(TaskKind::RemoteUploadTask),
                         RemoteOpFileKind::Layer,
                         RemoteOpKind::Upload,
-                        Arc::clone(&self.metrics),
+                      
                     )
                     .await
                 }
@@ -2197,7 +2164,7 @@ impl RemoteTimelineClient {
                         Some(TaskKind::RemoteUploadTask),
                         RemoteOpFileKind::Index,
                         RemoteOpKind::Upload,
-                        Arc::clone(&self.metrics),
+                    
                     )
                     .await;
                     if res.is_ok() {
@@ -2343,10 +2310,7 @@ impl RemoteTimelineClient {
                     upload_queue.clean.1 = Some(task.task_id);
 
                     let lsn = upload_queue.clean.0.metadata.disk_consistent_lsn();
-                    self.metrics
-                        .projected_remote_consistent_lsn_gauge
-                        .set(lsn.0);
-
+                   
                     if self.generation.is_none() {
                         // Legacy mode: skip validating generation
                         upload_queue.visible_remote_consistent_lsn.store(lsn);
@@ -2387,64 +2351,6 @@ impl RemoteTimelineClient {
                 .await;
         }
 
-        self.metric_end(&task.op);
-        for coalesced_op in &task.coalesced_ops {
-            self.metric_end(coalesced_op);
-        }
-    }
-
-    fn metric_impl(
-        &self,
-        op: &UploadOp,
-    ) -> Option<(
-        RemoteOpFileKind,
-        RemoteOpKind,
-        RemoteTimelineClientMetricsCallTrackSize,
-    )> {
-        use RemoteTimelineClientMetricsCallTrackSize::DontTrackSize;
-        let res = match op {
-            UploadOp::UploadLayer(_, m, _) => (
-                RemoteOpFileKind::Layer,
-                RemoteOpKind::Upload,
-                RemoteTimelineClientMetricsCallTrackSize::Bytes(m.file_size),
-            ),
-            UploadOp::UploadMetadata { .. } => (
-                RemoteOpFileKind::Index,
-                RemoteOpKind::Upload,
-                DontTrackSize {
-                    reason: "metadata uploads are tiny",
-                },
-            ),
-            UploadOp::Delete(_delete) => (
-                RemoteOpFileKind::Layer,
-                RemoteOpKind::Delete,
-                DontTrackSize {
-                    reason: "should we track deletes? positive or negative sign?",
-                },
-            ),
-            UploadOp::Barrier(..) | UploadOp::Shutdown => {
-                // we do not account these
-                return None;
-            }
-        };
-        Some(res)
-    }
-
-    fn metric_begin(&self, op: &UploadOp) {
-        let (file_kind, op_kind, track_bytes) = match self.metric_impl(op) {
-            Some(x) => x,
-            None => return,
-        };
-        let guard = self.metrics.call_begin(&file_kind, &op_kind, track_bytes);
-        guard.will_decrement_manually(); // in metric_end(), see right below
-    }
-
-    fn metric_end(&self, op: &UploadOp) {
-        let (file_kind, op_kind, track_bytes) = match self.metric_impl(op) {
-            Some(x) => x,
-            None => return,
-        };
-        self.metrics.call_end(&file_kind, &op_kind, track_bytes);
     }
 
     /// Close the upload queue for new operations and cancel queued operations.
@@ -2524,7 +2430,6 @@ impl RemoteTimelineClient {
 
                 // Tear down queued ops
                 for op in qi.queued_operations.into_iter() {
-                    self.metric_end(&op);
                     // Dropping UploadOp::Barrier() here will make wait_completion() return with an Err()
                     // which is exactly what we want to happen.
                     drop(op);
@@ -2834,10 +2739,6 @@ mod tests {
                 storage_impl: self.harness.remote_storage.clone(),
                 deletion_queue_client: self.harness.deletion_queue.new_client(),
                 upload_queue: Mutex::new(UploadQueue::Uninitialized),
-                metrics: Arc::new(RemoteTimelineClientMetrics::new(
-                    &self.harness.tenant_shard_id,
-                    &TIMELINE_ID,
-                )),
                 config: std::sync::RwLock::new(RemoteTimelineClientConfig::from(&location_conf)),
                 cancel: CancellationToken::new(),
             })
@@ -3064,99 +2965,7 @@ mod tests {
         );
     }
 
-    #[tokio::test]
-    async fn bytes_unfinished_gauge_for_layer_file_uploads() {
-        // Setup
-
-        let TestSetup {
-            harness,
-            tenant: _tenant,
-            timeline,
-            ..
-        } = TestSetup::new("metrics").await.unwrap();
-        let client = &timeline.remote_client;
-
-        let layer_file_name_1: LayerName = "000000000000000000000000000000000000-FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF__00000000016B59D8-00000000016B5A51".parse().unwrap();
-        let local_path = local_layer_path(
-            harness.conf,
-            &timeline.tenant_shard_id,
-            &timeline.timeline_id,
-            &layer_file_name_1,
-            &harness.generation,
-        );
-        let content_1 = dummy_contents("foo");
-        std::fs::write(&local_path, &content_1).unwrap();
-
-        let layer_file_1 = Layer::for_resident(
-            harness.conf,
-            &timeline,
-            local_path,
-            layer_file_name_1.clone(),
-            LayerFileMetadata::new(content_1.len() as u64, harness.generation, harness.shard),
-        );
-
-        #[derive(Debug, PartialEq, Clone, Copy)]
-        struct BytesStartedFinished {
-            started: Option<usize>,
-            finished: Option<usize>,
-        }
-        impl std::ops::Add for BytesStartedFinished {
-            type Output = Self;
-            fn add(self, rhs: Self) -> Self::Output {
-                Self {
-                    started: self.started.map(|v| v + rhs.started.unwrap_or(0)),
-                    finished: self.finished.map(|v| v + rhs.finished.unwrap_or(0)),
-                }
-            }
-        }
-        let get_bytes_started_stopped = || {
-            let started = client
-                .metrics
-                .get_bytes_started_counter_value(&RemoteOpFileKind::Layer, &RemoteOpKind::Upload)
-                .map(|v| v.try_into().unwrap());
-            let stopped = client
-                .metrics
-                .get_bytes_finished_counter_value(&RemoteOpFileKind::Layer, &RemoteOpKind::Upload)
-                .map(|v| v.try_into().unwrap());
-            BytesStartedFinished {
-                started,
-                finished: stopped,
-            }
-        };
-
-        // Test
-        tracing::info!("now doing actual test");
-
-        let actual_a = get_bytes_started_stopped();
-
-        client
-            .schedule_layer_file_upload(layer_file_1.clone())
-            .unwrap();
-
-        let actual_b = get_bytes_started_stopped();
-
-        client.wait_completion().await.unwrap();
-
-        let actual_c = get_bytes_started_stopped();
-
-        // Validate
-
-        let expected_b = actual_a
-            + BytesStartedFinished {
-                started: Some(content_1.len()),
-                // assert that the _finished metric is created eagerly so that subtractions work on first sample
-                finished: Some(0),
-            };
-        assert_eq!(actual_b, expected_b);
-
-        let expected_c = actual_a
-            + BytesStartedFinished {
-                started: Some(content_1.len()),
-                finished: Some(content_1.len()),
-            };
-        assert_eq!(actual_c, expected_c);
-    }
-
+    
     async fn inject_index_part(test_state: &TestSetup, generation: Generation) -> IndexPart {
         // An empty IndexPart, just sufficient to ensure deserialization will succeed
         let example_index_part = IndexPart::example();

pageserver/src/tenant/secondary.rs

@@ -6,7 +6,6 @@ mod scheduler;
 use std::sync::Arc;
 use std::time::SystemTime;
 
-use metrics::UIntGauge;
 use pageserver_api::models;
 use pageserver_api::shard::{ShardIdentity, TenantShardId};
 use remote_storage::GenericRemoteStorage;
@@ -26,7 +25,6 @@ use super::span::debug_assert_current_span_has_tenant_id;
 use super::storage_layer::LayerName;
 use crate::context::RequestContext;
 use crate::disk_usage_eviction_task::DiskUsageEvictionInfo;
-use crate::metrics::{SECONDARY_HEATMAP_TOTAL_SIZE, SECONDARY_RESIDENT_PHYSICAL_SIZE};
 use crate::task_mgr::{self, BACKGROUND_RUNTIME, TaskKind};
 
 enum DownloadCommand {
@@ -109,12 +107,7 @@ pub(crate) struct SecondaryTenant {
 
     // Public state indicating overall progress of downloads relative to the last heatmap seen
     pub(crate) progress: std::sync::Mutex<models::SecondaryProgress>,
-
-    // Sum of layer sizes on local disk
-    pub(super) resident_size_metric: UIntGauge,
-
-    // Sum of layer sizes in the most recently downloaded heatmap
-    pub(super) heatmap_total_size_metric: UIntGauge,
+   
 }
 
 impl SecondaryTenant {
@@ -124,16 +117,8 @@ impl SecondaryTenant {
         tenant_conf: pageserver_api::models::TenantConfig,
         config: &SecondaryLocationConfig,
     ) -> Arc<Self> {
-        let tenant_id = tenant_shard_id.tenant_id.to_string();
-        let shard_id = format!("{}", tenant_shard_id.shard_slug());
-        let resident_size_metric = SECONDARY_RESIDENT_PHYSICAL_SIZE
-            .get_metric_with_label_values(&[&tenant_id, &shard_id])
-            .unwrap();
-
-        let heatmap_total_size_metric = SECONDARY_HEATMAP_TOTAL_SIZE
-            .get_metric_with_label_values(&[&tenant_id, &shard_id])
-            .unwrap();
-
+    
+    
         Arc::new(Self {
             tenant_shard_id,
             // todo: shall we make this a descendent of the
@@ -150,14 +135,10 @@ impl SecondaryTenant {
 
             progress: std::sync::Mutex::default(),
 
-            resident_size_metric,
-            heatmap_total_size_metric,
         })
     }
 
-    pub(crate) fn tenant_shard_id(&self) -> TenantShardId {
-        self.tenant_shard_id
-    }
+    
 
     pub(crate) async fn shutdown(&self) {
         self.cancel.cancel();
@@ -169,15 +150,10 @@ impl SecondaryTenant {
 
         // Metrics are subtracted from and/or removed eagerly.
         // Deletions are done in the background via [`BackgroundPurges::spawn`].
-        let tenant_id = self.tenant_shard_id.tenant_id.to_string();
-        let shard_id = format!("{}", self.tenant_shard_id.shard_slug());
-        let _ = SECONDARY_RESIDENT_PHYSICAL_SIZE.remove_label_values(&[&tenant_id, &shard_id]);
-        let _ = SECONDARY_HEATMAP_TOTAL_SIZE.remove_label_values(&[&tenant_id, &shard_id]);
-
         self.detail
             .lock()
             .unwrap()
-            .drain_timelines(&self.tenant_shard_id, &self.resident_size_metric);
+            .drain_timelines(&self.tenant_shard_id);
     }
 
     pub(crate) fn set_config(&self, config: &SecondaryLocationConfig) {
@@ -255,7 +231,7 @@ impl SecondaryTenant {
             // of the cache.
             let mut detail = this.detail.lock().unwrap();
             if let Some(removed) =
-                detail.evict_layer(name, &timeline_id, now, &this.resident_size_metric)
+                detail.evict_layer(name, &timeline_id, now)
             {
                 // We might race with removal of the same layer during downloads, so finding the layer we
                 // were trying to remove is optional.  Only issue the disk I/O to remove it if we found it.
@@ -269,10 +245,9 @@ impl SecondaryTenant {
     /// Exhaustive check that incrementally updated metrics match the actual state.
     #[cfg(feature = "testing")]
     fn validate_metrics(&self) {
-        let detail = self.detail.lock().unwrap();
-        let resident_size = detail.total_resident_size();
+        
 
-        assert_eq!(resident_size, self.resident_size_metric.get());
+        
     }
 
     #[cfg(not(feature = "testing"))]

pageserver/src/tenant/secondary/downloader.rs

@@ -4,11 +4,9 @@ use std::str::FromStr;
 use std::sync::Arc;
 use std::time::{Duration, Instant, SystemTime};
 
-use crate::metrics::{STORAGE_IO_SIZE, StorageIoSizeOperation};
 use camino::Utf8PathBuf;
 use chrono::format::{DelayedFormat, StrftimeItems};
 use futures::Future;
-use metrics::UIntGauge;
 use pageserver_api::models::SecondaryProgress;
 use pageserver_api::shard::TenantShardId;
 use remote_storage::{DownloadError, DownloadKind, DownloadOpts, Etag, GenericRemoteStorage};
@@ -33,7 +31,6 @@ use crate::context::RequestContext;
 use crate::disk_usage_eviction_task::{
     DiskUsageEvictionInfo, EvictionCandidate, EvictionLayer, EvictionSecondaryLayer, finite_f32,
 };
-use crate::metrics::SECONDARY_MODE;
 use crate::tenant::config::SecondaryLocationConfig;
 use crate::tenant::debug_assert_current_span_has_tenant_and_timeline_id;
 use crate::tenant::ephemeral_file::is_ephemeral_file;
@@ -120,9 +117,6 @@ impl OnDiskState {
             .fatal_err("Deleting secondary layer")
     }
 
-    pub(crate) fn file_size(&self) -> u64 {
-        self.metadata.file_size
-    }
 }
 
 pub(super) struct SecondaryDetailTimeline {
@@ -175,13 +169,9 @@ impl SecondaryDetailTimeline {
     pub(super) fn remove_layer(
         &mut self,
         name: &LayerName,
-        resident_metric: &UIntGauge,
     ) -> Option<OnDiskState> {
-        let removed = self.on_disk_layers.remove(name);
-        if let Some(removed) = &removed {
-            resident_metric.sub(removed.file_size());
-        }
-        removed
+        self.on_disk_layers.remove(name)
+        
     }
 
     /// `local_path`
@@ -191,7 +181,6 @@ impl SecondaryDetailTimeline {
         tenant_shard_id: &TenantShardId,
         timeline_id: &TimelineId,
         touched: &HeatMapLayer,
-        resident_metric: &UIntGauge,
         local_path: F,
     ) where
         F: FnOnce() -> Utf8PathBuf,
@@ -211,7 +200,6 @@ impl SecondaryDetailTimeline {
                     touched.access_time,
                     local_path(),
                 ));
-                resident_metric.add(touched.metadata.file_size);
             }
         }
     }
@@ -267,28 +255,16 @@ impl SecondaryDetail {
         }
     }
 
-    #[cfg(feature = "testing")]
-    pub(crate) fn total_resident_size(&self) -> u64 {
-        self.timelines
-            .values()
-            .map(|tl| {
-                tl.on_disk_layers
-                    .values()
-                    .map(|v| v.metadata.file_size)
-                    .sum::<u64>()
-            })
-            .sum::<u64>()
-    }
 
     pub(super) fn evict_layer(
         &mut self,
         name: LayerName,
         timeline_id: &TimelineId,
         now: SystemTime,
-        resident_metric: &UIntGauge,
+
     ) -> Option<OnDiskState> {
         let timeline = self.timelines.get_mut(timeline_id)?;
-        let removed = timeline.remove_layer(&name, resident_metric);
+        let removed = timeline.remove_layer(&name);
         if removed.is_some() {
             timeline.evicted_at.insert(name, now);
         }
@@ -297,52 +273,21 @@ impl SecondaryDetail {
 
     pub(super) fn remove_timeline(
         &mut self,
-        tenant_shard_id: &TenantShardId,
+        _tenant_shard_id: &TenantShardId,
         timeline_id: &TimelineId,
-        resident_metric: &UIntGauge,
     ) {
-        let removed = self.timelines.remove(timeline_id);
-        if let Some(removed) = removed {
-            Self::clear_timeline_metrics(tenant_shard_id, timeline_id, removed, resident_metric);
-        }
+        self.timelines.remove(timeline_id);
+        
     }
 
     pub(super) fn drain_timelines(
         &mut self,
-        tenant_shard_id: &TenantShardId,
-        resident_metric: &UIntGauge,
+        _tenant_shard_id: &TenantShardId,
+
     ) {
-        for (timeline_id, removed) in self.timelines.drain() {
-            Self::clear_timeline_metrics(tenant_shard_id, &timeline_id, removed, resident_metric);
-        }
+        
     }
 
-    fn clear_timeline_metrics(
-        tenant_shard_id: &TenantShardId,
-        timeline_id: &TimelineId,
-        detail: SecondaryDetailTimeline,
-        resident_metric: &UIntGauge,
-    ) {
-        resident_metric.sub(
-            detail
-                .on_disk_layers
-                .values()
-                .map(|l| l.metadata.file_size)
-                .sum(),
-        );
-
-        let shard_id = format!("{}", tenant_shard_id.shard_slug());
-        let tenant_id = tenant_shard_id.tenant_id.to_string();
-        let timeline_id = timeline_id.to_string();
-        for op in StorageIoSizeOperation::VARIANTS {
-            let _ = STORAGE_IO_SIZE.remove_label_values(&[
-                op,
-                tenant_id.as_str(),
-                shard_id.as_str(),
-                timeline_id.as_str(),
-            ]);
-        }
-    }
 
