Merge pull request #9291 from neondatabase/rc/2024-10-07

Storage & Compute release 2024-10-07

.github/workflows/build_and_test.yml

@@ -341,7 +341,7 @@ jobs:
           PERF_TEST_RESULT_CONNSTR: "${{ secrets.PERF_TEST_RESULT_CONNSTR }}"
           TEST_RESULT_CONNSTR: "${{ secrets.REGRESS_TEST_RESULT_CONNSTR_NEW }}"
           PAGESERVER_VIRTUAL_FILE_IO_ENGINE: tokio-epoll-uring
-          SYNC_AFTER_EACH_TEST: true
+          SYNC_BETWEEN_TESTS: true
       # XXX: no coverage data handling here, since benchmarks are run on release builds,
       # while coverage is currently collected for the debug ones
 

.github/workflows/trigger-e2e-tests.yml

@@ -102,12 +102,17 @@ jobs:
           # Default set of platforms to run e2e tests on
           platforms='["docker", "k8s"]'
 
-          # If the PR changes vendor/, pgxn/ or libs/vm_monitor/ directories, or compute/Dockerfile.compute-node, add k8s-neonvm to the list of platforms.
+          # If a PR changes anything that affects computes, add k8s-neonvm to the list of platforms.
           # If the workflow run is not a pull request, add k8s-neonvm to the list.
           if [ "$GITHUB_EVENT_NAME" == "pull_request" ]; then
             for f in $(gh api "/repos/${GITHUB_REPOSITORY}/pulls/${PR_NUMBER}/files" --paginate --jq '.[].filename'); do
               case "$f" in
-                vendor/*|pgxn/*|libs/vm_monitor/*|compute/Dockerfile.compute-node)
+                # List of directories that contain code which affect compute images.
+                #
+                # This isn't exhaustive, just the paths that are most directly compute-related.
+                # For example, compute_ctl also depends on libs/utils, but we don't trigger
+                # an e2e run on that.
+                vendor/*|pgxn/*|compute_tools/*|libs/vm_monitor/*|compute/Dockerfile.compute-node)
                   platforms=$(echo "${platforms}" | jq --compact-output '. += ["k8s-neonvm"] | unique')
                   ;;
                 *)

Cargo.lock

@@ -673,8 +673,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "3b829e4e32b91e643de6eafe82b1d90675f5874230191a4ffbc1b336dec4d6bf"
 dependencies = [
  "async-trait",
- "axum-core",
- "base64 0.21.1",
+ "axum-core 0.3.4",
  "bitflags 1.3.2",
  "bytes",
  "futures-util",
@@ -689,16 +688,47 @@ dependencies = [
  "pin-project-lite",
  "rustversion",
  "serde",
+ "sync_wrapper 0.1.2",
+ "tower",
+ "tower-layer",
+ "tower-service",
+]
+
+[[package]]
+name = "axum"
+version = "0.7.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3a6c9af12842a67734c9a2e355436e5d03b22383ed60cf13cd0c18fbfe3dcbcf"
+dependencies = [
+ "async-trait",
+ "axum-core 0.4.5",
+ "base64 0.21.1",
+ "bytes",
+ "futures-util",
+ "http 1.1.0",
+ "http-body 1.0.0",
+ "http-body-util",
+ "hyper 1.4.1",
+ "hyper-util",
+ "itoa",
+ "matchit 0.7.0",
+ "memchr",
+ "mime",
+ "percent-encoding",
+ "pin-project-lite",
+ "rustversion",
+ "serde",
  "serde_json",
  "serde_path_to_error",
  "serde_urlencoded",
  "sha1",
- "sync_wrapper",
+ "sync_wrapper 1.0.1",
  "tokio",
  "tokio-tungstenite",
  "tower",
  "tower-layer",
  "tower-service",
+ "tracing",
 ]
 
 [[package]]
@@ -718,6 +748,27 @@ dependencies = [
  "tower-service",
 ]
 
+[[package]]
+name = "axum-core"
+version = "0.4.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "09f2bd6146b97ae3359fa0cc6d6b376d9539582c7b4220f041a33ec24c226199"
+dependencies = [
+ "async-trait",
+ "bytes",
+ "futures-util",
+ "http 1.1.0",
+ "http-body 1.0.0",
+ "http-body-util",
+ "mime",
+ "pin-project-lite",
+ "rustversion",
+ "sync_wrapper 1.0.1",
+ "tower-layer",
+ "tower-service",
+ "tracing",
+]
+
 [[package]]
 name = "azure_core"
 version = "0.19.0"
@@ -1214,6 +1265,7 @@ version = "0.1.0"
 dependencies = [
  "anyhow",
  "bytes",
+ "camino",
  "cfg-if",
  "chrono",
  "clap",
@@ -2039,7 +2091,7 @@ dependencies = [
  "futures-core",
  "futures-sink",
  "http-body-util",
- "hyper 1.2.0",
+ "hyper 1.4.1",
  "hyper-util",
  "pin-project",
  "rand 0.8.5",
@@ -2402,6 +2454,15 @@ dependencies = [
  "digest",
 ]
 