     /// Additionally returns the total number of layers, used for more stable relative access time
     /// based eviction.
@@ -797,7 +742,6 @@ impl<'a> TenantDownloader<'a> {
                         tenant_shard_id,
                         last_heatmap,
                         timeline,
-                        &self.secondary_state.resident_size_metric,
                         ctx,
                     )
                     .await;
@@ -920,11 +864,7 @@ impl<'a> TenantDownloader<'a> {
             bytes_downloaded: 0,
         };
 
-        // Also expose heatmap bytes_total as a metric
-        self.secondary_state
-            .heatmap_total_size_metric
-            .set(heatmap_stats.bytes);
-
+       
         // Accumulate list of things to delete while holding the detail lock, for execution after dropping the lock
         let mut delete_layers = Vec::new();
         let mut delete_timelines = Vec::new();
@@ -991,7 +931,6 @@ impl<'a> TenantDownloader<'a> {
                 detail.remove_timeline(
                     self.secondary_state.get_tenant_shard_id(),
                     delete_timeline,
-                    &self.secondary_state.resident_size_metric,
                 );
             }
         }
@@ -1010,7 +949,7 @@ impl<'a> TenantDownloader<'a> {
             let Some(timeline_state) = detail.timelines.get_mut(&timeline_id) else {
                 continue;
             };
-            timeline_state.remove_layer(&layer_name, &self.secondary_state.resident_size_metric);
+            timeline_state.remove_layer(&layer_name);
         }
 
         for timeline_id in delete_timelines {
@@ -1077,7 +1016,7 @@ impl<'a> TenantDownloader<'a> {
         .await
         .ok_or_else(|| UpdateError::Cancelled)
         .and_then(|x| x)
-        .inspect(|_| SECONDARY_MODE.download_heatmap.inc())
+        .inspect(|_|{} )
     }
 
     /// Download heatmap layers that are not present on local disk, or update their
@@ -1252,7 +1191,6 @@ impl<'a> TenantDownloader<'a> {
                     tenant_shard_id,
                     &timeline_id,
                     &t,
-                    &self.secondary_state.resident_size_metric,
                     || {
                         local_layer_path(
                             self.conf,
@@ -1364,7 +1302,6 @@ impl<'a> TenantDownloader<'a> {
             progress.layers_downloaded += 1;
         }
 
-        SECONDARY_MODE.download_layer.inc();
 
         Ok(Some(layer))
     }
@@ -1376,7 +1313,6 @@ async fn init_timeline_state(
     tenant_shard_id: &TenantShardId,
     last_heatmap: Option<&HeatMapTimeline>,
     heatmap: &HeatMapTimeline,
-    resident_metric: &UIntGauge,
     ctx: &RequestContext,
 ) -> SecondaryDetailTimeline {
     let ctx = ctx.with_scope_secondary_timeline(tenant_shard_id, &heatmap.timeline_id);
@@ -1480,7 +1416,6 @@ async fn init_timeline_state(
                                 tenant_shard_id,
                                 &heatmap.timeline_id,
                                 remote_meta,
-                                resident_metric,
                                 || file_path,
                             );
                         }

pageserver/src/tenant/secondary/heatmap_uploader.rs

@@ -20,7 +20,6 @@ use super::scheduler::{
 };
 use super::{CommandRequest, SecondaryTenantError, UploadCommand};
 use crate::TEMP_FILE_SUFFIX;
-use crate::metrics::SECONDARY_MODE;
 use crate::tenant::Tenant;
 use crate::tenant::config::AttachmentMode;
 use crate::tenant::mgr::{GetTenantError, TenantManager};
@@ -221,14 +220,10 @@ impl JobGenerator<UploadPending, WriteInProgress, WriteComplete, UploadCommand>
             // Guard for the barrier in [`WriteInProgress`]
             let _completion = completion;
 
-            let started_at = Instant::now();
+            
             let uploaded = match upload_tenant_heatmap(remote_storage, &tenant, last_upload.clone()).await {
                 Ok(UploadHeatmapOutcome::Uploaded(uploaded)) => {
-                    let duration = Instant::now().duration_since(started_at);
-                    SECONDARY_MODE
-                        .upload_heatmap_duration
-                        .observe(duration.as_secs_f64());
-                    SECONDARY_MODE.upload_heatmap.inc();
+                    
                     Some(uploaded)
                 }
                 Ok(UploadHeatmapOutcome::NoChange | UploadHeatmapOutcome::Skipped) => last_upload,
@@ -237,11 +232,8 @@ impl JobGenerator<UploadPending, WriteInProgress, WriteComplete, UploadCommand>
                         "Failed to upload heatmap for tenant {}: {e:#}",
                         tenant.get_tenant_shard_id(),
                     );
-                    let duration = Instant::now().duration_since(started_at);
-                    SECONDARY_MODE
-                        .upload_heatmap_duration
-                        .observe(duration.as_secs_f64());
-                    SECONDARY_MODE.upload_heatmap_errors.inc();
+                   
+
                     last_upload
                 }
                 Err(UploadHeatmapError::Cancelled) => {

pageserver/src/tenant/storage_layer.rs

@@ -715,13 +715,34 @@ pub(crate) enum LayerId {
 }
 
 /// Uniquely identify a layer visit by the layer
-/// and LSN floor (or start LSN) of the reads.
-/// The layer itself is not enough since we may
-/// have different LSN lower bounds for delta layer reads.
+/// and LSN range of the reads. Note that the end of the range is exclusive.
+///
+/// The layer itself is not enough since we may have different LSN lower
+/// bounds for delta layer reads. Scenarios where this can happen are:
+///
+/// 1. Layer overlaps: imagine an image layer inside and in-memory layer
+///    and a query that only partially hits the image layer. Part of the query
+///    needs to read the whole in-memory layer and the other part needs to read
+///    only up to the image layer. Hence, they'll have different LSN floor values
+///    for the read.
+///
+/// 2. Scattered reads: the read path supports starting at different LSNs. Imagine
+///    The start LSN for one range is inside a layer and the start LSN for another range
+///    Is above the layer (includes all of it). Both ranges need to read the layer all the
+///    Way to the end but starting at different points. Hence, they'll have different LSN
+///    Ceil values.
+///
+/// The implication is that we might visit the same layer multiple times
+/// in order to read different LSN ranges from it. In practice, this isn't very concerning
+/// because:
+/// 1. Layer overlaps are rare and generally not intended
+/// 2. Scattered reads will stabilise after the first few layers provided their starting LSNs
+///    are grouped tightly enough (likely the case).
 #[derive(Debug, PartialEq, Eq, Clone, Hash)]
 struct LayerToVisitId {
     layer_id: LayerId,
     lsn_floor: Lsn,
+    lsn_ceil: Lsn,
 }
 
 #[derive(Debug, PartialEq, Eq, Hash)]
@@ -805,6 +826,7 @@ impl LayerFringe {
         let layer_to_visit_id = LayerToVisitId {
             layer_id: layer.id(),
             lsn_floor: lsn_range.start,
+            lsn_ceil: lsn_range.end,
         };
 
         let entry = self.visit_reads.entry(layer_to_visit_id.clone());

pageserver/src/tenant/storage_layer/batch_split_writer.rs

@@ -5,6 +5,7 @@ use std::sync::Arc;
 use bytes::Bytes;
 use pageserver_api::key::{KEY_SIZE, Key};
 use pageserver_api::value::Value;
+use tokio_util::sync::CancellationToken;
 use utils::id::TimelineId;
 use utils::lsn::Lsn;
 use utils::shard::TenantShardId;
@@ -179,7 +180,7 @@ impl BatchLayerWriter {
 
 /// An image writer that takes images and produces multiple image layers.
 #[must_use]
-pub struct SplitImageLayerWriter {
+pub struct SplitImageLayerWriter<'a> {
     inner: ImageLayerWriter,
     target_layer_size: u64,
     lsn: Lsn,
@@ -188,9 +189,12 @@ pub struct SplitImageLayerWriter {
     tenant_shard_id: TenantShardId,
     batches: BatchLayerWriter,
     start_key: Key,
+    gate: &'a utils::sync::gate::Gate,
+    cancel: CancellationToken,
 }
 
-impl SplitImageLayerWriter {
+impl<'a> SplitImageLayerWriter<'a> {
+    #[allow(clippy::too_many_arguments)]
     pub async fn new(
         conf: &'static PageServerConf,
         timeline_id: TimelineId,
@@ -198,6 +202,8 @@ impl SplitImageLayerWriter {
         start_key: Key,
         lsn: Lsn,
         target_layer_size: u64,
+        gate: &'a utils::sync::gate::Gate,
+        cancel: CancellationToken,
         ctx: &RequestContext,
     ) -> anyhow::Result<Self> {
         Ok(Self {
@@ -208,6 +214,8 @@ impl SplitImageLayerWriter {
                 tenant_shard_id,
                 &(start_key..Key::MAX),
                 lsn,
+                gate,
+                cancel.clone(),
                 ctx,
             )
             .await?,
@@ -217,6 +225,8 @@ impl SplitImageLayerWriter {
             batches: BatchLayerWriter::new(conf).await?,
             lsn,
             start_key,
+            gate,
+            cancel,
         })
     }
 
@@ -239,6 +249,8 @@ impl SplitImageLayerWriter {
                 self.tenant_shard_id,
                 &(key..Key::MAX),
                 self.lsn,
+                self.gate,
+                self.cancel.clone(),
                 ctx,
             )
             .await?;
@@ -291,7 +303,7 @@ impl SplitImageLayerWriter {
 /// into a single file. This behavior might change in the future. For reference, the legacy compaction algorithm
 /// will split them into multiple files based on size.
 #[must_use]
-pub struct SplitDeltaLayerWriter {
+pub struct SplitDeltaLayerWriter<'a> {
     inner: Option<(Key, DeltaLayerWriter)>,
     target_layer_size: u64,
     conf: &'static PageServerConf,
@@ -300,15 +312,19 @@ pub struct SplitDeltaLayerWriter {
     lsn_range: Range<Lsn>,
     last_key_written: Key,
     batches: BatchLayerWriter,
+    gate: &'a utils::sync::gate::Gate,
+    cancel: CancellationToken,
 }
 
-impl SplitDeltaLayerWriter {
+impl<'a> SplitDeltaLayerWriter<'a> {
     pub async fn new(
         conf: &'static PageServerConf,
         timeline_id: TimelineId,
         tenant_shard_id: TenantShardId,
         lsn_range: Range<Lsn>,
         target_layer_size: u64,
+        gate: &'a utils::sync::gate::Gate,
+        cancel: CancellationToken,
     ) -> anyhow::Result<Self> {
         Ok(Self {
             target_layer_size,
@@ -319,6 +335,8 @@ impl SplitDeltaLayerWriter {
             lsn_range,
             last_key_written: Key::MIN,
             batches: BatchLayerWriter::new(conf).await?,
+            gate,
+            cancel,
         })
     }
 
@@ -344,6 +362,8 @@ impl SplitDeltaLayerWriter {
                     self.tenant_shard_id,
                     key,
                     self.lsn_range.clone(),
+                    self.gate,
+                    self.cancel.clone(),
                     ctx,
                 )
                 .await?,
@@ -362,6 +382,8 @@ impl SplitDeltaLayerWriter {
                     self.tenant_shard_id,
                     key,
                     self.lsn_range.clone(),
+                    self.gate,
+                    self.cancel.clone(),
                     ctx,
                 )
                 .await?;
@@ -469,6 +491,8 @@ mod tests {
             get_key(0),
             Lsn(0x18),
             4 * 1024 * 1024,
+            &tline.gate,
+            tline.cancel.clone(),
             &ctx,
         )
         .await
@@ -480,6 +504,8 @@ mod tests {
             tenant.tenant_shard_id,
             Lsn(0x18)..Lsn(0x20),
             4 * 1024 * 1024,
+            &tline.gate,
+            tline.cancel.clone(),
         )
         .await
         .unwrap();
@@ -546,6 +572,8 @@ mod tests {
             get_key(0),
             Lsn(0x18),
             4 * 1024 * 1024,
+            &tline.gate,
+            tline.cancel.clone(),
             &ctx,
         )
         .await
@@ -556,6 +584,8 @@ mod tests {
             tenant.tenant_shard_id,
             Lsn(0x18)..Lsn(0x20),
             4 * 1024 * 1024,
+            &tline.gate,
+            tline.cancel.clone(),
         )
         .await
         .unwrap();
@@ -643,6 +673,8 @@ mod tests {
             get_key(0),
             Lsn(0x18),
             4 * 1024,
+            &tline.gate,
+            tline.cancel.clone(),
             &ctx,
         )
         .await
@@ -654,6 +686,8 @@ mod tests {
             tenant.tenant_shard_id,
             Lsn(0x18)..Lsn(0x20),
             4 * 1024,
+            &tline.gate,
+            tline.cancel.clone(),
         )
         .await
         .unwrap();
@@ -730,6 +764,8 @@ mod tests {
             tenant.tenant_shard_id,
             Lsn(0x10)..Lsn(N as u64 * 16 + 0x10),
             4 * 1024 * 1024,
+            &tline.gate,
+            tline.cancel.clone(),
         )
         .await
         .unwrap();

pageserver/src/tenant/storage_layer/delta_layer.rs

@@ -50,6 +50,7 @@ use rand::distributions::Alphanumeric;
 use serde::{Deserialize, Serialize};
 use tokio::sync::OnceCell;
 use tokio_epoll_uring::IoBuf;
+use tokio_util::sync::CancellationToken;
 use tracing::*;
 use utils::bin_ser::BeSer;
 use utils::id::{TenantId, TimelineId};
@@ -400,12 +401,15 @@ impl DeltaLayerWriterInner {
     ///
     /// Start building a new delta layer.
     ///
+    #[allow(clippy::too_many_arguments)]
     async fn new(
         conf: &'static PageServerConf,
         timeline_id: TimelineId,
         tenant_shard_id: TenantShardId,
         key_start: Key,
         lsn_range: Range<Lsn>,
+        gate: &utils::sync::gate::Gate,
+        cancel: CancellationToken,
         ctx: &RequestContext,
     ) -> anyhow::Result<Self> {
         // Create the file initially with a temporary filename. We don't know
@@ -420,7 +424,7 @@ impl DeltaLayerWriterInner {
         let mut file = VirtualFile::create(&path, ctx).await?;
         // make room for the header block
         file.seek(SeekFrom::Start(PAGE_SZ as u64)).await?;
-        let blob_writer = BlobWriter::new(file, PAGE_SZ as u64);
+        let blob_writer = BlobWriter::new(file, PAGE_SZ as u64, gate, cancel, ctx);
 
         // Initialize the b-tree index builder
         let block_buf = BlockBuf::new();
@@ -628,12 +632,15 @@ impl DeltaLayerWriter {
     ///
     /// Start building a new delta layer.
     ///
+    #[allow(clippy::too_many_arguments)]
     pub async fn new(
         conf: &'static PageServerConf,
         timeline_id: TimelineId,
         tenant_shard_id: TenantShardId,
         key_start: Key,
         lsn_range: Range<Lsn>,
+        gate: &utils::sync::gate::Gate,
+        cancel: CancellationToken,
         ctx: &RequestContext,
     ) -> anyhow::Result<Self> {
         Ok(Self {
@@ -644,6 +651,8 @@ impl DeltaLayerWriter {
                     tenant_shard_id,
                     key_start,
                     lsn_range,
+                    gate,
+                    cancel,
                     ctx,
                 )
                 .await?,
@@ -1885,6 +1894,8 @@ pub(crate) mod test {
             harness.tenant_shard_id,
             entries_meta.key_range.start,
             entries_meta.lsn_range.clone(),
+            &timeline.gate,
+            timeline.cancel.clone(),
             &ctx,
         )
         .await?;
@@ -2079,6 +2090,8 @@ pub(crate) mod test {
                 tenant.tenant_shard_id,
                 Key::MIN,
                 Lsn(0x11)..truncate_at,
+                &branch.gate,
+                branch.cancel.clone(),
                 ctx,
             )
             .await
@@ -2213,6 +2226,8 @@ pub(crate) mod test {
             tenant.tenant_shard_id,
             *key_start,
             (*lsn_min)..lsn_end,
+            &tline.gate,
+            tline.cancel.clone(),
             ctx,
         )
         .await?;

pageserver/src/tenant/storage_layer/image_layer.rs

@@ -48,6 +48,7 @@ use rand::distributions::Alphanumeric;
 use serde::{Deserialize, Serialize};
 use tokio::sync::OnceCell;
 use tokio_stream::StreamExt;
+use tokio_util::sync::CancellationToken;
 use tracing::*;
 use utils::bin_ser::BeSer;
 use utils::id::{TenantId, TimelineId};
@@ -748,12 +749,15 @@ impl ImageLayerWriterInner {
     ///
     /// Start building a new image layer.
     ///
+    #[allow(clippy::too_many_arguments)]
     async fn new(
         conf: &'static PageServerConf,
         timeline_id: TimelineId,
         tenant_shard_id: TenantShardId,
         key_range: &Range<Key>,
         lsn: Lsn,
+        gate: &utils::sync::gate::Gate,
+        cancel: CancellationToken,
         ctx: &RequestContext,
     ) -> anyhow::Result<Self> {
         // Create the file initially with a temporary filename.
@@ -780,7 +784,7 @@ impl ImageLayerWriterInner {
         };
         // make room for the header block
         file.seek(SeekFrom::Start(PAGE_SZ as u64)).await?;
-        let blob_writer = BlobWriter::new(file, PAGE_SZ as u64);
+        let blob_writer = BlobWriter::new(file, PAGE_SZ as u64, gate, cancel, ctx);
 