+[[package]]
+name = "home"
+version = "0.5.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e3d1354bf6b7235cb4a0576c2619fd4ed18183f689b12b006a0ee7329eeff9a5"
+dependencies = [
+ "windows-sys 0.52.0",
+]
+
 [[package]]
 name = "hostname"
 version = "0.4.0"
@@ -2458,9 +2519,9 @@ dependencies = [
 
 [[package]]
 name = "http-body-util"
-version = "0.1.0"
+version = "0.1.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "41cb79eb393015dadd30fc252023adb0b2400a0caee0fa2a077e6e21a551e840"
+checksum = "793429d76616a256bcb62c2a2ec2bed781c8307e797e2598c50010f2bee2544f"
 dependencies = [
  "bytes",
  "futures-util",
@@ -2543,9 +2604,9 @@ dependencies = [
 
 [[package]]
 name = "hyper"
-version = "1.2.0"
+version = "1.4.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "186548d73ac615b32a73aafe38fb4f56c0d340e110e5a200bcadbaf2e199263a"
+checksum = "50dfd22e0e76d0f662d429a5f80fcaf3855009297eab6a0a9f8543834744ba05"
 dependencies = [
  "bytes",
  "futures-channel",
@@ -2585,7 +2646,7 @@ checksum = "a0bea761b46ae2b24eb4aef630d8d1c398157b6fc29e6350ecf090a0b70c952c"
 dependencies = [
  "futures-util",
  "http 1.1.0",
- "hyper 1.2.0",
+ "hyper 1.4.1",
  "hyper-util",
  "rustls 0.22.4",
  "rustls-pki-types",
@@ -2608,16 +2669,16 @@ dependencies = [
 
 [[package]]
 name = "hyper-util"
-version = "0.1.3"
+version = "0.1.7"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ca38ef113da30126bbff9cd1705f9273e15d45498615d138b0c20279ac7a76aa"
+checksum = "cde7055719c54e36e95e8719f95883f22072a48ede39db7fc17a4e1d5281e9b9"
 dependencies = [
  "bytes",
  "futures-channel",
  "futures-util",
  "http 1.1.0",
  "http-body 1.0.0",
- "hyper 1.2.0",
+ "hyper 1.4.1",
  "pin-project-lite",
  "socket2",
  "tokio",
@@ -3423,7 +3484,7 @@ dependencies = [
  "opentelemetry",
  "opentelemetry_sdk",
  "prost 0.13.3",
- "tonic 0.12.2",
+ "tonic 0.12.3",
 ]
 
 [[package]]
@@ -4312,7 +4373,7 @@ dependencies = [
  "humantime",
  "humantime-serde",
  "hyper 0.14.30",
- "hyper 1.2.0",
+ "hyper 1.4.1",
  "hyper-util",
  "indexmap 2.0.1",
  "ipnet",
@@ -4678,7 +4739,7 @@ dependencies = [
  "scopeguard",
  "serde",
  "serde_json",
- "sync_wrapper",
+ "sync_wrapper 0.1.2",
  "test-context",
  "tokio",
  "tokio-stream",
@@ -4743,7 +4804,7 @@ dependencies = [
  "http 1.1.0",
  "http-body 1.0.0",
  "http-body-util",
- "hyper 1.2.0",
+ "hyper 1.4.1",
  "hyper-rustls 0.26.0",
  "hyper-util",
  "ipnet",
@@ -4759,7 +4820,7 @@ dependencies = [
  "serde",
  "serde_json",
  "serde_urlencoded",
- "sync_wrapper",
+ "sync_wrapper 0.1.2",
  "tokio",
  "tokio-rustls 0.25.0",
  "tokio-util",
@@ -4800,7 +4861,7 @@ dependencies = [
  "futures",
  "getrandom 0.2.11",
  "http 1.1.0",
- "hyper 1.2.0",
+ "hyper 1.4.1",
  "parking_lot 0.11.2",
  "reqwest 0.12.4",
  "reqwest-middleware",
@@ -5897,6 +5958,12 @@ dependencies = [
  "futures-core",
 ]
 
+[[package]]
+name = "sync_wrapper"
+version = "1.0.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a7065abeca94b6a8a577f9bd45aa0867a2238b74e8eb67cf10d492bc39351394"
+
 [[package]]
 name = "synstructure"
 version = "0.12.6"
@@ -6241,9 +6308,9 @@ dependencies = [
 
 [[package]]
 name = "tokio-stream"
-version = "0.1.14"
+version = "0.1.16"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "397c988d37662c7dda6d2208364a706264bf3d6138b11d436cbac0ad38832842"
+checksum = "4f4e6ce100d0eb49a2734f8c0812bcd324cf357d21810932c5df6b96ef2b86f1"
 dependencies = [
  "futures-core",
  "pin-project-lite",
@@ -6267,9 +6334,9 @@ dependencies = [
 
 [[package]]
 name = "tokio-tungstenite"
-version = "0.20.0"
+version = "0.21.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2b2dbec703c26b00d74844519606ef15d09a7d6857860f84ad223dec002ddea2"
+checksum = "c83b561d025642014097b66e6c1bb422783339e0909e4429cde4749d1990bc38"
 dependencies = [
  "futures-util",
  "log",
@@ -6336,7 +6403,7 @@ checksum = "3082666a3a6433f7f511c7192923fa1fe07c69332d3c6a2e6bb040b569199d5a"
 dependencies = [
  "async-stream",
  "async-trait",
- "axum",
+ "axum 0.6.20",
  "base64 0.21.1",
  "bytes",
  "futures-core",
@@ -6362,9 +6429,9 @@ dependencies = [
 
 [[package]]
 name = "tonic"
-version = "0.12.2"
+version = "0.12.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c6f6ba989e4b2c58ae83d862d3a3e27690b6e3ae630d0deb59f3697f32aa88ad"
+checksum = "877c5b330756d856ffcc4553ab34a5684481ade925ecc54bcd1bf02b1d0d4d52"
 dependencies = [
  "async-trait",
  "base64 0.22.1",
@@ -6562,14 +6629,14 @@ checksum = "e421abadd41a4225275504ea4d6566923418b7f05506fbc9c0fe86ba7396114b"
 
 [[package]]
 name = "tungstenite"
-version = "0.20.1"
+version = "0.21.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9e3dac10fd62eaf6617d3a904ae222845979aec67c615d1c842b4002c7666fb9"
+checksum = "9ef1a641ea34f399a848dea702823bbecfb4c486f911735368f1f137cb8257e1"
 dependencies = [
  "byteorder",
  "bytes",
  "data-encoding",
- "http 0.2.9",
+ "http 1.1.0",
  "httparse",
  "log",
  "rand 0.8.5",
@@ -6797,7 +6864,7 @@ name = "vm_monitor"
 version = "0.1.0"
 dependencies = [
  "anyhow",
- "axum",
+ "axum 0.7.5",
  "cgroups-rs",
  "clap",
  "futures",
@@ -7030,13 +7097,14 @@ dependencies = [
 
 [[package]]
 name = "which"
-version = "4.4.0"
+version = "4.4.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2441c784c52b289a054b7201fc93253e288f094e2f4be9058343127c4226a269"
+checksum = "87ba24419a2078cd2b0f2ede2691b6c66d8e47836da3b6db8265ebad47afbfc7"
 dependencies = [
  "either",
- "libc",
+ "home",
  "once_cell",
+ "rustix",
 ]
 
 [[package]]
@@ -7267,15 +7335,9 @@ version = "0.1.0"
 dependencies = [
  "ahash",
  "anyhow",
- "aws-config",
- "aws-runtime",
- "aws-sigv4",
- "aws-smithy-async",
- "aws-smithy-http",
- "aws-smithy-types",
- "axum",
  "base64 0.21.1",
  "base64ct",
+ "bitflags 2.4.1",
  "bytes",
  "camino",
  "cc",
@@ -7300,6 +7362,8 @@ dependencies = [
  "hex",
  "hmac",
  "hyper 0.14.30",
+ "hyper 1.4.1",
+ "hyper-util",
  "indexmap 1.9.3",
  "itertools 0.10.5",
  "itertools 0.12.1",
@@ -7332,7 +7396,7 @@ dependencies = [
  "subtle",
  "syn 1.0.109",
  "syn 2.0.52",
- "sync_wrapper",
+ "sync_wrapper 0.1.2",
  "tikv-jemalloc-sys",
  "time",
  "time-macros",
@@ -7344,7 +7408,6 @@ dependencies = [
  "tracing",
  "tracing-core",
  "url",
- "uuid",
  "zeroize",
  "zstd",
  "zstd-safe",

Cargo.toml

@@ -53,7 +53,7 @@ azure_storage_blobs = { version = "0.19", default-features = false, features = [
 flate2 = "1.0.26"
 async-stream = "0.3"
 async-trait = "0.1"
-aws-config = { version = "1.5", default-features = false, features=["rustls"] }
+aws-config = { version = "1.5", default-features = false, features=["rustls", "sso"] }
 aws-sdk-s3 = "1.52"
 aws-sdk-iam = "1.46.0"
 aws-smithy-async = { version = "1.2.1", default-features = false, features=["rt-tokio"] }
@@ -61,7 +61,7 @@ aws-smithy-types = "1.2"
 aws-credential-types = "1.2.0"
 aws-sigv4 = { version = "1.2", features = ["sign-http"] }
 aws-types = "1.3"
-axum = { version = "0.6.20", features = ["ws"] }
+axum = { version = "0.7.5", features = ["ws"] }
 base64 = "0.13.0"
 bincode = "1.3"
 bindgen = "0.70"
@@ -96,10 +96,13 @@ hmac = "0.12.1"
 hostname = "0.4"
 http = {version = "1.1.0", features = ["std"]}
 http-types = { version = "2", default-features = false }
+http-body-util = "0.1.2"
 humantime = "2.1"
 humantime-serde = "1.1.1"
-hyper = "0.14"
-tokio-tungstenite = "0.20.0"
+hyper0 = { package = "hyper", version = "0.14" }
+hyper = "1.4"
+hyper-util = "0.1"
+tokio-tungstenite = "0.21.0"
 indexmap = "2"
 indoc = "2"
 ipnet = "2.9.0"