         // Initialize the b-tree index builder
         let block_buf = BlockBuf::new();
@@ -878,14 +882,6 @@ impl ImageLayerWriterInner {
     ) -> anyhow::Result<(PersistentLayerDesc, Utf8PathBuf)> {
         let index_start_blk = self.blob_writer.size().div_ceil(PAGE_SZ as u64) as u32;
 
-        // Calculate compression ratio
-        let compressed_size = self.blob_writer.size() - PAGE_SZ as u64; // Subtract PAGE_SZ for header
-        crate::metrics::COMPRESSION_IMAGE_INPUT_BYTES.inc_by(self.uncompressed_bytes);
-        crate::metrics::COMPRESSION_IMAGE_INPUT_BYTES_CONSIDERED
-            .inc_by(self.uncompressed_bytes_eligible);
-        crate::metrics::COMPRESSION_IMAGE_INPUT_BYTES_CHOSEN.inc_by(self.uncompressed_bytes_chosen);
-        crate::metrics::COMPRESSION_IMAGE_OUTPUT_BYTES.inc_by(compressed_size);
-
         let mut file = self.blob_writer.into_inner();
 
         // Write out the index
@@ -988,18 +984,30 @@ impl ImageLayerWriter {
     ///
     /// Start building a new image layer.
     ///
+    #[allow(clippy::too_many_arguments)]
     pub async fn new(
         conf: &'static PageServerConf,
         timeline_id: TimelineId,
         tenant_shard_id: TenantShardId,
         key_range: &Range<Key>,
         lsn: Lsn,
+        gate: &utils::sync::gate::Gate,
+        cancel: CancellationToken,
         ctx: &RequestContext,
     ) -> anyhow::Result<ImageLayerWriter> {
         Ok(Self {
             inner: Some(
-                ImageLayerWriterInner::new(conf, timeline_id, tenant_shard_id, key_range, lsn, ctx)
-                    .await?,
+                ImageLayerWriterInner::new(
+                    conf,
+                    timeline_id,
+                    tenant_shard_id,
+                    key_range,
+                    lsn,
+                    gate,
+                    cancel,
+                    ctx,
+                )
+                .await?,
             ),
         })
     }
@@ -1192,7 +1200,7 @@ mod test {
 
         // This key range contains several 0x8000 page stripes, only one of which belongs to shard zero
         let input_start = Key::from_hex("000000067f00000001000000ae0000000000").unwrap();
-        let input_end = Key::from_hex("000000067f00000001000000ae0000020000").unwrap();
+        let input_end = Key::from_hex("000000067f00000001000000ae0000002000").unwrap();
         let range = input_start..input_end;
 
         // Build an image layer to filter
@@ -1203,6 +1211,8 @@ mod test {
                 harness.tenant_shard_id,
                 &range,
                 lsn,
+                &timeline.gate,
+                timeline.cancel.clone(),
                 &ctx,
             )
             .await
@@ -1235,7 +1245,7 @@ mod test {
             let shard_identity = ShardIdentity::new(
                 ShardNumber(shard_number),
                 shard_count,
-                ShardStripeSize(0x8000),
+                ShardStripeSize(0x800),
             )
             .unwrap();
             let harness = TenantHarness::create_custom(
@@ -1268,6 +1278,8 @@ mod test {
                 harness.tenant_shard_id,
                 &range,
                 lsn,
+                &timeline.gate,
+                timeline.cancel.clone(),
                 &ctx,
             )
             .await
@@ -1287,12 +1299,12 @@ mod test {
 
             // This exact size and those below will need updating as/when the layer encoding changes, but
             // should be deterministic for a given version of the format, as we used no randomness generating the input.
-            assert_eq!(original_size, 1597440);
+            assert_eq!(original_size, 122880);
 
             match shard_number {
                 0 => {
                     // We should have written out just one stripe for our shard identity
-                    assert_eq!(wrote_keys, 0x8000);
+                    assert_eq!(wrote_keys, 0x800);
                     let replacement = replacement.unwrap();
 
                     // We should have dropped some of the data
@@ -1300,7 +1312,7 @@ mod test {
                     assert!(replacement.metadata().file_size > 0);
 
                     // Assert that we dropped ~3/4 of the data.
-                    assert_eq!(replacement.metadata().file_size, 417792);
+                    assert_eq!(replacement.metadata().file_size, 49152);
                 }
                 1 => {
                     // Shard 1 has no keys in our input range
@@ -1309,19 +1321,19 @@ mod test {
                 }
                 2 => {
                     // Shard 2 has one stripes in the input range
-                    assert_eq!(wrote_keys, 0x8000);
+                    assert_eq!(wrote_keys, 0x800);
                     let replacement = replacement.unwrap();
                     assert!(replacement.metadata().file_size < original_size);
                     assert!(replacement.metadata().file_size > 0);
-                    assert_eq!(replacement.metadata().file_size, 417792);
+                    assert_eq!(replacement.metadata().file_size, 49152);
                 }
                 3 => {
                     // Shard 3 has two stripes in the input range
-                    assert_eq!(wrote_keys, 0x10000);
+                    assert_eq!(wrote_keys, 0x1000);
                     let replacement = replacement.unwrap();
                     assert!(replacement.metadata().file_size < original_size);
                     assert!(replacement.metadata().file_size > 0);
-                    assert_eq!(replacement.metadata().file_size, 811008);
+                    assert_eq!(replacement.metadata().file_size, 73728);
                 }
                 _ => unreachable!(),
             }
@@ -1346,6 +1358,8 @@ mod test {
             tenant.tenant_shard_id,
             &key_range,
             lsn,
+            &tline.gate,
+            tline.cancel.clone(),
             ctx,
         )
         .await?;

pageserver/src/tenant/storage_layer/inmemory_layer.rs

@@ -32,7 +32,6 @@ use crate::config::PageServerConf;
 use crate::context::{PageContentKind, RequestContext, RequestContextBuilder};
 // avoid binding to Write (conflicts with std::io::Write)
 // while being able to use std::fmt::Write's methods
-use crate::metrics::TIMELINE_EPHEMERAL_BYTES;
 use crate::tenant::ephemeral_file::EphemeralFile;
 use crate::tenant::storage_layer::{OnDiskValue, OnDiskValueIo};
 use crate::tenant::timeline::GetVectoredError;
@@ -307,11 +306,7 @@ impl GlobalResourceUnits {
             }
         };
 
-        // This is a sloppy update: concurrent updates to the counter will race, and the exact
-        // value of the metric might not be the exact latest value of GLOBAL_RESOURCES::dirty_bytes.
-        // That's okay: as long as the metric contains some recent value, it doesn't have to always
-        // be literally the last update.
-        TIMELINE_EPHEMERAL_BYTES.set(new_global_dirty_bytes);
+       
 
         self.dirty_bytes = size;
 
@@ -719,6 +714,8 @@ impl InMemoryLayer {
         ctx: &RequestContext,
         key_range: Option<Range<Key>>,
         l0_flush_global_state: &l0_flush::Inner,
+        gate: &utils::sync::gate::Gate,
+        cancel: CancellationToken,
     ) -> Result<Option<(PersistentLayerDesc, Utf8PathBuf)>> {
         // Grab the lock in read-mode. We hold it over the I/O, but because this
         // layer is not writeable anymore, no one should be trying to acquire the
@@ -759,6 +756,8 @@ impl InMemoryLayer {
             self.tenant_shard_id,
             Key::MIN,
             self.start_lsn..end_lsn,
+            gate,
+            cancel,
             ctx,
         )
         .await?;

pageserver/src/tenant/storage_layer/layer.rs

@@ -231,9 +231,7 @@ impl Layer {
 
         debug_assert!(owner.0.needs_download_blocking().unwrap().is_none());
 
-        timeline
-            .metrics
-            .resident_physical_size_add(metadata.file_size);
+        
 
         ResidentLayer { downloaded, owner }
     }
@@ -526,12 +524,6 @@ impl Layer {
                 }
             }
 
-            // Update the timeline's visible bytes count
-            if let Some(tl) = self.0.timeline.upgrade() {
-                tl.metrics
-                    .visible_physical_size_gauge
-                    .add(self.0.desc.file_size)
-            }
         }
     }
 
@@ -540,23 +532,10 @@ impl Layer {
         use LayerVisibilityHint::*;
         match (old_visibility, visibility) {
             (Visible, Covered) => {
-                // Subtract this layer's contribution to the visible size metric
-                if let Some(tl) = self.0.timeline.upgrade() {
-                    debug_assert!(
-                        tl.metrics.visible_physical_size_gauge.get() >= self.0.desc.file_size
-                    );
-                    tl.metrics
-                        .visible_physical_size_gauge
-                        .sub(self.0.desc.file_size)
-                }
+                
             }
             (Covered, Visible) => {
-                // Add this layer's contribution to the visible size metric
-                if let Some(tl) = self.0.timeline.upgrade() {
-                    tl.metrics
-                        .visible_physical_size_gauge
-                        .add(self.0.desc.file_size)
-                }
+                
             }
             (Covered, Covered) | (Visible, Visible) => {
                 // no change
@@ -609,7 +588,6 @@ impl ResidentOrWantedEvicted {
             ResidentOrWantedEvicted::Resident(strong) => Some((strong.clone(), false)),
             ResidentOrWantedEvicted::WantedEvicted(weak, _) => match weak.upgrade() {
                 Some(strong) => {
-                    LAYER_IMPL_METRICS.inc_raced_wanted_evicted_accesses();
 
                     *self = ResidentOrWantedEvicted::Resident(strong.clone());
 
@@ -741,17 +719,8 @@ enum Status {
 
 impl Drop for LayerInner {
     fn drop(&mut self) {
-        // if there was a pending eviction, mark it cancelled here to balance metrics
-        if let Some((ResidentOrWantedEvicted::WantedEvicted(..), _)) = self.inner.take_and_deinit()
-        {
-            // eviction has already been started
-            LAYER_IMPL_METRICS.inc_eviction_cancelled(EvictionCancelled::LayerGone);
-
-            // eviction request is intentionally not honored as no one is present to wait for it
-            // and we could be delaying shutdown for nothing.
-        }
-
-        let timeline = self.timeline.upgrade();
+        
+        let timeline: Option<Arc<Timeline>> = self.timeline.upgrade();
 
         if let Some(timeline) = timeline.as_ref() {
             // Only need to decrement metrics if the timeline still exists: otherwise
@@ -759,13 +728,6 @@ impl Drop for LayerInner {
             timeline.metrics.dec_layer(&self.desc);
 
             if matches!(self.access_stats.visibility(), LayerVisibilityHint::Visible) {
-                debug_assert!(
-                    timeline.metrics.visible_physical_size_gauge.get() >= self.desc.file_size
-                );
-                timeline
-                    .metrics
-                    .visible_physical_size_gauge
-                    .sub(self.desc.file_size);
             }
         }
 
@@ -777,7 +739,6 @@ impl Drop for LayerInner {
 
         let path = std::mem::take(&mut self.path);
         let file_name = self.layer_desc().layer_name();
-        let file_size = self.layer_desc().file_size;
         let meta = self.metadata();
         let status = self.status.take();
 
@@ -786,20 +747,13 @@ impl Drop for LayerInner {
 
             // carry this until we are finished for [`Layer::wait_drop`] support
             let _status = status;
-
             let Some(timeline) = timeline else {
                 // no need to nag that timeline is gone: under normal situation on
                 // task_mgr::remove_tenant_from_memory the timeline is gone before we get dropped.
-                LAYER_IMPL_METRICS.inc_deletes_failed(DeleteFailed::TimelineGone);
                 return;
             };
 
-            let Ok(_guard) = timeline.gate.enter() else {
-                LAYER_IMPL_METRICS.inc_deletes_failed(DeleteFailed::TimelineGone);
-                return;
-            };
-
-            let removed = match std::fs::remove_file(path) {
+           match std::fs::remove_file(path) {
                 Ok(()) => true,
                 Err(e) if e.kind() == std::io::ErrorKind::NotFound => {
                     // until we no longer do detaches by removing all local files before removing the
@@ -810,34 +764,16 @@ impl Drop for LayerInner {
                     // layers.
                     false
                 }
-                Err(e) => {
-                    tracing::error!("failed to remove wanted deleted layer: {e}");
-                    LAYER_IMPL_METRICS.inc_delete_removes_failed();
+                Err(_e) => {
                     false
                 }
             };
 
-            if removed {
-                timeline.metrics.resident_physical_size_sub(file_size);
-            }
-            let res = timeline
+            
+            let _a=timeline
                 .remote_client
                 .schedule_deletion_of_unlinked(vec![(file_name, meta)]);
 
-            if let Err(e) = res {
-                // test_timeline_deletion_with_files_stuck_in_upload_queue is good at
-                // demonstrating this deadlock (without spawn_blocking): stop will drop
-                // queued items, which will have ResidentLayer's, and those drops would try
-                // to re-entrantly lock the RemoteTimelineClient inner state.
-                if !timeline.is_active() {
-                    tracing::info!("scheduling deletion on drop failed: {e:#}");
-                } else {
-                    tracing::warn!("scheduling deletion on drop failed: {e:#}");
-                }
-                LAYER_IMPL_METRICS.inc_deletes_failed(DeleteFailed::DeleteSchedulingFailed);
-            } else {
-                LAYER_IMPL_METRICS.inc_completed_deletes();
-            }
         });
     }
 }
@@ -868,12 +804,6 @@ impl LayerInner {
         // This object acts as a RAII guard on these metrics: increment on construction
         timeline.metrics.inc_layer(&desc);
 
-        // New layers are visible by default. This metric is later updated on drop or in set_visibility
-        timeline
-            .metrics
-            .visible_physical_size_gauge
-            .add(desc.file_size);
-
         LayerInner {
             conf,
             path: local_path,
@@ -894,13 +824,9 @@ impl LayerInner {
     }
 
     fn delete_on_drop(&self) {
-        let res =
-            self.wanted_deleted
+                    let _a=self.wanted_deleted
                 .compare_exchange(false, true, Ordering::Release, Ordering::Relaxed);
 
-        if res.is_ok() {
-            LAYER_IMPL_METRICS.inc_started_deletes();
-        }
     }
 
     /// Cancellation safe, however dropping the future and calling this method again might result
@@ -938,12 +864,6 @@ impl LayerInner {
             // drop the DownloadedLayer outside of the holding the guard
             drop(strong);
 
-            // idea here is that only one evicter should ever get to witness a strong reference,
-            // which means whenever get_or_maybe_download upgrades a weak, it must mark up a
-            // cancelled eviction and signal us, like it currently does.
-            //
-            // a second concurrent evict_and_wait will not see a strong reference.
-            LAYER_IMPL_METRICS.inc_started_evictions();
         }
 
         let changed = rx.changed();
@@ -983,15 +903,13 @@ impl LayerInner {
             // get_or_init_detached can:
             // - be fast (mutex lock) OR uncontested semaphore permit acquire
             // - be slow (wait for semaphore permit or closing)
-            let init_cancelled = scopeguard::guard((), |_| LAYER_IMPL_METRICS.inc_init_cancelled());
-
             let locked = self
                 .inner
                 .get_or_init_detached_measured(Some(&mut wait_for_download_recorder))
                 .await
                 .map(|mut guard| guard.get_and_upgrade().ok_or(guard));
 
-            scopeguard::ScopeGuard::into_inner(init_cancelled);
+
 
             match locked {
                 // this path could had been a RwLock::read
@@ -1004,8 +922,7 @@ impl LayerInner {
                     // note that we also have dropped the Guard; this is fine, because we just made
                     // a state change and are holding a strong reference to be returned.
                     self.status.as_ref().unwrap().send_replace(Status::Resident);
-                    LAYER_IMPL_METRICS
-                        .inc_eviction_cancelled(EvictionCancelled::UpgradedBackOnAccess);
+        
 
                     return Ok(strong);
                 }
@@ -1032,8 +949,7 @@ impl LayerInner {
             .upgrade()
             .ok_or(DownloadError::TimelineShutdown)?;
 
-        // count cancellations, which currently remain largely unexpected
-        let init_cancelled = scopeguard::guard((), |_| LAYER_IMPL_METRICS.inc_init_cancelled());
+        
 
         // check if we really need to be downloaded: this can happen if a read access won the
         // semaphore before eviction.
@@ -1045,7 +961,6 @@ impl LayerInner {
             .await
             .map_err(DownloadError::PreStatFailed);
 
-        scopeguard::ScopeGuard::into_inner(init_cancelled);
 
         let needs_download = needs_download?;
 
@@ -1056,7 +971,7 @@ impl LayerInner {
             self.failpoint(failpoints::FailpointKind::AfterDeterminingLayerNeedsNoDownload)
                 .await?;
 
-            LAYER_IMPL_METRICS.inc_init_needed_no_download();
+            
 
             return Ok(self.initialize_after_layer_is_on_disk(permit));
         };
@@ -1097,13 +1012,13 @@ impl LayerInner {
         async move {
             tracing::info!(%reason, "downloading on-demand");
 