README.md

@@ -58,7 +58,7 @@ curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
 1. Install XCode and dependencies
 ```
 xcode-select --install
-brew install protobuf openssl flex bison icu4c pkg-config
+brew install protobuf openssl flex bison icu4c pkg-config m4
 
 # add openssl to PATH, required for ed25519 keys generation in neon_local
 echo 'export PATH="$(brew --prefix openssl)/bin:$PATH"' >> ~/.zshrc

compute/Dockerfile.compute-node

@@ -27,8 +27,8 @@ RUN case $DEBIAN_FLAVOR in \
       ;; \
     esac && \
     apt update &&  \
-    apt install -y git autoconf automake libtool build-essential bison flex libreadline-dev \
-    zlib1g-dev libxml2-dev libcurl4-openssl-dev libossp-uuid-dev wget pkg-config libssl-dev \
+    apt install --no-install-recommends -y git autoconf automake libtool build-essential bison flex libreadline-dev \
+    zlib1g-dev libxml2-dev libcurl4-openssl-dev libossp-uuid-dev wget ca-certificates pkg-config libssl-dev \
     libicu-dev libxslt1-dev liblz4-dev libzstd-dev zstd \
     $VERSION_INSTALLS
 
@@ -104,7 +104,7 @@ FROM build-deps AS postgis-build
 ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 RUN apt update && \
-    apt install -y gdal-bin libboost-dev libboost-thread-dev libboost-filesystem-dev \
+    apt install --no-install-recommends -y gdal-bin libboost-dev libboost-thread-dev libboost-filesystem-dev \
     libboost-system-dev libboost-iostreams-dev libboost-program-options-dev libboost-timer-dev \
     libcgal-dev libgdal-dev libgmp-dev libmpfr-dev libopenscenegraph-dev libprotobuf-c-dev \
     protobuf-c-compiler xsltproc
@@ -182,7 +182,7 @@ RUN case "${PG_VERSION}" in "v17") \
     echo "v17 extensions are not supported yet. Quit" && exit 0;; \
     esac && \
     apt update && \
-    apt install -y ninja-build python3-dev libncurses5 binutils clang
+    apt install --no-install-recommends -y ninja-build python3-dev libncurses5 binutils clang
 
 RUN case "${PG_VERSION}" in "v17") \
     echo "v17 extensions are not supported yet. Quit" && exit 0;; \
@@ -587,7 +587,7 @@ RUN case "${PG_VERSION}" in "v17") \
     echo "v17 extensions are not supported yet. Quit" && exit 0;; \
     esac && \
     apt-get update && \
-    apt-get install -y \
+    apt-get install --no-install-recommends -y \
         libboost-iostreams1.74-dev \
         libboost-regex1.74-dev \
         libboost-serialization1.74-dev \
@@ -752,7 +752,7 @@ ARG PG_VERSION
 COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
 
 RUN apt-get update && \
-    apt-get install -y curl libclang-dev && \
+    apt-get install --no-install-recommends -y curl libclang-dev && \
     useradd -ms /bin/bash nonroot -b /home
 
 ENV HOME=/home/nonroot
@@ -880,9 +880,6 @@ RUN case "${PG_VERSION}" in "v17") \
     mkdir pg_session_jwt-src && cd pg_session_jwt-src && tar xzf ../pg_session_jwt.tar.gz --strip-components=1 -C . && \
     sed -i 's/pgrx = "=0.11.3"/pgrx = { version = "=0.11.3", features = [ "unsafe-postgres" ] }/g' Cargo.toml && \
     cargo pgrx install --release
-    # it's needed to enable extension because it uses untrusted C language
-    # sed -i 's/superuser = false/superuser = true/g' /usr/local/pgsql/share/extension/pg_session_jwt.control && \
-    # echo "trusted = true" >> /usr/local/pgsql/share/extension/pg_session_jwt.control
 
 #########################################################################################
 #
@@ -1058,9 +1055,12 @@ FROM debian:$DEBIAN_FLAVOR AS pgbouncer
 ARG DEBIAN_FLAVOR
 RUN set -e \
     && apt-get update \
-    && apt-get install -y \
+    && apt-get install --no-install-recommends -y \
         build-essential \
         git \
+        ca-certificates \
+        autoconf \
+        automake \
         libevent-dev \
         libtool \
         pkg-config
@@ -1075,6 +1075,20 @@ RUN set -e \
     && make -j $(nproc) dist_man_MANS= \
     && make install dist_man_MANS=
 
+#########################################################################################
+#
+# Compile the Neon-specific `local_proxy` binary
+#
+#########################################################################################
+FROM $REPOSITORY/$IMAGE:$TAG AS local_proxy
+ARG BUILD_TAG
+ENV BUILD_TAG=$BUILD_TAG
+
+USER nonroot
+# Copy entire project to get Cargo.* files with proper dependencies for the whole project
+COPY --chown=nonroot . .
+RUN mold -run cargo build --locked --profile release-line-debug-size-lto --bin local_proxy
+
 #########################################################################################