-            let init_cancelled = scopeguard::guard((), |_| LAYER_IMPL_METRICS.inc_init_cancelled());
+            
             let res = self
                 .download_init_and_wait(timeline, permit, ctx.attached_child())
                 .maybe_perf_instrument(&ctx, |crnt_perf_span| crnt_perf_span.clone())
                 .await?;
 
-            scopeguard::ScopeGuard::into_inner(init_cancelled);
+      
             Ok(res)
         }
         .instrument(tracing::info_span!("get_or_maybe_download", layer=%self))
@@ -1121,7 +1036,6 @@ impl LayerInner {
                     "unexpectedly on-demand downloading for task kind {:?}",
                     ctx.task_kind()
                 );
-                crate::metrics::UNEXPECTED_ONDEMAND_DOWNLOADS.inc();
 
                 let really_error =
                     matches!(b, Error) && !self.conf.ondemand_download_behavior_treat_error_as_warn;
@@ -1173,20 +1087,7 @@ impl LayerInner {
 
                 let res = this.download_and_init(timeline, permit, &ctx).await;
 
-                if let Err(res) = tx.send(res) {
-                    match res {
-                        Ok(_res) => {
-                            tracing::debug!("layer initialized, but caller has been cancelled");
-                            LAYER_IMPL_METRICS.inc_init_completed_without_requester();
-                        }
-                        Err(e) => {
-                            tracing::info!(
-                                "layer file download failed, and caller has been cancelled: {e:?}"
-                            );
-                            LAYER_IMPL_METRICS.inc_download_failed_without_requester();
-                        }
-                    }
-                }
+                let _a =tx.send(res);
             }
             .in_current_span(),
         );
@@ -1238,21 +1139,9 @@ impl LayerInner {
                     }
                 };
                 tracing::info!(size=%self.desc.file_size, %latency_millis, "on-demand download successful");
-                timeline
-                    .metrics
-                    .resident_physical_size_add(self.desc.file_size);
+    
                 self.consecutive_failures.store(0, Ordering::Relaxed);
 
-                let since_last_eviction = self
-                    .last_evicted_at
-                    .lock()
-                    .unwrap()
-                    .take()
-                    .map(|ts| ts.elapsed());
-                if let Some(since_last_eviction) = since_last_eviction {
-                    LAYER_IMPL_METRICS.record_redownloaded_after(since_last_eviction);
-                }
-
                 self.access_stats.record_residence_event();
 
                 Ok(self.initialize_after_layer_is_on_disk(permit))
@@ -1407,12 +1296,8 @@ impl LayerInner {
 
             tracing::debug!("eviction started");
 
-            let res = self.wait_for_turn_and_evict(only_version).await;
-            // metrics: ignore the Ok branch, it is not done yet
-            if let Err(e) = res {
-                tracing::debug!(res=?Err::<(), _>(&e), "eviction completed");
-                LAYER_IMPL_METRICS.inc_eviction_cancelled(e);
-            }
+           let _a = self.wait_for_turn_and_evict(only_version).await;
+            
         };
 
         Self::spawn(start_evicting.instrument(span));
@@ -1532,21 +1417,13 @@ impl LayerInner {
         Self::spawn_blocking(move || {
             let _span = span.entered();
 
-            let res = self.evict_blocking(&timeline, &gate, &permit);
+            let res = self.evict_blocking( &gate, &permit);
 
             let waiters = self.inner.initializer_count();
 
-            if waiters > 0 {
-                LAYER_IMPL_METRICS.inc_evicted_with_waiters();
-            }
-
+            
             let completed_in = spawned_at.elapsed();
-            LAYER_IMPL_METRICS.record_time_to_evict(completed_in);
-
-            match res {
-                Ok(()) => LAYER_IMPL_METRICS.inc_completed_evictions(),
-                Err(e) => LAYER_IMPL_METRICS.inc_eviction_cancelled(e),
-            }
+           
 
             tracing::debug!(?res, elapsed_ms=%completed_in.as_millis(), %waiters, "eviction completed");
         });
@@ -1557,7 +1434,6 @@ impl LayerInner {
     /// This is blocking only to do just one spawn_blocking hop compared to multiple via tokio::fs.
     fn evict_blocking(
         &self,
-        timeline: &Timeline,
         _gate: &gate::GateGuard,
         _permit: &heavier_once_cell::InitPermit,
     ) -> Result<(), EvictionCancelled> {
@@ -1570,17 +1446,7 @@ impl LayerInner {
                     Ok(elapsed) => {
                         let accessed_and_visible = self.access_stats.accessed()
                             && self.access_stats.visibility() == LayerVisibilityHint::Visible;
-                        if accessed_and_visible {
-                            // Only layers used for reads contribute to our "low residence" metric that is used
-                            // to detect thrashing.  Layers promoted for other reasons (e.g. compaction) are allowed
-                            // to be rapidly evicted without contributing to this metric.
-                            timeline
-                                .metrics
-                                .evictions_with_low_residence_duration
-                                .read()
-                                .unwrap()
-                                .observe(elapsed);
-                        }
+                        
 
                         tracing::info!(
                             residence_millis = elapsed.as_millis(),
@@ -1592,10 +1458,6 @@ impl LayerInner {
                         tracing::info!("evicted layer after unknown residence period");
                     }
                 }
-                timeline.metrics.evictions.inc();
-                timeline
-                    .metrics
-                    .resident_physical_size_sub(self.desc.file_size);
             }
             Err(e) if e.kind() == std::io::ErrorKind::NotFound => {
                 tracing::error!(
@@ -1812,13 +1674,7 @@ impl DownloadedLayer {
             match res {
                 Ok(layer) => Ok(layer),
                 Err(err) => {
-                    LAYER_IMPL_METRICS.inc_permanent_loading_failures();
-                    // We log this message once over the lifetime of `Self`
-                    // => Ok and good to log backtrace and path here.
-                    tracing::error!(
-                        "layer load failed, assuming permanent failure: {}: {err:?}",
-                        owner.path
-                    );
+                    
                     Err(err)
                 }
             }
@@ -2026,218 +1882,6 @@ impl From<ResidentLayer> for Layer {
     }
 }
 
-use metrics::IntCounter;
-
-pub(crate) struct LayerImplMetrics {
-    started_evictions: IntCounter,
-    completed_evictions: IntCounter,
-    cancelled_evictions: enum_map::EnumMap<EvictionCancelled, IntCounter>,
-
-    started_deletes: IntCounter,
-    completed_deletes: IntCounter,
-    failed_deletes: enum_map::EnumMap<DeleteFailed, IntCounter>,
-
-    rare_counters: enum_map::EnumMap<RareEvent, IntCounter>,
-    inits_cancelled: metrics::core::GenericCounter<metrics::core::AtomicU64>,
-    redownload_after: metrics::Histogram,
-    time_to_evict: metrics::Histogram,
-}
-
-impl Default for LayerImplMetrics {
-    fn default() -> Self {
-        use enum_map::Enum;
-
-        // reminder: these will be pageserver_layer_* with "_total" suffix
-
-        let started_evictions = metrics::register_int_counter!(
-            "pageserver_layer_started_evictions",
-            "Evictions started in the Layer implementation"
-        )
-        .unwrap();
-        let completed_evictions = metrics::register_int_counter!(
-            "pageserver_layer_completed_evictions",
-            "Evictions completed in the Layer implementation"
-        )
-        .unwrap();
-
-        let cancelled_evictions = metrics::register_int_counter_vec!(
-            "pageserver_layer_cancelled_evictions_count",
-            "Different reasons for evictions to have been cancelled or failed",
-            &["reason"]
-        )
-        .unwrap();
-
-        let cancelled_evictions = enum_map::EnumMap::from_array(std::array::from_fn(|i| {
-            let reason = EvictionCancelled::from_usize(i);
-            let s = reason.as_str();
-            cancelled_evictions.with_label_values(&[s])
-        }));
-
-        let started_deletes = metrics::register_int_counter!(
-            "pageserver_layer_started_deletes",
-            "Deletions on drop pending in the Layer implementation"
-        )
-        .unwrap();
-        let completed_deletes = metrics::register_int_counter!(
-            "pageserver_layer_completed_deletes",
-            "Deletions on drop completed in the Layer implementation"
-        )
-        .unwrap();
-
-        let failed_deletes = metrics::register_int_counter_vec!(
-            "pageserver_layer_failed_deletes_count",
-            "Different reasons for deletions on drop to have failed",
-            &["reason"]
-        )
-        .unwrap();
-
-        let failed_deletes = enum_map::EnumMap::from_array(std::array::from_fn(|i| {
-            let reason = DeleteFailed::from_usize(i);
-            let s = reason.as_str();
-            failed_deletes.with_label_values(&[s])
-        }));
-
-        let rare_counters = metrics::register_int_counter_vec!(
-            "pageserver_layer_assumed_rare_count",
-            "Times unexpected or assumed rare event happened",
-            &["event"]
-        )
-        .unwrap();
-
-        let rare_counters = enum_map::EnumMap::from_array(std::array::from_fn(|i| {
-            let event = RareEvent::from_usize(i);
-            let s = event.as_str();
-            rare_counters.with_label_values(&[s])
-        }));
-
-        let inits_cancelled = metrics::register_int_counter!(
-            "pageserver_layer_inits_cancelled_count",
-            "Times Layer initialization was cancelled",
-        )
-        .unwrap();
-
-        let redownload_after = {
-            let minute = 60.0;
-            let hour = 60.0 * minute;
-            metrics::register_histogram!(
-                "pageserver_layer_redownloaded_after",
-                "Time between evicting and re-downloading.",
-                vec![
-                    10.0,
-                    30.0,
-                    minute,
-                    5.0 * minute,
-                    15.0 * minute,
-                    30.0 * minute,
-                    hour,
-                    12.0 * hour,
-                ]
-            )
-            .unwrap()
-        };
-
-        let time_to_evict = metrics::register_histogram!(
-            "pageserver_layer_eviction_held_permit_seconds",
-            "Time eviction held the permit.",
-            vec![0.001, 0.010, 0.100, 0.500, 1.000, 5.000]
-        )
-        .unwrap();
-
-        Self {
-            started_evictions,
-            completed_evictions,
-            cancelled_evictions,
-
-            started_deletes,
-            completed_deletes,
-            failed_deletes,
-
-            rare_counters,
-            inits_cancelled,
-            redownload_after,
-            time_to_evict,
-        }
-    }
-}
-
-impl LayerImplMetrics {
-    fn inc_started_evictions(&self) {
-        self.started_evictions.inc();
-    }
-    fn inc_completed_evictions(&self) {
-        self.completed_evictions.inc();
-    }
-    fn inc_eviction_cancelled(&self, reason: EvictionCancelled) {
-        self.cancelled_evictions[reason].inc()
-    }
-
-    fn inc_started_deletes(&self) {
-        self.started_deletes.inc();
-    }
-    fn inc_completed_deletes(&self) {
-        self.completed_deletes.inc();
-    }
-    fn inc_deletes_failed(&self, reason: DeleteFailed) {
-        self.failed_deletes[reason].inc();
-    }
-
-    /// Counted separatedly from failed layer deletes because we will complete the layer deletion
-    /// attempt regardless of failure to delete local file.
-    fn inc_delete_removes_failed(&self) {
-        self.rare_counters[RareEvent::RemoveOnDropFailed].inc();
-    }
-
-    /// Expected rare just as cancellations are rare, but we could have cancellations separate from
-    /// the single caller which can start the download, so use this counter to separte them.
-    fn inc_init_completed_without_requester(&self) {
-        self.rare_counters[RareEvent::InitCompletedWithoutRequester].inc();
-    }
-
-    /// Expected rare because cancellations are unexpected, and failures are unexpected
-    fn inc_download_failed_without_requester(&self) {
-        self.rare_counters[RareEvent::DownloadFailedWithoutRequester].inc();
-    }
-
-    /// The Weak in ResidentOrWantedEvicted::WantedEvicted was successfully upgraded.
-    ///
-    /// If this counter is always zero, we should replace ResidentOrWantedEvicted type with an
-    /// Option.
-    fn inc_raced_wanted_evicted_accesses(&self) {
-        self.rare_counters[RareEvent::UpgradedWantedEvicted].inc();
-    }
-
-    /// These are only expected for [`Self::inc_init_cancelled`] amount when
-    /// running with remote storage.
-    fn inc_init_needed_no_download(&self) {
-        self.rare_counters[RareEvent::InitWithoutDownload].inc();
-    }
-
-    /// Expected rare because all layer files should be readable and good
-    fn inc_permanent_loading_failures(&self) {
-        self.rare_counters[RareEvent::PermanentLoadingFailure].inc();
-    }
-
-    fn inc_init_cancelled(&self) {
-        self.inits_cancelled.inc()
-    }
-
-    fn record_redownloaded_after(&self, duration: std::time::Duration) {
-        self.redownload_after.observe(duration.as_secs_f64())
-    }
-
-    /// This would be bad if it ever happened, or mean extreme disk pressure. We should probably
-    /// instead cancel eviction if we would have read waiters. We cannot however separate reads
-    /// from other evictions, so this could have noise as well.
-    fn inc_evicted_with_waiters(&self) {
-        self.rare_counters[RareEvent::EvictedWithWaiters].inc();
-    }
-
-    /// Recorded at least initially as the permit is now acquired in async context before
-    /// spawn_blocking action.
-    fn record_time_to_evict(&self, duration: std::time::Duration) {
-        self.time_to_evict.observe(duration.as_secs_f64())
-    }
-}
 
 #[derive(Debug, Clone, Copy, enum_map::Enum)]
 enum EvictionCancelled {
@@ -2254,21 +1898,6 @@ enum EvictionCancelled {
     UnexpectedEvictedState,
 }
 
-impl EvictionCancelled {
-    fn as_str(&self) -> &'static str {
-        match self {
-            EvictionCancelled::LayerGone => "layer_gone",
-            EvictionCancelled::TimelineGone => "timeline_gone",
-            EvictionCancelled::VersionCheckFailed => "version_check_fail",
-            EvictionCancelled::FileNotFound => "file_not_found",
-            EvictionCancelled::RemoveFailed => "remove_failed",
-            EvictionCancelled::AlreadyReinitialized => "already_reinitialized",
-            EvictionCancelled::LostToDownload => "lost_to_download",
-            EvictionCancelled::UpgradedBackOnAccess => "upgraded_back_on_access",
-            EvictionCancelled::UnexpectedEvictedState => "unexpected_evicted_state",
-        }
-    }
-}
 
 #[derive(enum_map::Enum)]
 enum DeleteFailed {
@@ -2276,15 +1905,6 @@ enum DeleteFailed {
     DeleteSchedulingFailed,
 }
 
-impl DeleteFailed {
-    fn as_str(&self) -> &'static str {
-        match self {
-            DeleteFailed::TimelineGone => "timeline_gone",
-            DeleteFailed::DeleteSchedulingFailed => "delete_scheduling_failed",
-        }
-    }
-}
-
 #[derive(enum_map::Enum)]
 enum RareEvent {
     RemoveOnDropFailed,
@@ -2296,21 +1916,3 @@ enum RareEvent {
     EvictedWithWaiters,
 }
 
-impl RareEvent {
-    fn as_str(&self) -> &'static str {
-        use RareEvent::*;
-
-        match self {
-            RemoveOnDropFailed => "remove_on_drop_failed",
-            InitCompletedWithoutRequester => "init_completed_without",
-            DownloadFailedWithoutRequester => "download_failed_without",
-            UpgradedWantedEvicted => "raced_wanted_evicted",
-            InitWithoutDownload => "init_needed_no_download",
-            PermanentLoadingFailure => "permanent_loading_failure",
-            EvictedWithWaiters => "evicted_with_waiters",
-        }
-    }
-}
-
-pub(crate) static LAYER_IMPL_METRICS: once_cell::sync::Lazy<LayerImplMetrics> =
-    once_cell::sync::Lazy::new(LayerImplMetrics::default);

pageserver/src/tenant/storage_layer/layer/tests.rs

@@ -238,7 +238,7 @@ async fn smoke_test() {
         rtc.get_remote_physical_size(),
         dummy_layer.metadata().file_size
     );
-    assert_eq!(0, LAYER_IMPL_METRICS.inits_cancelled.get())
+   
 }
 
 /// This test demonstrates a previous hang when a eviction and deletion were requested at the same
@@ -311,11 +311,6 @@ async fn evict_and_wait_on_wanted_deleted() {
 
     SpawnBlockingPoolHelper::consume_and_release_all_of_spawn_blocking_threads(&handle).await;
 
-    assert_eq!(1, LAYER_IMPL_METRICS.started_deletes.get());
-    assert_eq!(1, LAYER_IMPL_METRICS.completed_deletes.get());
-    assert_eq!(1, LAYER_IMPL_METRICS.started_evictions.get());
-    assert_eq!(1, LAYER_IMPL_METRICS.completed_evictions.get());
-    assert_eq!(0, LAYER_IMPL_METRICS.inits_cancelled.get())
 }
 
 /// This test ensures we are able to read the layer while the layer eviction has been
@@ -366,7 +361,7 @@ fn read_wins_pending_eviction() {
         tokio::time::timeout(ADVANCE, &mut evict_and_wait)
             .await
             .expect_err("should had been a timeout since we are holding the layer resident");
-        assert_eq!(1, LAYER_IMPL_METRICS.started_evictions.get());
+       
 
         let (completion, barrier) = utils::completion::channel();
         let (arrival, arrived_at_barrier) = utils::completion::channel();
@@ -398,18 +393,7 @@ fn read_wins_pending_eviction() {
 
         // works as intended: evictions lose to "downloads"
         assert!(matches!(e, EvictionError::Downloaded), "{e:?}");
-        assert_eq!(0, LAYER_IMPL_METRICS.completed_evictions.get());
-
-        // this is not wrong: the eviction is technically still "on the way" as it's still queued
-        // because of a failpoint
-        assert_eq!(
-            0,
-            LAYER_IMPL_METRICS
-                .cancelled_evictions
-                .values()
-                .map(|ctr| ctr.get())
-                .sum::<u64>()
-        );
+        
 
         drop(completion);
 
@@ -417,26 +401,9 @@ fn read_wins_pending_eviction() {
         SpawnBlockingPoolHelper::consume_and_release_all_of_spawn_blocking_threads0(&handle, 1)
             .await;
 
-        assert_eq!(0, LAYER_IMPL_METRICS.completed_evictions.get());
+        
 
-        // now we finally can observe the original eviction failing
-        // it would had been possible to observe it earlier, but here it is guaranteed to have
-        // happened.
-        assert_eq!(
-            1,
-            LAYER_IMPL_METRICS
-                .cancelled_evictions
-                .values()
-                .map(|ctr| ctr.get())
-                .sum::<u64>()
-        );
-
-        assert_eq!(
-            1,
-            LAYER_IMPL_METRICS.cancelled_evictions[EvictionCancelled::AlreadyReinitialized].get()
-        );
-
-        assert_eq!(0, LAYER_IMPL_METRICS.inits_cancelled.get())
+        
     });
 }
 
@@ -499,7 +466,7 @@ fn multiple_pending_evictions_scenario(name: &'static str, in_order: bool) {
         tokio::time::timeout(ADVANCE, &mut evict_and_wait)
             .await
             .expect_err("should had been a timeout since we are holding the layer resident");
-        assert_eq!(1, LAYER_IMPL_METRICS.started_evictions.get());
+        
 
         let (completion1, barrier) = utils::completion::channel();
         let mut completion1 = Some(completion1);
@@ -534,20 +501,9 @@ fn multiple_pending_evictions_scenario(name: &'static str, in_order: bool) {
 
         // works as intended: evictions lose to "downloads"
         assert!(matches!(e, EvictionError::Downloaded), "{e:?}");
-        assert_eq!(0, LAYER_IMPL_METRICS.completed_evictions.get());
+       
 
-        // this is not wrong: the eviction is technically still "on the way" as it's still queued
-        // because of a failpoint
-        assert_eq!(
-            0,
-            LAYER_IMPL_METRICS
-                .cancelled_evictions
-                .values()
-                .map(|ctr| ctr.get())
-                .sum::<u64>()
-        );
-
-        assert_eq!(0, LAYER_IMPL_METRICS.completed_evictions.get());
+       
 
         // configure another failpoint for the second eviction -- evictions are per initialization,
         // so now that we've reinitialized the inner, we get to run two of them at the same time.
@@ -567,13 +523,10 @@ fn multiple_pending_evictions_scenario(name: &'static str, in_order: bool) {
 
         arrived_at_barrier.wait().await;
 
-        assert_eq!(2, LAYER_IMPL_METRICS.started_evictions.get());
+       
 
-        let mut release_earlier_eviction = |expected_reason| {
-            assert_eq!(
-                0,
-                LAYER_IMPL_METRICS.cancelled_evictions[expected_reason].get(),
-            );
+        let mut release_earlier_eviction = |_expected_reason| {
+            
 
             drop(completion1.take().unwrap());
 
@@ -586,10 +539,7 @@ fn multiple_pending_evictions_scenario(name: &'static str, in_order: bool) {
                 )
                 .await;
 
-                assert_eq!(
-                    1,
-                    LAYER_IMPL_METRICS.cancelled_evictions[expected_reason].get(),
-                );
+                
             }
         };
 
@@ -612,19 +562,7 @@ fn multiple_pending_evictions_scenario(name: &'static str, in_order: bool) {
             .expect("eviction goes through now that spawn_blocking is unclogged")
             .expect("eviction should succeed, because version matches");
 
-        assert_eq!(1, LAYER_IMPL_METRICS.completed_evictions.get());
-
-        // ensure the cancelled are unchanged
-        assert_eq!(
-            1,
-            LAYER_IMPL_METRICS
-                .cancelled_evictions
-                .values()
-                .map(|ctr| ctr.get())
-                .sum::<u64>()
-        );
-
-        assert_eq!(0, LAYER_IMPL_METRICS.inits_cancelled.get())
+       
     });
 }
 
@@ -714,8 +652,7 @@ async fn cancelled_get_or_maybe_download_does_not_cancel_eviction() {
         .unwrap_err();
     assert!(matches!(e, DownloadError::DownloadRequired), "{e:?}");
 
-    // failpoint is not counted as cancellation either
-    assert_eq!(0, LAYER_IMPL_METRICS.inits_cancelled.get())
+    
 }
 
 #[tokio::test(start_paused = true)]
@@ -892,8 +829,7 @@ async fn eviction_cancellation_on_drop() {
                 .expect_err("should had been a timeout since we are holding the layer resident");
         }
 
-        // 1 == we only evict one of the layers
-        assert_eq!(1, LAYER_IMPL_METRICS.started_evictions.get());
+        
 
         drop(resident);
 
@@ -902,10 +838,7 @@ async fn eviction_cancellation_on_drop() {
 
         SpawnBlockingPoolHelper::consume_and_release_all_of_spawn_blocking_threads(&handle).await;
 
-        assert_eq!(
-            1,
-            LAYER_IMPL_METRICS.cancelled_evictions[EvictionCancelled::LayerGone].get()
-        );
+        
     }
 }
 

pageserver/src/tenant/tasks.rs

@@ -10,7 +10,7 @@ use std::time::{Duration, Instant};
 use once_cell::sync::Lazy;
 use pageserver_api::config::tenant_conf_defaults::DEFAULT_COMPACTION_PERIOD;
 use rand::Rng;
-use scopeguard::defer;
+
 use tokio::sync::{Semaphore, SemaphorePermit};
 use tokio_util::sync::CancellationToken;
 use tracing::*;
@@ -19,7 +19,6 @@ use utils::completion::Barrier;
 use utils::pausable_failpoint;
 
 use crate::context::{DownloadBehavior, RequestContext};
-use crate::metrics::{self, BackgroundLoopSemaphoreMetricsRecorder, TENANT_TASK_EVENTS};
 use crate::task_mgr::{self, BACKGROUND_RUNTIME, TOKIO_WORKER_THREADS, TaskKind};
 use crate::tenant::throttle::Stats;
 use crate::tenant::timeline::CompactionError;
@@ -85,17 +84,15 @@ pub(crate) enum BackgroundLoopKind {
     SecondaryDownload,
 }
 
-pub struct BackgroundLoopSemaphorePermit<'a> {
+pub struct BackgroundLoopSemaphorePermit {
     _permit: SemaphorePermit<'static>,
-    _recorder: BackgroundLoopSemaphoreMetricsRecorder<'a>,
 }
 
 /// Acquires a semaphore permit, to limit concurrent background jobs.
 pub(crate) async fn acquire_concurrency_permit(
     loop_kind: BackgroundLoopKind,
     _ctx: &RequestContext,
-) -> BackgroundLoopSemaphorePermit<'static> {
-    let mut recorder = metrics::BACKGROUND_LOOP_SEMAPHORE.record(loop_kind);
+) -> BackgroundLoopSemaphorePermit {
 
     if loop_kind == BackgroundLoopKind::InitialLogicalSizeCalculation {
         pausable_failpoint!("initial-size-calculation-permit-pause");
@@ -108,11 +105,9 @@ pub(crate) async fn acquire_concurrency_permit(
     };
     let permit = semaphore.acquire().await.expect("should never close");
 
-    recorder.acquired();
 
     BackgroundLoopSemaphorePermit {
         _permit: permit,
-        _recorder: recorder,
     }
 }
 
@@ -135,8 +130,7 @@ pub fn start_background_loops(tenant: &Arc<Tenant>, can_start: Option<&Barrier>)
                     _ = cancel.cancelled() => return Ok(()),
                     _ = Barrier::maybe_wait(can_start) => {}
                 };
-                TENANT_TASK_EVENTS.with_label_values(&["start"]).inc();
-                defer!(TENANT_TASK_EVENTS.with_label_values(&["stop"]).inc());
+
                 compaction_loop(tenant, cancel)
                     // If you rename this span, change the RUST_LOG env variable in test_runner/performance/test_branch_creation.py
                     .instrument(info_span!("compaction_loop", tenant_id = %tenant_shard_id.tenant_id, shard_id = %tenant_shard_id.shard_slug()))
@@ -161,8 +155,6 @@ pub fn start_background_loops(tenant: &Arc<Tenant>, can_start: Option<&Barrier>)
                     _ = cancel.cancelled() => return Ok(()),
                     _ = Barrier::maybe_wait(can_start) => {}
                 };
-                TENANT_TASK_EVENTS.with_label_values(&["start"]).inc();
-                defer!(TENANT_TASK_EVENTS.with_label_values(&["stop"]).inc());
                 gc_loop(tenant, cancel)
                     .instrument(info_span!("gc_loop", tenant_id = %tenant_shard_id.tenant_id, shard_id = %tenant_shard_id.shard_slug()))
                     .await;
@@ -186,8 +178,7 @@ pub fn start_background_loops(tenant: &Arc<Tenant>, can_start: Option<&Barrier>)
                     _ = cancel.cancelled() => return Ok(()),
                     _ = Barrier::maybe_wait(can_start) => {}
                 };
-                TENANT_TASK_EVENTS.with_label_values(&["start"]).inc();
-                defer!(TENANT_TASK_EVENTS.with_label_values(&["stop"]).inc());
+
                 tenant_housekeeping_loop(tenant, cancel)
                     .instrument(info_span!("tenant_housekeeping_loop", tenant_id = %tenant_shard_id.tenant_id, shard_id = %tenant_shard_id.shard_slug()))
                     .await;
@@ -592,8 +583,5 @@ pub(crate) fn warn_when_period_overrun(
             ?task,
             "task iteration took longer than the configured period"
         );
-        metrics::BACKGROUND_LOOP_PERIOD_OVERRUN_COUNT
-            .with_label_values(&[task.into(), &format!("{}", period.as_secs())])
-            .inc();
     }
 }

pageserver/src/tenant/throttle.rs

@@ -45,8 +45,8 @@ pub struct Stats {
 }
 
 pub enum ThrottleResult {
-    NotThrottled { end: Instant },
-    Throttled { end: Instant },
+    NotThrottled {  },
+    Throttled {  },
 }
 
 impl Throttle {
@@ -114,7 +114,7 @@ impl Throttle {
         let inner = self.inner.load_full(); // clones the `Inner` Arc
 
         if !inner.enabled {
-            return ThrottleResult::NotThrottled { end: start };
+            return ThrottleResult::NotThrottled { };
         }
 
         self.count_accounted_start.fetch_add(1, Ordering::Relaxed);
@@ -127,9 +127,9 @@ impl Throttle {
             let wait_time = end - start;
             self.sum_throttled_usecs
                 .fetch_add(wait_time.as_micros() as u64, Ordering::Relaxed);
-            ThrottleResult::Throttled { end }
+            ThrottleResult::Throttled {  }
         } else {
-            ThrottleResult::NotThrottled { end: start }
+            ThrottleResult::NotThrottled { }
         }
     }
 }

pageserver/src/tenant/timeline/compaction.rs

@@ -56,7 +56,8 @@ use crate::tenant::storage_layer::batch_split_writer::{
 use crate::tenant::storage_layer::filter_iterator::FilterIterator;
 use crate::tenant::storage_layer::merge_iterator::MergeIterator;
 use crate::tenant::storage_layer::{
-    AsLayerDesc, PersistentLayerDesc, PersistentLayerKey, ValueReconstructState,
+    AsLayerDesc, LayerVisibilityHint, PersistentLayerDesc, PersistentLayerKey,
+    ValueReconstructState,
 };
 use crate::tenant::tasks::log_compaction_error;
 use crate::tenant::timeline::{
@@ -69,6 +70,13 @@ use crate::virtual_file::{MaybeFatalIo, VirtualFile};
 /// Maximum number of deltas before generating an image layer in bottom-most compaction.
 const COMPACTION_DELTA_THRESHOLD: usize = 5;
 
+/// Ratio of shard-local pages below which we trigger shard ancestor layer rewrites. 0.3 means that
+/// <= 30% of layer pages must belong to the descendant shard to rewrite the layer.
+///
+/// We choose a value < 0.5 to avoid rewriting all visible layers every time we do a power-of-two
+/// shard split, which gets expensive for large tenants.
+const ANCESTOR_COMPACTION_REWRITE_THRESHOLD: f64 = 0.3;
+
 #[derive(Debug, Clone, Copy, Hash, PartialEq, Eq)]
 pub struct GcCompactionJobId(pub usize);
 
@@ -80,6 +88,7 @@ impl std::fmt::Display for GcCompactionJobId {
 
 pub struct GcCompactionCombinedSettings {
     pub gc_compaction_enabled: bool,
+    pub gc_compaction_verification: bool,
     pub gc_compaction_initial_threshold_kb: u64,
     pub gc_compaction_ratio_percent: u64,
 }
@@ -225,6 +234,7 @@ impl GcCompactionQueue {
             gc_compaction_enabled,
             gc_compaction_initial_threshold_kb,
             gc_compaction_ratio_percent,
+            ..
         } = timeline.get_gc_compaction_settings();
         if !gc_compaction_enabled {
             return Ok(());
@@ -747,8 +757,8 @@ impl KeyHistoryRetention {
     async fn pipe_to(
         self,
         key: Key,
-        delta_writer: &mut SplitDeltaLayerWriter,
-        mut image_writer: Option<&mut SplitImageLayerWriter>,
+        delta_writer: &mut SplitDeltaLayerWriter<'_>,
+        mut image_writer: Option<&mut SplitImageLayerWriter<'_>>,
         stat: &mut CompactionStatistics,
         ctx: &RequestContext,
     ) -> anyhow::Result<()> {
@@ -788,6 +798,123 @@ impl KeyHistoryRetention {
         }
         Ok(())
     }
+
+    /// Verify if every key in the retention is readable by replaying the logs.
+    async fn verify(
+        &self,
+        key: Key,
+        base_img_from_ancestor: &Option<(Key, Lsn, Bytes)>,
+        full_history: &[(Key, Lsn, Value)],
+        tline: &Arc<Timeline>,
+    ) -> anyhow::Result<()> {
+        // Usually the min_lsn should be the first record but we do a full iteration to be safe.
+        let Some(min_lsn) = full_history.iter().map(|(_, lsn, _)| *lsn).min() else {
+            // This should never happen b/c if we don't have any history of a key, we won't even do `generate_key_retention`.
+            return Ok(());
+        };
+        let Some(max_lsn) = full_history.iter().map(|(_, lsn, _)| *lsn).max() else {
+            // This should never happen b/c if we don't have any history of a key, we won't even do `generate_key_retention`.
+            return Ok(());
+        };
+        let mut base_img = base_img_from_ancestor
+            .as_ref()
+            .map(|(_, lsn, img)| (*lsn, img));
+        let mut history = Vec::new();
+
+        async fn collect_and_verify(
+            key: Key,
+            lsn: Lsn,
+            base_img: &Option<(Lsn, &Bytes)>,
+            history: &[(Lsn, &NeonWalRecord)],
+            tline: &Arc<Timeline>,
+            skip_empty: bool,
+        ) -> anyhow::Result<()> {
+            if base_img.is_none() && history.is_empty() {
+                if skip_empty {
+                    return Ok(());
+                }
+                anyhow::bail!("verification failed: key {} has no history at {}", key, lsn);
+            };
+
+            let mut records = history
+                .iter()
+                .map(|(lsn, val)| (*lsn, (*val).clone()))
+                .collect::<Vec<_>>();
+
+            // WAL redo requires records in the reverse LSN order
+            records.reverse();
+            let data = ValueReconstructState {
+                img: base_img.as_ref().map(|(lsn, img)| (*lsn, (*img).clone())),
+                records,
+            };
+
+            tline
+                .reconstruct_value(key, lsn, data, RedoAttemptType::GcCompaction)
+                .await
+                .with_context(|| format!("verification failed for key {} at lsn {}", key, lsn))?;
+
+            Ok(())
+        }
+
+        for (retain_lsn, KeyLogAtLsn(logs)) in &self.below_horizon {
+            for (lsn, val) in logs {
+                match val {
+                    Value::Image(img) => {
+                        base_img = Some((*lsn, img));
+                        history.clear();
+                    }
+                    Value::WalRecord(rec) if val.will_init() => {
+                        base_img = None;
+                        history.clear();
+                        history.push((*lsn, rec));
+                    }
+                    Value::WalRecord(rec) => {
+                        history.push((*lsn, rec));
+                    }
+                }
+            }
+            if *retain_lsn >= min_lsn {
+                // Only verify after the key appears in the full history for the first time.
+
+                // We don't modify history: in theory, we could replace the history with a single
+                // image as in `generate_key_retention` to make redos at later LSNs faster. But we
+                // want to verify everything as if they are read from the real layer map.
+                collect_and_verify(key, *retain_lsn, &base_img, &history, tline, false)
+                    .await
+                    .context("below horizon retain_lsn")?;
+            }
+        }
+
+        for (lsn, val) in &self.above_horizon.0 {
+            match val {
+                Value::Image(img) => {
+                    // Above the GC horizon, we verify every time we see an image.
+                    collect_and_verify(key, *lsn, &base_img, &history, tline, true)
+                        .await
+                        .context("above horizon full image")?;
+                    base_img = Some((*lsn, img));
+                    history.clear();
+                }
+                Value::WalRecord(rec) if val.will_init() => {
+                    // Above the GC horizon, we verify every time we see an init record.
+                    collect_and_verify(key, *lsn, &base_img, &history, tline, true)
+                        .await
+                        .context("above horizon init record")?;
+                    base_img = None;
+                    history.clear();
+                    history.push((*lsn, rec));
+                }
+                Value::WalRecord(rec) => {
+                    history.push((*lsn, rec));
+                }
+            }
+        }
+        // Ensure the latest record is readable.
+        collect_and_verify(key, max_lsn, &base_img, &history, tline, false)
+            .await
+            .context("latest record")?;
+        Ok(())
+    }
 }
 