 #
 # Layers "postgres-exporter" and "sql-exporter"
@@ -1213,6 +1227,10 @@ COPY --from=compute-tools --chown=postgres /home/nonroot/target/release-line-deb
 COPY --from=pgbouncer         /usr/local/pgbouncer/bin/pgbouncer /usr/local/bin/pgbouncer
 COPY --chmod=0666 --chown=postgres compute/etc/pgbouncer.ini /etc/pgbouncer.ini
 
+# local_proxy and its config
+COPY --from=local_proxy --chown=postgres /home/nonroot/target/release-line-debug-size-lto/local_proxy /usr/local/bin/local_proxy
+RUN mkdir -p /etc/local_proxy && chown postgres:postgres /etc/local_proxy
+
 # Metrics exporter binaries and  configuration files
 COPY --from=postgres-exporter /bin/postgres_exporter /bin/postgres_exporter
 COPY --from=sql-exporter      /bin/sql_exporter      /bin/sql_exporter

compute/etc/neon_collector.yml

@@ -148,7 +148,7 @@ metrics:
   values: [pageserver_send_flushes_total]
   query_ref: neon_perf_counters
 
-- metric_name: getpage_wait_seconds_buckets
+- metric_name: getpage_wait_seconds_bucket
   type: counter
   help: 'Histogram buckets of getpage request latency'
   key_labels:

compute/vm-image-spec.yaml

@@ -19,6 +19,10 @@ commands:
     user: postgres
     sysvInitAction: respawn
     shell: '/usr/local/bin/pgbouncer /etc/pgbouncer.ini'
+  - name: local_proxy
+    user: postgres
+    sysvInitAction: respawn
+    shell: '/usr/local/bin/local_proxy --config-path /etc/local_proxy/config.json --pid-path /etc/local_proxy/pid --http 0.0.0.0:10432'
   - name: postgres-exporter
     user: nobody
     sysvInitAction: respawn

compute_tools/Cargo.toml

@@ -11,12 +11,13 @@ testing = []
 
 [dependencies]
 anyhow.workspace = true
+camino.workspace = true
 chrono.workspace = true
 cfg-if.workspace = true
 clap.workspace = true
 flate2.workspace = true
 futures.workspace = true
-hyper = { workspace = true, features = ["full"] }
+hyper0 = { workspace = true, features = ["full"] }
 nix.workspace = true
 notify.workspace = true
 num_cpus.workspace = true

compute_tools/src/compute.rs

@@ -34,6 +34,7 @@ use nix::sys::signal::{kill, Signal};
 use remote_storage::{DownloadError, RemotePath};
 
 use crate::checker::create_availability_check_data;
+use crate::local_proxy;
 use crate::logger::inlinify;
 use crate::pg_helpers::*;
 use crate::spec::*;
@@ -886,6 +887,11 @@ impl ComputeNode {
         // 'Close' connection
         drop(client);
 
+        if let Some(ref local_proxy) = spec.local_proxy_config {
+            info!("configuring local_proxy");
+            local_proxy::configure(local_proxy).context("apply_config local_proxy")?;
+        }
+
         // Run migrations separately to not hold up cold starts
         thread::spawn(move || {
             let mut connstr = connstr.clone();
@@ -936,6 +942,19 @@ impl ComputeNode {
             });
         }
 
+        if let Some(ref local_proxy) = spec.local_proxy_config {
+            info!("configuring local_proxy");
+
+            // Spawn a thread to do the configuration,
+            // so that we don't block the main thread that starts Postgres.
+            let local_proxy = local_proxy.clone();
+            let _handle = Some(thread::spawn(move || {
+                if let Err(err) = local_proxy::configure(&local_proxy) {
+                    error!("error while configuring local_proxy: {err:?}");
+                }
+            }));
+        }
+
         // Write new config
         let pgdata_path = Path::new(&self.pgdata);
         let postgresql_conf_path = pgdata_path.join("postgresql.conf");
@@ -1023,6 +1042,19 @@ impl ComputeNode {
             });
         }
 
+        if let Some(local_proxy) = &pspec.spec.local_proxy_config {
+            info!("configuring local_proxy");
+
+            // Spawn a thread to do the configuration,
+            // so that we don't block the main thread that starts Postgres.
+            let local_proxy = local_proxy.clone();
+            let _handle = thread::spawn(move || {
+                if let Err(err) = local_proxy::configure(&local_proxy) {
+                    error!("error while configuring local_proxy: {err:?}");
+                }
+            });
+        }
+
         info!(
             "start_compute spec.remote_extensions {:?}",
             pspec.spec.remote_extensions

compute_tools/src/lib.rs

@@ -2,6 +2,9 @@
 //! configuration.
 #![deny(unsafe_code)]
 #![deny(clippy::undocumented_unsafe_blocks)]
+
+extern crate hyper0 as hyper;
+
 pub mod checker;
 pub mod config;
 pub mod configurator;
@@ -12,6 +15,7 @@ pub mod catalog;
 pub mod compute;
 pub mod disk_quota;
 pub mod extension_server;
+pub mod local_proxy;
 pub mod lsn_lease;
 mod migration;
 pub mod monitor;

compute_tools/src/local_proxy.rs