 #[derive(Debug, Serialize, Default)]
@@ -1006,16 +1133,15 @@ impl Timeline {
 
         // 1. L0 Compact
         let l0_outcome = {
-            let timer = self.metrics.compact_time_histo.start_timer();
-            let l0_outcome = self
+            
+            self
                 .compact_level0(
                     target_file_size,
                     options.flags.contains(CompactFlags::ForceL0Compaction),
                     ctx,
                 )
-                .await?;
-            timer.stop_and_record();
-            l0_outcome
+                .await?
+        
         };
 
         if options.flags.contains(CompactFlags::OnlyL0Compaction) {
@@ -1112,14 +1238,23 @@ impl Timeline {
         let partition_count = self.partitioning.read().0.0.parts.len();
 
         // 4. Shard ancestor compaction
-
-        if self.shard_identity.count >= ShardCount::new(2) {
+        if self.get_compaction_shard_ancestor() && self.shard_identity.count >= ShardCount::new(2) {
             // Limit the number of layer rewrites to the number of partitions: this means its
             // runtime should be comparable to a full round of image layer creations, rather than
             // being potentially much longer.
             let rewrite_max = partition_count;
 
-            self.compact_shard_ancestors(rewrite_max, ctx).await?;
+            let outcome = self
+                .compact_shard_ancestors(
+                    rewrite_max,
+                    options.flags.contains(CompactFlags::YieldForL0),
+                    ctx,
+                )
+                .await?;
+            match outcome {
+                CompactionOutcome::Pending | CompactionOutcome::YieldForL0 => return Ok(outcome),
+                CompactionOutcome::Done | CompactionOutcome::Skipped => {}
+            }
         }
 
         Ok(CompactionOutcome::Done)
@@ -1136,8 +1271,10 @@ impl Timeline {
     async fn compact_shard_ancestors(
         self: &Arc<Self>,
         rewrite_max: usize,
+        yield_for_l0: bool,
         ctx: &RequestContext,
-    ) -> Result<(), CompactionError> {
+    ) -> Result<CompactionOutcome, CompactionError> {
+        let mut outcome = CompactionOutcome::Done;
         let mut drop_layers = Vec::new();
         let mut layers_to_rewrite: Vec<Layer> = Vec::new();
 
@@ -1148,15 +1285,13 @@ impl Timeline {
         // Holding this read guard also blocks [`Self::gc_timeline`] from entering while we
         // are rewriting layers.
         let latest_gc_cutoff = self.get_applied_gc_cutoff_lsn();
-
-        tracing::info!(
-            "starting shard ancestor compaction, latest_gc_cutoff: {}, pitr cutoff {}",
-            *latest_gc_cutoff,
-            self.gc_info.read().unwrap().cutoffs.time
-        );
+        let pitr_cutoff = self.gc_info.read().unwrap().cutoffs.time;
 
         let layers = self.layers.read().await;
-        for layer_desc in layers.layer_map()?.iter_historic_layers() {
+        let layers_iter = layers.layer_map()?.iter_historic_layers();
+        let (layers_total, mut layers_checked) = (layers_iter.len(), 0);
+        for layer_desc in layers_iter {
+            layers_checked += 1;
             let layer = layers.get_from_desc(&layer_desc);
             if layer.metadata().shard.shard_count == self.shard_identity.count {
                 // This layer does not belong to a historic ancestor, no need to re-image it.
@@ -1171,8 +1306,8 @@ impl Timeline {
                 // This ancestral layer only covers keys that belong to other shards.
                 // We include the full metadata in the log: if we had some critical bug that caused
                 // us to incorrectly drop layers, this would simplify manually debugging + reinstating those layers.
-                info!(%layer, old_metadata=?layer.metadata(),
-                    "dropping layer after shard split, contains no keys for this shard.",
+                debug!(%layer, old_metadata=?layer.metadata(),
+                    "dropping layer after shard split, contains no keys for this shard",
                 );
 
                 if cfg!(debug_assertions) {
@@ -1200,14 +1335,15 @@ impl Timeline {
                 continue;
             }
 
-            // Don't bother re-writing a layer unless it will at least halve its size
+            // Only rewrite a layer if we can reclaim significant space.
             if layer_local_page_count != u32::MAX
-                && layer_local_page_count > layer_raw_page_count / 2
+                && layer_local_page_count as f64 / layer_raw_page_count as f64
+                    <= ANCESTOR_COMPACTION_REWRITE_THRESHOLD
             {
                 debug!(%layer,
-                    "layer is already mostly local ({}/{}), not rewriting",
-                    layer_local_page_count,
-                    layer_raw_page_count
+                    "layer has a large share of local pages \
+                        ({layer_local_page_count}/{layer_raw_page_count} > \
+                        {ANCESTOR_COMPACTION_REWRITE_THRESHOLD}), not rewriting",
                 );
             }
 
@@ -1219,12 +1355,19 @@ impl Timeline {
                 continue;
             }
 
+            // We do not yet implement rewrite of delta layers.
             if layer_desc.is_delta() {
-                // We do not yet implement rewrite of delta layers
                 debug!(%layer, "Skipping rewrite of delta layer");
                 continue;
             }
 
+            // We don't bother rewriting layers that aren't visible, since these won't be needed by
+            // reads and will likely be garbage collected soon.
+            if layer.visibility() != LayerVisibilityHint::Visible {
+                debug!(%layer, "Skipping rewrite of invisible layer");
+                continue;
+            }
+
             // Only rewrite layers if their generations differ.  This guarantees:
             //  - that local rewrite is safe, as local layer paths will differ between existing layer and rewritten one
             //  - that the layer is persistent in remote storage, as we only see old-generation'd layer via loading from remote storage
@@ -1234,19 +1377,36 @@ impl Timeline {
             }
 
             if layers_to_rewrite.len() >= rewrite_max {
-                tracing::info!(%layer, "Will rewrite layer on a future compaction, already rewrote {}",
+                debug!(%layer, "Will rewrite layer on a future compaction, already rewrote {}",
                     layers_to_rewrite.len()
                 );
-                continue;
+                outcome = CompactionOutcome::Pending;
+                break;
             }
 
             // Fall through: all our conditions for doing a rewrite passed.
             layers_to_rewrite.push(layer);
         }
 
-        // Drop read lock on layer map before we start doing time-consuming I/O
+        // Drop read lock on layer map before we start doing time-consuming I/O.
         drop(layers);
 
+        // Drop out early if there's nothing to do.
+        if layers_to_rewrite.is_empty() && drop_layers.is_empty() {
+            return Ok(CompactionOutcome::Done);
+        }
+
+        info!(
+            "starting shard ancestor compaction, rewriting {} layers and dropping {} layers, \
+                checked {layers_checked}/{layers_total} layers \
+                (latest_gc_cutoff={} pitr_cutoff={})",
+            layers_to_rewrite.len(),
+            drop_layers.len(),
+            *latest_gc_cutoff,
+            pitr_cutoff,
+        );
+        let started = Instant::now();
+
         let mut replace_image_layers = Vec::new();
 
         for layer in layers_to_rewrite {
@@ -1254,13 +1414,15 @@ impl Timeline {
                 return Err(CompactionError::ShuttingDown);
             }
 
-            tracing::info!(layer=%layer, "Rewriting layer after shard split...");
+            info!(layer=%layer, "rewriting layer after shard split");
             let mut image_layer_writer = ImageLayerWriter::new(
                 self.conf,
                 self.timeline_id,
                 self.tenant_shard_id,
                 &layer.layer_desc().key_range,
                 layer.layer_desc().image_layer_lsn(),
+                &self.gate,
+                self.cancel.clone(),
                 ctx,
             )
             .await
@@ -1292,7 +1454,7 @@ impl Timeline {
                     .map_err(CompactionError::Other)?;
                 let new_layer = Layer::finish_creating(self.conf, self, desc, &path)
                     .map_err(CompactionError::Other)?;
-                tracing::info!(layer=%new_layer, "Rewrote layer, {} -> {} bytes",
+                info!(layer=%new_layer, "rewrote layer, {} -> {} bytes",
                     layer.metadata().file_size,
                     new_layer.metadata().file_size);
 
@@ -1302,6 +1464,26 @@ impl Timeline {
                 // the layer has no data for us with the ShardedRange check above, but
                 drop_layers.push(layer);
             }
+
+            // Yield for L0 compaction if necessary, but make sure we update the layer map below
+            // with the work we've already done.
+            if yield_for_l0
+                && self
+                    .l0_compaction_trigger
+                    .notified()
+                    .now_or_never()
+                    .is_some()
+            {
+                info!("shard ancestor compaction yielding for L0 compaction");
+                outcome = CompactionOutcome::YieldForL0;
+                break;
+            }
+        }
+
+        for layer in &drop_layers {
+            info!(%layer, old_metadata=?layer.metadata(),
+                "dropping layer after shard split (no keys for this shard)",
+            );
         }
 
         // At this point, we have replaced local layer files with their rewritten form, but not yet uploaded
@@ -1319,17 +1501,36 @@ impl Timeline {
         // necessary for correctness, but it simplifies testing, and avoids proceeding with another
         // Timeline's compaction while this timeline's uploads may be generating lots of disk I/O
         // load.
-        match self.remote_client.wait_completion().await {
-            Ok(()) => (),
-            Err(WaitCompletionError::NotInitialized(ni)) => return Err(CompactionError::from(ni)),
-            Err(WaitCompletionError::UploadQueueShutDownOrStopped) => {
-                return Err(CompactionError::ShuttingDown);
+        if outcome != CompactionOutcome::YieldForL0 {
+            info!("shard ancestor compaction waiting for uploads");
+            tokio::select! {
+                result = self.remote_client.wait_completion() => match result {
+                    Ok(()) => {},
+                    Err(WaitCompletionError::NotInitialized(ni)) => return Err(CompactionError::from(ni)),
+                    Err(WaitCompletionError::UploadQueueShutDownOrStopped) => {
+                        return Err(CompactionError::ShuttingDown);
+                    }
+                },
+                // Don't wait if there's L0 compaction to do. We don't need to update the outcome
+                // here, because we've already done the actual work.
+                _ = self.l0_compaction_trigger.notified(), if yield_for_l0 => {},
             }
         }
 
+        info!(
+            "shard ancestor compaction done in {:.3}s{}",
+            started.elapsed().as_secs_f64(),
+            match outcome {
+                CompactionOutcome::Pending =>
+                    format!(", with pending work (rewrite_max={rewrite_max})"),
+                CompactionOutcome::YieldForL0 => String::from(", yielding for L0 compaction"),
+                CompactionOutcome::Skipped | CompactionOutcome::Done => String::new(),
+            }
+        );
+
         fail::fail_point!("compact-shard-ancestors-persistent");
 
-        Ok(())
+        Ok(outcome)
     }
 
     /// Update the LayerVisibilityHint of layers covered by image layers, based on whether there is
@@ -1861,6 +2062,8 @@ impl Timeline {
                                 debug!("Create new layer {}..{}", lsn_range.start, lsn_range.end);
                                 lsn_range.clone()
                             },
+                            &self.gate,
+                            self.cancel.clone(),
                             ctx,
                         )
                         .await
@@ -2148,6 +2351,7 @@ impl Timeline {
     /// ```
     ///
     /// Note that `accumulated_values` must be sorted by LSN and should belong to a single key.
+    #[allow(clippy::too_many_arguments)]
     pub(crate) async fn generate_key_retention(
         self: &Arc<Timeline>,
         key: Key,
@@ -2156,6 +2360,7 @@ impl Timeline {
         retain_lsn_below_horizon: &[Lsn],
         delta_threshold_cnt: usize,
         base_img_from_ancestor: Option<(Key, Lsn, Bytes)>,
+        verification: bool,
     ) -> anyhow::Result<KeyHistoryRetention> {
         // Pre-checks for the invariants
 
@@ -2242,8 +2447,8 @@ impl Timeline {
             "should have at least below + above horizon batches"
         );
         let mut replay_history: Vec<(Key, Lsn, Value)> = Vec::new();
-        if let Some((key, lsn, img)) = base_img_from_ancestor {
-            replay_history.push((key, lsn, Value::Image(img)));
+        if let Some((key, lsn, ref img)) = base_img_from_ancestor {
+            replay_history.push((key, lsn, Value::Image(img.clone())));
         }
 
         /// Generate debug information for the replay history
@@ -2357,22 +2562,15 @@ impl Timeline {
             // Whether to reconstruct the image. In debug mode, we will generate an image
             // at every retain_lsn to ensure data is not corrupted, but we won't put the
             // image into the final layer.
-            let generate_image = produce_image || debug_mode;
-            if produce_image {
+            let img_and_lsn = if produce_image {
                 records_since_last_image = 0;
-            }
-            let img_and_lsn = if generate_image {
                 let replay_history_for_debug = if debug_mode {
                     Some(replay_history.clone())
                 } else {
                     None
                 };
                 let replay_history_for_debug_ref = replay_history_for_debug.as_deref();
-                let history = if produce_image {
-                    std::mem::take(&mut replay_history)
-                } else {
-                    replay_history.clone()
-                };
+                let history = std::mem::take(&mut replay_history);
                 let mut img = None;
                 let mut records = Vec::with_capacity(history.len());
                 if let (_, lsn, Value::Image(val)) = history.first().as_ref().unwrap() {
@@ -2407,6 +2605,7 @@ impl Timeline {
                         records.push((lsn, rec));
                     }
                 }
+                // WAL redo requires records in the reverse LSN order
                 records.reverse();
                 let state = ValueReconstructState { img, records };
                 // last batch does not generate image so i is always in range, unless we force generate
@@ -2439,10 +2638,16 @@ impl Timeline {
         assert_eq!(retention.len(), lsn_split_points.len() + 1);
         for (idx, logs) in retention.into_iter().enumerate() {
             if idx == lsn_split_points.len() {
-                return Ok(KeyHistoryRetention {
+                let retention = KeyHistoryRetention {
                     below_horizon: result,
                     above_horizon: KeyLogAtLsn(logs),
-                });
+                };
+                if verification {
+                    retention
+                        .verify(key, &base_img_from_ancestor, full_history, self)
+                        .await?;
+                }
+                return Ok(retention);
             } else {
                 result.push((lsn_split_points[idx], KeyLogAtLsn(logs)));
             }
@@ -2909,6 +3114,9 @@ impl Timeline {
             }
             (false, res)
         };
+
+        let verification = self.get_gc_compaction_settings().gc_compaction_verification;
+
         info!(
             "picked {} layers for compaction ({} layers need rewriting) with max_layer_lsn={} min_layer_lsn={} gc_cutoff={} lowest_retain_lsn={}, key_range={}..{}, has_data_below={}",
             job_desc.selected_layers.len(),
@@ -3055,6 +3263,8 @@ impl Timeline {
                     job_desc.compaction_key_range.start,
                     lowest_retain_lsn,
                     self.get_compaction_target_size(),
+                    &self.gate,
+                    self.cancel.clone(),
                     ctx,
                 )
                 .await
@@ -3071,6 +3281,8 @@ impl Timeline {
             self.tenant_shard_id,
             lowest_retain_lsn..end_lsn,
             self.get_compaction_target_size(),
+            &self.gate,
+            self.cancel.clone(),
         )
         .await
         .context("failed to create delta layer writer")
@@ -3167,6 +3379,8 @@ impl Timeline {
                                 self.tenant_shard_id,
                                 desc.key_range.start,
                                 desc.lsn_range.clone(),
+                                &self.gate,
+                                self.cancel.clone(),
                                 ctx,
                             )
                             .await
@@ -3184,6 +3398,8 @@ impl Timeline {
                                 self.tenant_shard_id,
                                 job_desc.compaction_key_range.end,
                                 desc.lsn_range.clone(),
+                                &self.gate,
+                                self.cancel.clone(),
                                 ctx,
                             )
                             .await
@@ -3225,6 +3441,7 @@ impl Timeline {
                             .await
                             .context("failed to get ancestor image")
                             .map_err(CompactionError::Other)?,
+                        verification,
                     )
                     .await
                     .context("failed to generate key retention")
@@ -3265,6 +3482,7 @@ impl Timeline {
                     .await
                     .context("failed to get ancestor image")
                     .map_err(CompactionError::Other)?,
+                verification,
             )
             .await
             .context("failed to generate key retention")
@@ -3753,6 +3971,8 @@ impl CompactionJobExecutor for TimelineAdaptor {
             self.timeline.tenant_shard_id,
             key_range.start,
             lsn_range.clone(),
+            &self.timeline.gate,
+            self.timeline.cancel.clone(),
             ctx,
         )
         .await?;
@@ -3820,7 +4040,7 @@ impl TimelineAdaptor {
         key_range: &Range<Key>,
         ctx: &RequestContext,
     ) -> Result<(), CreateImageLayersError> {
-        let timer = self.timeline.metrics.create_images_time_histo.start_timer();
+       
 
         let image_layer_writer = ImageLayerWriter::new(
             self.timeline.conf,
@@ -3828,6 +4048,8 @@ impl TimelineAdaptor {
             self.timeline.tenant_shard_id,
             key_range,
             lsn,
+            &self.timeline.gate,
+            self.timeline.cancel.clone(),
             ctx,
         )
         .await?;
@@ -3864,7 +4086,7 @@ impl TimelineAdaptor {
             self.new_images.push(image_layer);
         }
 