@@ -0,0 +1,56 @@
+//! Local Proxy is a feature of our BaaS Neon Authorize project.
+//!
+//! Local Proxy validates JWTs and manages the pg_session_jwt extension.
+//! It also maintains a connection pool to postgres.
+
+use anyhow::{Context, Result};
+use camino::Utf8Path;
+use compute_api::spec::LocalProxySpec;
+use nix::sys::signal::Signal;
+use utils::pid_file::{self, PidFileRead};
+
+pub fn configure(local_proxy: &LocalProxySpec) -> Result<()> {
+    write_local_proxy_conf("/etc/local_proxy/config.json".as_ref(), local_proxy)?;
+    notify_local_proxy("/etc/local_proxy/pid".as_ref())?;
+
+    Ok(())
+}
+
+/// Create or completely rewrite configuration file specified by `path`
+fn write_local_proxy_conf(path: &Utf8Path, local_proxy: &LocalProxySpec) -> Result<()> {
+    let config =
+        serde_json::to_string_pretty(local_proxy).context("serializing LocalProxySpec to json")?;
+    std::fs::write(path, config).with_context(|| format!("writing {path}"))?;
+
+    Ok(())
+}
+
+/// Notify local proxy about a new config file.
+fn notify_local_proxy(path: &Utf8Path) -> Result<()> {
+    match pid_file::read(path)? {
+        // if the file doesn't exist, or isn't locked, local_proxy isn't running
+        // and will naturally pick up our config later
+        PidFileRead::NotExist | PidFileRead::NotHeldByAnyProcess(_) => {}
+        PidFileRead::LockedByOtherProcess(pid) => {
+            // From the pid_file docs:
+            //
+            // > 1. The other process might exit at any time, turning the given PID stale.
+            // > 2. There is a small window in which `claim_for_current_process` has already
+            // >    locked the file but not yet updates its contents. [`read`] will return
+            // >    this variant here, but with the old file contents, i.e., a stale PID.
+            // >
+            // > The kernel is free to recycle PID once it has been `wait(2)`ed upon by
+            // > its creator. Thus, acting upon a stale PID, e.g., by issuing a `kill`
+            // > system call on it, bears the risk of killing an unrelated process.
+            // > This is an inherent limitation of using pidfiles.
+            // > The only race-free solution is to have a supervisor-process with a lifetime
+            // > that exceeds that of all of its child-processes (e.g., `runit`, `supervisord`).
+            //
+            // This is an ok risk as we only send a SIGHUP which likely won't actually
+            // kill the process, only reload config.
+            nix::sys::signal::kill(pid, Signal::SIGHUP).context("sending signal to local_proxy")?;
+        }
+    }
+
+    Ok(())
+}

control_plane/Cargo.toml

@@ -14,7 +14,7 @@ humantime.workspace = true
 nix.workspace = true
 once_cell.workspace = true
 humantime-serde.workspace = true
-hyper.workspace = true
+hyper0.workspace = true
 regex.workspace = true
 reqwest = { workspace = true, features = ["blocking", "json"] }
 scopeguard.workspace = true

control_plane/src/endpoint.rs

@@ -599,6 +599,7 @@ impl Endpoint {
             remote_extensions,
             pgbouncer_settings: None,
             shard_stripe_size: Some(shard_stripe_size),
+            local_proxy_config: None,
         };
         let spec_path = self.endpoint_path().join("spec.json");
         std::fs::write(spec_path, serde_json::to_string_pretty(&spec)?)?;

control_plane/src/local_env.rs

@@ -168,6 +168,9 @@ pub struct NeonStorageControllerConf {
 
     #[serde(with = "humantime_serde")]
     pub heartbeat_interval: Duration,
+
+    #[serde(with = "humantime_serde")]
+    pub long_reconcile_threshold: Option<Duration>,
 }
 
 impl NeonStorageControllerConf {
@@ -190,6 +193,7 @@ impl Default for NeonStorageControllerConf {
             split_threshold: None,
             max_secondary_lag_bytes: None,
             heartbeat_interval: Self::DEFAULT_HEARTBEAT_INTERVAL,
+            long_reconcile_threshold: None,
         }
     }
 }

control_plane/src/storage_controller.rs

@@ -3,7 +3,7 @@ use crate::{
     local_env::{LocalEnv, NeonStorageControllerConf},
 };
 use camino::{Utf8Path, Utf8PathBuf};
-use hyper::Uri;
+use hyper0::Uri;
 use nix::unistd::Pid;
 use pageserver_api::{
     controller_api::{
@@ -517,6 +517,13 @@ impl StorageController {
             args.push(format!("--max-secondary-lag-bytes={lag}"))
         }
 
+        if let Some(threshold) = self.config.long_reconcile_threshold {
+            args.push(format!(
+                "--long-reconcile-threshold={}",
+                humantime::Duration::from(threshold)
+            ))
+        }
+
         args.push(format!(
             "--neon-local-repo-dir={}",
             self.env.base_data_dir.display()

deny.toml

@@ -27,10 +27,6 @@ yanked = "warn"
 id = "RUSTSEC-2023-0071"
 reason = "the marvin attack only affects private key decryption, not public key signature verification"
 
-[[advisories.ignore]]
-id = "RUSTSEC-2024-0376"
-reason = "gRPC endpoints in Neon are not exposed externally"
-
 # This section is considered when running `cargo deny check licenses`
 # More documentation for the licenses section can be found here:
 # https://embarkstudios.github.io/cargo-deny/checks/licenses/cfg.html

docs/docker.md

@@ -5,7 +5,7 @@
 Currently we build two main images:
 
 - [neondatabase/neon](https://hub.docker.com/repository/docker/neondatabase/neon) — image with pre-built `pageserver`, `safekeeper` and `proxy` binaries and all the required runtime dependencies. Built from [/Dockerfile](/Dockerfile).
-- [neondatabase/compute-node-v16](https://hub.docker.com/repository/docker/neondatabase/compute-node-v16) — compute node image with pre-built Postgres binaries from [neondatabase/postgres](https://github.com/neondatabase/postgres). Similar images exist for v15 and v14.
+- [neondatabase/compute-node-v16](https://hub.docker.com/repository/docker/neondatabase/compute-node-v16) — compute node image with pre-built Postgres binaries from [neondatabase/postgres](https://github.com/neondatabase/postgres). Similar images exist for v15 and v14. Built from [/compute-node/Dockerfile](/compute/Dockerfile.compute-node).
 
 And additional intermediate image:
 

libs/compute_api/src/spec.rs

@@ -106,6 +106,10 @@ pub struct ComputeSpec {
     // Stripe size for pageserver sharding, in pages
     #[serde(default)]
     pub shard_stripe_size: Option<usize>,
+
+    /// Local Proxy configuration used for JWT authentication
+    #[serde(default)]
+    pub local_proxy_config: Option<LocalProxySpec>,
 }
 
 /// Feature flag to signal `compute_ctl` to enable certain experimental functionality.
@@ -278,11 +282,13 @@ pub struct GenericOption {
 /// declare a `trait` on it.
 pub type GenericOptions = Option<Vec<GenericOption>>;
 
-/// Configured the local-proxy application with the relevant JWKS and roles it should
+/// Configured the local_proxy application with the relevant JWKS and roles it should
 /// use for authorizing connect requests using JWT.
 #[derive(Clone, Debug, Deserialize, Serialize)]
 pub struct LocalProxySpec {
-    pub jwks: Vec<JwksSettings>,
+    #[serde(default)]
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub jwks: Option<Vec<JwksSettings>>,
 }
 
 #[derive(Clone, Debug, Deserialize, Serialize)]

libs/pageserver_api/src/config.rs

@@ -296,7 +296,14 @@ pub mod defaults {
 
     pub const DEFAULT_INGEST_BATCH_SIZE: u64 = 100;
 
-    pub const DEFAULT_MAX_VECTORED_READ_BYTES: usize = 128 * 1024; // 128 KiB
+    /// Soft limit for the maximum size of a vectored read.
+    ///
+    /// This is determined by the largest NeonWalRecord that can exist (minus dbdir and reldir keys
+    /// which are bounded by the blob io limits only). As of this writing, that is a `NeonWalRecord::ClogSetCommitted` record,
+    /// with 32k xids. That's the max number of XIDS on a single CLOG page. The size of such a record
+    /// is `sizeof(Transactionid) * 32768 + (some fixed overhead from 'timestamp`, the Vec length and whatever extra serde serialization adds)`.
+    /// That is, slightly above 128 kB.
+    pub const DEFAULT_MAX_VECTORED_READ_BYTES: usize = 130 * 1024; // 130 KiB
 
     pub const DEFAULT_IMAGE_COMPRESSION: ImageCompressionAlgorithm =
         ImageCompressionAlgorithm::Zstd { level: Some(1) };

libs/pageserver_api/src/key.rs

@@ -748,6 +748,16 @@ impl Key {
         self.field1 == 0x00 && self.field4 != 0 && self.field6 != 0xffffffff
     }
 
+    #[inline(always)]
+    pub fn is_rel_dir_key(&self) -> bool {
+        self.field1 == 0x00
+            && self.field2 != 0
+            && self.field3 != 0
+            && self.field4 == 0
+            && self.field5 == 0
+            && self.field6 == 1
+    }
+
     /// 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)> {

libs/remote_storage/Cargo.toml

@@ -16,7 +16,7 @@ aws-sdk-s3.workspace = true
 bytes.workspace = true
 camino = { workspace = true, features = ["serde1"] }
 humantime-serde.workspace = true
-hyper = { workspace = true, features = ["stream"] }
+hyper0 = { workspace = true, features = ["stream"] }
 futures.workspace = true
 serde.workspace = true
 serde_json.workspace = true

libs/remote_storage/src/azure_blob.rs

@@ -14,7 +14,7 @@ use std::time::SystemTime;
 