-        timer.stop_and_record();
+        
 
         Ok(())
     }

pageserver/src/tenant/timeline/detach_ancestor.rs

@@ -30,6 +30,7 @@ use crate::tenant::storage_layer::{
     AsLayerDesc as _, DeltaLayerWriter, ImageLayerWriter, IoConcurrency, Layer, ResidentLayer,
     ValuesReconstructState,
 };
+use crate::tenant::timeline::VersionedKeySpaceQuery;
 use crate::virtual_file::{MaybeFatalIo, VirtualFile};
 
 #[derive(Debug, thiserror::Error)]
@@ -212,13 +213,9 @@ async fn generate_tombstone_image_layer(
         }
     }
 
+    let query = VersionedKeySpaceQuery::uniform(KeySpace::single(key_range.clone()), image_lsn);
     let data = ancestor
-        .get_vectored_impl(
-            KeySpace::single(key_range.clone()),
-            image_lsn,
-            &mut reconstruct_state,
-            ctx,
-        )
+        .get_vectored_impl(query, &mut reconstruct_state, ctx)
         .await
         .context("failed to retrieve aux keys")
         .map_err(|e| Error::launder(e, Error::Prepare))?;
@@ -231,6 +228,8 @@ async fn generate_tombstone_image_layer(
             detached.tenant_shard_id,
             &key_range,
             image_lsn,
+            &detached.gate,
+            detached.cancel.clone(),
             ctx,
         )
         .await
@@ -779,6 +778,8 @@ async fn copy_lsn_prefix(
         target_timeline.tenant_shard_id,
         layer.layer_desc().key_range.start,
         layer.layer_desc().lsn_range.start..end_lsn,
+        &target_timeline.gate,
+        target_timeline.cancel.clone(),
         ctx,
     )
     .await

pageserver/src/tenant/timeline/eviction_task.rs

@@ -126,7 +126,7 @@ impl Timeline {
     ) -> ControlFlow<(), Instant> {
         debug!("eviction iteration: {policy:?}");
         let start = Instant::now();
-        let (period, threshold) = match policy {
+        let (period, _) = match policy {
             EvictionPolicy::NoEviction => {
                 // check again in 10 seconds; XXX config watch mechanism
                 return ControlFlow::Continue(Instant::now() + Duration::from_secs(10));
@@ -159,16 +159,6 @@ impl Timeline {
             period,
             BackgroundLoopKind::Eviction,
         );
-        // FIXME: if we were to mix policies on a pageserver, we would have no way to sense this. I
-        // don't think that is a relevant fear however, and regardless the imitation should be the
-        // most costly part.
-        crate::metrics::EVICTION_ITERATION_DURATION
-            .get_metric_with_label_values(&[
-                &format!("{}", period.as_secs()),
-                &format!("{}", threshold.as_secs()),
-            ])
-            .unwrap()
-            .observe(elapsed.as_secs_f64());
 
         ControlFlow::Continue(start + period)
     }
@@ -325,7 +315,7 @@ impl Timeline {
         &self,
         cancel: &CancellationToken,
         ctx: &RequestContext,
-    ) -> ControlFlow<(), BackgroundLoopSemaphorePermit<'static>> {
+    ) -> ControlFlow<(), BackgroundLoopSemaphorePermit> {
         let acquire_permit =
             crate::tenant::tasks::acquire_concurrency_permit(BackgroundLoopKind::Eviction, ctx);
 
@@ -367,7 +357,7 @@ impl Timeline {
         p: &EvictionPolicyLayerAccessThreshold,
         cancel: &CancellationToken,
         gate: &GateGuard,
-        permit: BackgroundLoopSemaphorePermit<'static>,
+        permit: BackgroundLoopSemaphorePermit,
         ctx: &RequestContext,
     ) -> ControlFlow<()> {
         if !self.tenant_shard_id.is_shard_zero() {

pageserver/src/tenant/timeline/heatmap_layers_downloader.rs

@@ -59,11 +59,7 @@ impl HeatmapLayersDownloader {
                     return;
                 };
 
-                tracing::info!(
-                    resident_size=%timeline.resident_physical_size(),
-                    heatmap_layers=%heatmap.all_layers().count(),
-                    "Starting heatmap layers download"
-                );
+                
 
                 let stream = futures::stream::iter(heatmap.all_layers().cloned().filter_map(
                     |layer| {
@@ -93,7 +89,7 @@ impl HeatmapLayersDownloader {
                 tokio::select! {
                     _ = stream.collect::<()>() => {
                         tracing::info!(
-                            resident_size=%timeline.resident_physical_size(),
+                            
                             "Heatmap layers download completed"
                         );
                     },

pageserver/src/tenant/timeline/import_pgdata/flow.rs

@@ -738,6 +738,8 @@ impl ChunkProcessingJob {
             self.timeline.tenant_shard_id,
             &self.range,
             self.pgdata_lsn,
+            &self.timeline.gate,
+            self.timeline.cancel.clone(),
             ctx,
         )
         .await?;

pageserver/src/tenant/timeline/layer_manager.rs

@@ -290,7 +290,7 @@ impl OpenLayerManager {
         lsn: Lsn,
         last_freeze_at: &AtomicLsn,
         write_lock: &mut tokio::sync::MutexGuard<'_, Option<TimelineWriterState>>,
-        metrics: &TimelineMetrics,
+        _metrics: &TimelineMetrics,
     ) -> bool {
         let Lsn(last_record_lsn) = lsn;
         let end_lsn = Lsn(last_record_lsn + 1);
@@ -299,10 +299,6 @@ impl OpenLayerManager {
             let open_layer_rc = Arc::clone(open_layer);
             open_layer.freeze(end_lsn).await;
 
-            // Increment the frozen layer metrics. This is decremented in `finish_flush_l0_layer()`.
-            // TODO: It would be nicer to do this via `InMemoryLayer::drop()`, but it requires a
-            // reference to the timeline metrics. Other methods use a metrics borrow as well.
-            metrics.inc_frozen_layer(open_layer);
 
             // The layer is no longer open, update the layer map to reflect this.
             // We will replace it with on-disk historics below.
@@ -334,16 +330,12 @@ impl OpenLayerManager {
     pub(crate) fn track_new_image_layers(
         &mut self,
         image_layers: &[ResidentLayer],
-        metrics: &TimelineMetrics,
+        _metrics: &TimelineMetrics,
     ) {
         let mut updates = self.layer_map.batch_update();
         for layer in image_layers {
             Self::insert_historic_layer(layer.as_ref().clone(), &mut updates, &mut self.layer_fmgr);
 
-            // record these here instead of Layer::finish_creating because otherwise partial
-            // failure with create_image_layers would balloon up the physical size gauge. downside
-            // is that all layers need to be created before metrics are updated.
-            metrics.record_new_file_metrics(layer.layer_desc().file_size);
         }
         updates.flush();
     }
@@ -353,14 +345,13 @@ impl OpenLayerManager {
         &mut self,
         delta_layer: Option<&ResidentLayer>,
         frozen_layer_for_check: &Arc<InMemoryLayer>,
-        metrics: &TimelineMetrics,
+        _metrics: &TimelineMetrics,
     ) {
         let inmem = self
             .layer_map
             .frozen_layers
             .pop_front()
             .expect("there must be a inmem layer to flush");
-        metrics.dec_frozen_layer(&inmem);
 
         // Only one task may call this function at a time (for this
         // timeline). If two tasks tried to flush the same frozen
@@ -370,7 +361,6 @@ impl OpenLayerManager {
         if let Some(l) = delta_layer {
             let mut updates = self.layer_map.batch_update();
             Self::insert_historic_layer(l.as_ref().clone(), &mut updates, &mut self.layer_fmgr);
-            metrics.record_new_file_metrics(l.layer_desc().file_size);
             updates.flush();
         }
     }
@@ -380,12 +370,11 @@ impl OpenLayerManager {
         &mut self,
         compact_from: &[Layer],
         compact_to: &[ResidentLayer],
-        metrics: &TimelineMetrics,
+        _metrics: &TimelineMetrics,
     ) {
         let mut updates = self.layer_map.batch_update();
         for l in compact_to {
             Self::insert_historic_layer(l.as_ref().clone(), &mut updates, &mut self.layer_fmgr);
-            metrics.record_new_file_metrics(l.layer_desc().file_size);
         }
         for l in compact_from {
             Self::delete_historic_layer(l, &mut updates, &mut self.layer_fmgr);
@@ -438,7 +427,7 @@ impl OpenLayerManager {
         rewrite_layers: &[(Layer, ResidentLayer)],
         drop_layers: &[Layer],
         add_layers: &[ResidentLayer],
-        metrics: &TimelineMetrics,
+        _metrics: &TimelineMetrics,
     ) {
         let mut updates = self.layer_map.batch_update();
         for (old_layer, new_layer) in rewrite_layers {
@@ -469,14 +458,12 @@ impl OpenLayerManager {
                 &mut self.layer_fmgr,
             );
 
-            metrics.record_new_file_metrics(new_layer.layer_desc().file_size);
         }
         for l in drop_layers {
             Self::delete_historic_layer(l, &mut updates, &mut self.layer_fmgr);
         }
         for l in add_layers {
             Self::insert_historic_layer(l.as_ref().clone(), &mut updates, &mut self.layer_fmgr);
-            metrics.record_new_file_metrics(l.layer_desc().file_size);
         }
         updates.flush();
     }

pageserver/src/tenant/timeline/logical_size.rs

@@ -23,7 +23,6 @@ pub(super) struct LogicalSize {
     /// the initial size at a different LSN.
     pub initial_logical_size: OnceCell<(
         u64,
-        crate::metrics::initial_logical_size::FinishedCalculationGuard,
     )>,
 
     /// Cancellation for the best-effort logical size calculation.
@@ -130,11 +129,7 @@ impl CurrentLogicalSize {
 impl LogicalSize {
     pub(super) fn empty_initial() -> Self {
         Self {
-            initial_logical_size: OnceCell::with_value((0, {
-                crate::metrics::initial_logical_size::START_CALCULATION
-                    .first(crate::metrics::initial_logical_size::StartCircumstances::EmptyInitial)
-                    .calculation_result_saved()
-            })),
+            initial_logical_size: OnceCell::with_value((0,)),
             cancel_wait_for_background_loop_concurrency_limit_semaphore: OnceCell::new(),
             initial_part_end: None,
             size_added_after_initial: AtomicI64::new(0),
@@ -159,7 +154,7 @@ impl LogicalSize {
         //                  ^^^ keep this type explicit so that the casts in this function break if
         //                  we change the type.
         match self.initial_logical_size.get() {
-            Some((initial_size, _)) => {
+            Some((initial_size, )) => {
                 CurrentLogicalSize::Exact(Exact(initial_size.checked_add_signed(size_increment)
                     .with_context(|| format!("Overflow during logical size calculation, initial_size: {initial_size}, size_increment: {size_increment}"))
                     .unwrap()))
@@ -181,7 +176,7 @@ impl LogicalSize {
     /// available for re-use. This doesn't contain the incremental part.
     pub(super) fn initialized_size(&self, lsn: Lsn) -> Option<u64> {
         match self.initial_part_end {
-            Some(v) if v == lsn => self.initial_logical_size.get().map(|(s, _)| *s),
+            Some(v) if v == lsn => self.initial_logical_size.get().map(|(s, )| *s),
             _ => None,
         }
     }

pageserver/src/tenant/timeline/walreceiver/connection_manager.rs

@@ -39,10 +39,6 @@ use utils::postgres_client::{
 use super::walreceiver_connection::{WalConnectionStatus, WalReceiverError};
 use super::{TaskEvent, TaskHandle, TaskStateUpdate, WalReceiverConf};
 use crate::context::{DownloadBehavior, RequestContext};
-use crate::metrics::{
-    WALRECEIVER_ACTIVE_MANAGERS, WALRECEIVER_BROKER_UPDATES, WALRECEIVER_CANDIDATES_ADDED,
-    WALRECEIVER_CANDIDATES_REMOVED, WALRECEIVER_SWITCHES,
-};
 use crate::task_mgr::TaskKind;
 use crate::tenant::{Timeline, debug_assert_current_span_has_tenant_and_timeline_id};
 
@@ -76,11 +72,6 @@ pub(super) async fn connection_manager_loop_step(
         }
     }
 
-    WALRECEIVER_ACTIVE_MANAGERS.inc();
-    scopeguard::defer! {
-        WALRECEIVER_ACTIVE_MANAGERS.dec();
-    }
-
     let id = TenantTimelineId {
         tenant_id: connection_manager_state.timeline.tenant_shard_id.tenant_id,
         timeline_id: connection_manager_state.timeline.timeline_id,
@@ -526,9 +517,6 @@ impl ConnectionManagerState {
 
     /// Shuts down the current connection (if any) and immediately starts another one with the given connection string.
     async fn change_connection(&mut self, new_sk: NewWalConnectionCandidate, ctx: &RequestContext) {
-        WALRECEIVER_SWITCHES
-            .with_label_values(&[new_sk.reason.name()])
-            .inc();
 
         self.drop_old_connection(true).await;
 
@@ -580,6 +568,7 @@ impl ConnectionManagerState {
                                 );
                                 Ok(())
                             }
+                            WalReceiverError::Cancelled => Ok(()),
                             WalReceiverError::Other(e) => {
                                 // give out an error to have task_mgr give it a really verbose logging
                                 if cancellation.is_cancelled() {
@@ -730,8 +719,6 @@ impl ConnectionManagerState {
             }
         };
 
-        WALRECEIVER_BROKER_UPDATES.inc();
-
         trace!(
             "safekeeper info update: standby_horizon(cutoff)={}",
             timeline_update.standby_horizon
@@ -741,10 +728,6 @@ impl ConnectionManagerState {
             self.timeline
                 .standby_horizon
                 .store(Lsn(timeline_update.standby_horizon));
-            self.timeline
-                .metrics
-                .standby_horizon_gauge
-                .set(timeline_update.standby_horizon as i64);
         }
 
         let new_safekeeper_id = NodeId(timeline_update.safekeeper_id);
@@ -762,7 +745,6 @@ impl ConnectionManagerState {
                 %new_safekeeper_id,
                 "New SK node was added",
             );
-            WALRECEIVER_CANDIDATES_ADDED.inc();
         }
     }
 
@@ -1050,7 +1032,6 @@ impl ConnectionManagerState {
                     "Safekeeper node {node_id} did not send events for over {lagging_wal_timeout:?}, not retrying the connections"
                 );
                 self.wal_connection_retries.remove(&node_id);
-                WALRECEIVER_CANDIDATES_REMOVED.inc();
             }
         }
     }
@@ -1077,6 +1058,7 @@ struct NewWalConnectionCandidate {
     safekeeper_id: NodeId,
     wal_source_connconf: PgConnectionConfig,
     availability_zone: Option<String>,
+    #[allow(dead_code)]
     reason: ReconnectReason,
 }
 
@@ -1105,18 +1087,6 @@ enum ReconnectReason {
     },
 }
 
-impl ReconnectReason {
-    fn name(&self) -> &str {
-        match self {
-            ReconnectReason::NoExistingConnection => "NoExistingConnection",
-            ReconnectReason::LaggingWal { .. } => "LaggingWal",
-            ReconnectReason::SwitchAvailabilityZone => "SwitchAvailabilityZone",
-            ReconnectReason::NoWalTimeout { .. } => "NoWalTimeout",
-            ReconnectReason::NoKeepAlives { .. } => "NoKeepAlives",
-        }
-    }
-}
-
 #[cfg(test)]
 mod tests {
     use pageserver_api::config::defaults::DEFAULT_WAL_RECEIVER_PROTOCOL;

pageserver/src/tenant/timeline/walreceiver/walreceiver_connection.rs

@@ -36,7 +36,6 @@ use wal_decoder::wire_format::FromWireFormat;
 
 use super::TaskStateUpdate;
 use crate::context::RequestContext;
-use crate::metrics::{LIVE_CONNECTIONS, WAL_INGEST, WALRECEIVER_STARTED_CONNECTIONS};
 use crate::pgdatadir_mapping::DatadirModification;
 use crate::task_mgr::{TaskKind, WALRECEIVER_RUNTIME};
 use crate::tenant::{
@@ -73,6 +72,7 @@ pub(super) enum WalReceiverError {
     /// Generic error
     Other(anyhow::Error),
     ClosedGate,
+    Cancelled,
 }
 
 impl From<tokio_postgres::Error> for WalReceiverError {
@@ -136,7 +136,7 @@ pub(super) async fn handle_walreceiver_connection(
         GateError::GateClosed => WalReceiverError::ClosedGate,
     })?;
 
-    WALRECEIVER_STARTED_CONNECTIONS.inc();
+  
 
     // Connect to the database in replication mode.
     info!("connecting to {wal_source_connconf:?}");
@@ -200,6 +200,9 @@ pub(super) async fn handle_walreceiver_connection(
                                 // with a similar error.
                             },
                             WalReceiverError::SuccessfulCompletion(_) => {}
+                            WalReceiverError::Cancelled => {
+                                debug!("Connection cancelled")
+                            }
                             WalReceiverError::ClosedGate => {
                                 // doesn't happen at runtime
                             }
@@ -219,10 +222,6 @@ pub(super) async fn handle_walreceiver_connection(
         .instrument(tracing::info_span!("poller")),
     );
 
-    let _guard = LIVE_CONNECTIONS
-        .with_label_values(&["wal_receiver"])
-        .guard();
-
     let identify = identify_system(&replication_client).await?;
     info!("{identify:?}");
 
@@ -273,7 +272,12 @@ pub(super) async fn handle_walreceiver_connection(
 
     let mut waldecoder = WalStreamDecoder::new(startpoint, timeline.pg_version);
 
-    let mut walingest = WalIngest::new(timeline.as_ref(), startpoint, &ctx).await?;
+    let mut walingest = WalIngest::new(timeline.as_ref(), startpoint, &ctx)
+        .await
+        .map_err(|e| match e.kind {
+            crate::walingest::WalIngestErrorKind::Cancelled => WalReceiverError::Cancelled,
+            _ => WalReceiverError::Other(e.into()),
+        })?;
 
     let shard = vec![*timeline.get_shard_identity()];
 
@@ -335,7 +339,7 @@ pub(super) async fn handle_walreceiver_connection(
 
         let status_update = match replication_message {
             ReplicationMessage::RawInterpretedWalRecords(raw) => {
-                WAL_INGEST.bytes_received.inc_by(raw.data().len() as u64);
+               
 
                 let mut uncommitted_records = 0;
 
@@ -408,21 +412,13 @@ pub(super) async fn handle_walreceiver_connection(
                     ctx: &RequestContext,
                     uncommitted: &mut u64,
                 ) -> anyhow::Result<()> {
-                    let stats = modification.stats();
+                  
                     modification.commit(ctx).await?;
-                    WAL_INGEST.records_committed.inc_by(*uncommitted);
-                    WAL_INGEST.inc_values_committed(&stats);
+                  
                     *uncommitted = 0;
                     Ok(())
                 }
 
-                if !records.is_empty() {
-                    timeline
-                        .metrics
-                        .wal_records_received
-                        .inc_by(records.len() as u64);
-                }
-
                 for interpreted in records {
                     if matches!(interpreted.flush_uncommitted, FlushUncommittedRecords::Yes)
                         && uncommitted_records > 0
@@ -432,9 +428,6 @@ pub(super) async fn handle_walreceiver_connection(
 
                     let local_next_record_lsn = interpreted.next_record_lsn;
 
-                    if interpreted.is_observed() {
-                        WAL_INGEST.records_observed.inc();
-                    }
 
                     walingest
                         .ingest_record(interpreted, &mut modification, &ctx)
@@ -506,12 +499,9 @@ pub(super) async fn handle_walreceiver_connection(
                     filtered: &mut u64,
                     ctx: &RequestContext,
                 ) -> anyhow::Result<()> {
-                    let stats = modification.stats();
+                   
                     modification.commit(ctx).await?;
-                    WAL_INGEST
-                        .records_committed
-                        .inc_by(*uncommitted - *filtered);
-                    WAL_INGEST.inc_values_committed(&stats);
+                   
                     *uncommitted = 0;
                     *filtered = 0;
                     Ok(())
@@ -525,7 +515,7 @@ pub(super) async fn handle_walreceiver_connection(
 
                 trace!("received XLogData between {startlsn} and {endlsn}");
 
-                WAL_INGEST.bytes_received.inc_by(data.len() as u64);
+            
                 waldecoder.feed_bytes(data);
 
                 {
@@ -567,7 +557,6 @@ pub(super) async fn handle_walreceiver_connection(
                         }
 
                         // Ingest the records without immediately committing them.
-                        timeline.metrics.wal_records_received.inc();
                         let ingested = walingest
                             .ingest_record(interpreted, &mut modification, &ctx)
                             .await
@@ -583,7 +572,7 @@ pub(super) async fn handle_walreceiver_connection(
                             })?;
                         if !ingested {
                             tracing::debug!("ingest: filtered out record @ LSN {next_record_lsn}");
-                            WAL_INGEST.records_filtered.inc();
+                          
                             filtered_records += 1;
                         }
 

pageserver/src/tenant/vectored_blob_io.rs

@@ -26,7 +26,7 @@ use utils::lsn::Lsn;
 use utils::vec_map::VecMap;
 
 use crate::context::RequestContext;
-use crate::tenant::blob_io::{BYTE_UNCOMPRESSED, BYTE_ZSTD, LEN_COMPRESSION_BIT_MASK};
+use crate::tenant::blob_io::{BYTE_UNCOMPRESSED, BYTE_ZSTD, Header};
 use crate::virtual_file::{self, IoBufferMut, VirtualFile};
 
 /// Metadata bundled with the start and end offset of a blob.
@@ -111,18 +111,20 @@ impl From<Bytes> for BufView<'_> {
 pub struct VectoredBlob {
     /// Blob metadata.
     pub meta: BlobMeta,
-    /// Start offset.
-    start: usize,
+    /// Header start offset.
+    header_start: usize,
+    /// Data start offset.
+    data_start: usize,
     /// End offset.
     end: usize,
-    /// Compression used on the the blob.
+    /// Compression used on the data, extracted from the header.
     compression_bits: u8,
 }
 
 impl VectoredBlob {
     /// Reads a decompressed view of the blob.
     pub(crate) async fn read<'a>(&self, buf: &BufView<'a>) -> Result<BufView<'a>, std::io::Error> {
-        let view = buf.view(self.start..self.end);
+        let view = buf.view(self.data_start..self.end);
 
         match self.compression_bits {
             BYTE_UNCOMPRESSED => Ok(view),
@@ -140,13 +142,19 @@ impl VectoredBlob {
                     std::io::ErrorKind::InvalidData,
                     format!(
                         "Failed to decompress blob for {}@{}, {}..{}: invalid compression byte {bits:x}",
-                        self.meta.key, self.meta.lsn, self.start, self.end
+                        self.meta.key, self.meta.lsn, self.data_start, self.end
                     ),
                 );
                 Err(error)
             }
         }
     }
+
+    /// Returns the raw blob including header.
+    #[allow(unused)]
+    pub(crate) fn raw_with_header<'a>(&self, buf: &BufView<'a>) -> BufView<'a> {
+        buf.view(self.header_start..self.end)
+    }
 }
 
 impl std::fmt::Display for VectoredBlob {
@@ -154,7 +162,7 @@ impl std::fmt::Display for VectoredBlob {
         write!(
             f,
             "{}@{}, {}..{}",
-            self.meta.key, self.meta.lsn, self.start, self.end
+            self.meta.key, self.meta.lsn, self.data_start, self.end
         )
     }
 }
@@ -493,50 +501,28 @@ impl<'a> VectoredBlobReader<'a> {
 
         let blobs_at = read.blobs_at.as_slice();
 
-        let start_offset = read.start;
-
-        let mut metas = Vec::with_capacity(blobs_at.len());
+        let mut blobs = Vec::with_capacity(blobs_at.len());
         // Blobs in `read` only provide their starting offset. The end offset
         // of a blob is implicit: the start of the next blob if one exists
         // or the end of the read.
 
-        for (blob_start, meta) in blobs_at {
-            let blob_start_in_buf = blob_start - start_offset;
-            let first_len_byte = buf[blob_start_in_buf as usize];
+        for (blob_start, meta) in blobs_at.iter().copied() {
+            let header_start = (blob_start - read.start) as usize;
+            let header = Header::decode(&buf[header_start..])?;
+            let data_start = header_start + header.header_len;
+            let end = data_start + header.data_len;
+            let compression_bits = header.compression_bits;
 
-            // Each blob is prefixed by a header containing its size and compression information.
-            // Extract the size and skip that header to find the start of the data.
-            // The size can be 1 or 4 bytes. The most significant bit is 0 in the
-            // 1 byte case and 1 in the 4 byte case.
-            let (size_length, blob_size, compression_bits) = if first_len_byte < 0x80 {
-                (1, first_len_byte as u64, BYTE_UNCOMPRESSED)
-            } else {
-                let mut blob_size_buf = [0u8; 4];
-                let offset_in_buf = blob_start_in_buf as usize;
-
-                blob_size_buf.copy_from_slice(&buf[offset_in_buf..offset_in_buf + 4]);
-                blob_size_buf[0] &= !LEN_COMPRESSION_BIT_MASK;
-
-                let compression_bits = first_len_byte & LEN_COMPRESSION_BIT_MASK;
-                (
-                    4,
-                    u32::from_be_bytes(blob_size_buf) as u64,
-                    compression_bits,
-                )
-            };
-
-            let start = (blob_start_in_buf + size_length) as usize;
-            let end = start + blob_size as usize;
-
-            metas.push(VectoredBlob {
-                start,
+            blobs.push(VectoredBlob {
+                header_start,
+                data_start,
                 end,
-                meta: *meta,
+                meta,
                 compression_bits,
             });
         }
 
-        Ok(VectoredBlobsBuf { buf, blobs: metas })
+        Ok(VectoredBlobsBuf { buf, blobs })
     }
 }
 
@@ -997,6 +983,15 @@ mod tests {
                 &read_buf[..],
                 "mismatch for idx={idx} at offset={offset}"
             );
+
+            // Check that raw_with_header returns a valid header.
+            let raw = read_blob.raw_with_header(&view);
+            let header = Header::decode(&raw)?;
+            if !compression || header.header_len == 1 {
+                assert_eq!(header.compression_bits, BYTE_UNCOMPRESSED);
+            }
+            assert_eq!(raw.len(), header.total_len());
+
             buf = result.buf;
         }
         Ok(())

pageserver/src/utilization.rs

@@ -10,7 +10,7 @@ use pageserver_api::models::PageserverUtilization;
 use utils::serde_percent::Percent;
 
 use crate::config::PageServerConf;
-use crate::metrics::NODE_UTILIZATION_SCORE;
+
 use crate::tenant::mgr::TenantManager;
 
 pub(crate) fn regenerate(
@@ -53,7 +53,7 @@ pub(crate) fn regenerate(
     // Express a static value for how many shards we may schedule on one node
     const MAX_SHARDS: u32 = 5000;
 
-    let mut doc = PageserverUtilization {
+    let doc = PageserverUtilization {
         disk_usage_bytes: used,
         free_space_bytes: free,
         disk_wanted_bytes,
@@ -63,10 +63,7 @@ pub(crate) fn regenerate(
         utilization_score: None,
         captured_at: utils::serde_system_time::SystemTime(captured_at),
     };
-
-    // Initialize `PageserverUtilization::utilization_score`
-    let score = doc.cached_score();
-    NODE_UTILIZATION_SCORE.set(score);
+    
 
     Ok(doc)
 }

pageserver/src/virtual_file.rs

@@ -27,12 +27,9 @@ use owned_buffers_io::io_buf_ext::FullSlice;
 use pageserver_api::config::defaults::DEFAULT_IO_BUFFER_ALIGNMENT;
 pub use pageserver_api::models::virtual_file as api;
 use tokio::sync::{RwLock, RwLockReadGuard, RwLockWriteGuard};
-use tokio::time::Instant;
 use tokio_epoll_uring::{BoundedBuf, IoBuf, IoBufMut, Slice};
 
-use crate::assert_u64_eq_usize::UsizeIsU64;
 use crate::context::RequestContext;
-use crate::metrics::{STORAGE_IO_TIME_METRIC, StorageIoOperation};
 use crate::page_cache::{PAGE_SZ, PageWriteGuard};
 pub(crate) mod io_engine;
 pub use io_engine::{
@@ -431,9 +428,7 @@ impl OpenFiles {
         if let Some(old_file) = slot_guard.file.take() {
             // the normal path of dropping VirtualFile uses "close", use "close-by-replace" here to
             // distinguish the two.
-            STORAGE_IO_TIME_METRIC
-                .get(StorageIoOperation::CloseByReplace)
-                .observe_closure_duration(|| drop(old_file));
+            drop(old_file);
         }
 
         // Prepare the slot for reuse and return it
@@ -532,13 +527,9 @@ impl<T> MaybeFatalIo<T> for std::io::Result<T> {
 /// where "support" means that we measure wall clock time.
 macro_rules! observe_duration {
     ($op:expr, $($body:tt)*) => {{
-        let instant = Instant::now();
-        let result = $($body)*;
-        let elapsed = instant.elapsed().as_secs_f64();
-        STORAGE_IO_TIME_METRIC
-            .get($op)
-            .observe(elapsed);
-        result
+        
+        $($body)*
+        
     }}
 }
 
@@ -913,7 +904,7 @@ impl VirtualFileInner {
         &self,
         buf: tokio_epoll_uring::Slice<Buf>,
         offset: u64,
-        ctx: &RequestContext,
+        _ctx: &RequestContext,
     ) -> (tokio_epoll_uring::Slice<Buf>, Result<usize, Error>)
     where
         Buf: tokio_epoll_uring::IoBufMut + Send,
@@ -930,9 +921,7 @@ impl VirtualFileInner {
         observe_duration!(StorageIoOperation::Read, {
             let ((_file_guard, buf), res) = io_engine::get().read_at(file_guard, offset, buf).await;
             let res = res.maybe_fatal_err("io_engine read_at inside VirtualFileInner::read_at");
-            if let Ok(size) = res {
-                ctx.io_size_metrics().read.add(size.into_u64());
-            }
+            
             (buf, res)
         })
     }
@@ -953,7 +942,7 @@ impl VirtualFileInner {
         &self,
         buf: FullSlice<B>,
         offset: u64,
-        ctx: &RequestContext,
+        _ctx: &RequestContext,
     ) -> (FullSlice<B>, Result<usize, Error>) {
         let file_guard = match self.lock_file().await {
             Ok(file_guard) => file_guard,
@@ -962,9 +951,7 @@ impl VirtualFileInner {
         observe_duration!(StorageIoOperation::Write, {
             let ((_file_guard, buf), result) =
                 io_engine::get().write_at(file_guard, offset, buf).await;
-            if let Ok(size) = result {
-                ctx.io_size_metrics().write.add(size.into_u64());
-            }
+
             (buf, result)
         })
     }
@@ -1263,9 +1250,7 @@ impl Drop for VirtualFileInner {
                 // there is also operation "close-by-replace" for closes done on eviction for
                 // comparison.
                 if let Some(fd) = slot_guard.file.take() {
-                    STORAGE_IO_TIME_METRIC
-                        .get(StorageIoOperation::Close)
-                        .observe_closure_duration(|| drop(fd));
+                    drop(fd);
                 }
             }
         }
@@ -1334,7 +1319,6 @@ pub fn init(num_slots: usize, engine: IoEngineKind, mode: IoMode, sync_mode: Syn
     set_io_mode(mode);
     io_engine::init(engine);
     SYNC_MODE.store(sync_mode as u8, std::sync::atomic::Ordering::Relaxed);
-    crate::metrics::virtual_file_descriptor_cache::SIZE_MAX.set(num_slots as u64);
 }
 
 const TEST_MAX_FILE_DESCRIPTORS: usize = 10;

pageserver/src/virtual_file/io_engine.rs

@@ -54,14 +54,7 @@ static IO_ENGINE: AtomicU8 = AtomicU8::new(IoEngine::NotSet as u8);
 pub(crate) fn set(engine_kind: IoEngineKind) {
     let engine: IoEngine = engine_kind.into();
     IO_ENGINE.store(engine as u8, std::sync::atomic::Ordering::Relaxed);
-    #[cfg(not(test))]
-    {
-        let metric = &crate::metrics::virtual_file_io_engine::KIND;
-        metric.reset();
-        metric
-            .with_label_values(&[&format!("{engine_kind}")])
-            .set(1);
-    }
+    
 }
 
 #[cfg(not(test))]