 use super::REMOTE_STORAGE_PREFIX_SEPARATOR;
 use anyhow::Result;
-use azure_core::request_options::{MaxResults, Metadata, Range};
+use azure_core::request_options::{IfMatchCondition, MaxResults, Metadata, Range};
 use azure_core::{Continuable, RetryOptions};
 use azure_identity::DefaultAzureCredential;
 use azure_storage::StorageCredentials;
@@ -33,10 +33,10 @@ use tracing::debug;
 use utils::backoff;
 
 use crate::metrics::{start_measuring_requests, AttemptOutcome, RequestKind};
-use crate::ListingObject;
 use crate::{
-    config::AzureConfig, error::Cancelled, ConcurrencyLimiter, Download, DownloadError, Listing,
-    ListingMode, RemotePath, RemoteStorage, StorageMetadata, TimeTravelError, TimeoutOrCancel,
+    config::AzureConfig, error::Cancelled, ConcurrencyLimiter, Download, DownloadError,
+    DownloadOpts, Listing, ListingMode, ListingObject, RemotePath, RemoteStorage, StorageMetadata,
+    TimeTravelError, TimeoutOrCancel,
 };
 
 pub struct AzureBlobStorage {
@@ -259,6 +259,7 @@ fn to_download_error(error: azure_core::Error) -> DownloadError {
     if let Some(http_err) = error.as_http_error() {
         match http_err.status() {
             StatusCode::NotFound => DownloadError::NotFound,
+            StatusCode::NotModified => DownloadError::Unmodified,
             StatusCode::BadRequest => DownloadError::BadInput(anyhow::Error::new(error)),
             _ => DownloadError::Other(anyhow::Error::new(error)),
         }
@@ -484,11 +485,16 @@ impl RemoteStorage for AzureBlobStorage {
     async fn download(
         &self,
         from: &RemotePath,
+        opts: &DownloadOpts,
         cancel: &CancellationToken,
     ) -> Result<Download, DownloadError> {
         let blob_client = self.client.blob_client(self.relative_path_to_name(from));
 
-        let builder = blob_client.get();
+        let mut builder = blob_client.get();
+
+        if let Some(ref etag) = opts.etag {
+            builder = builder.if_match(IfMatchCondition::NotMatch(etag.to_string()))
+        }
 
         self.download_for_builder(builder, cancel).await
     }

libs/remote_storage/src/error.rs

@@ -5,6 +5,8 @@ pub enum DownloadError {
     BadInput(anyhow::Error),
     /// The file was not found in the remote storage.
     NotFound,
+    /// The caller provided an ETag, and the file was not modified.
+    Unmodified,
     /// A cancellation token aborted the download, typically during
     /// tenant detach or process shutdown.
     Cancelled,
@@ -24,6 +26,7 @@ impl std::fmt::Display for DownloadError {
                 write!(f, "Failed to download a remote file due to user input: {e}")
             }
             DownloadError::NotFound => write!(f, "No file found for the remote object id given"),
+            DownloadError::Unmodified => write!(f, "File was not modified"),
             DownloadError::Cancelled => write!(f, "Cancelled, shutting down"),
             DownloadError::Timeout => write!(f, "timeout"),
             DownloadError::Other(e) => write!(f, "Failed to download a remote file: {e:?}"),
@@ -38,7 +41,7 @@ impl DownloadError {
     pub fn is_permanent(&self) -> bool {
         use DownloadError::*;
         match self {
-            BadInput(_) | NotFound | Cancelled => true,
+            BadInput(_) | NotFound | Unmodified | Cancelled => true,
             Timeout | Other(_) => false,
         }
     }

libs/remote_storage/src/lib.rs

@@ -161,6 +161,14 @@ pub struct Listing {
     pub keys: Vec<ListingObject>,
 }
 
+/// Options for downloads. The default value is a plain GET.
+#[derive(Default)]
+pub struct DownloadOpts {
+    /// If given, returns [`DownloadError::Unmodified`] if the object still has
+    /// the same ETag (using If-None-Match).
+    pub etag: Option<Etag>,
+}
+
 /// Storage (potentially remote) API to manage its state.
 /// This storage tries to be unaware of any layered repository context,
 /// providing basic CRUD operations for storage files.
@@ -245,6 +253,7 @@ pub trait RemoteStorage: Send + Sync + 'static {
     async fn download(
         &self,
         from: &RemotePath,
+        opts: &DownloadOpts,
         cancel: &CancellationToken,
     ) -> Result<Download, DownloadError>;
 
@@ -401,16 +410,18 @@ impl<Other: RemoteStorage> GenericRemoteStorage<Arc<Other>> {
         }
     }
 
+    /// See [`RemoteStorage::download`]
     pub async fn download(
         &self,
         from: &RemotePath,
+        opts: &DownloadOpts,
         cancel: &CancellationToken,
     ) -> Result<Download, DownloadError> {
         match self {
-            Self::LocalFs(s) => s.download(from, cancel).await,
-            Self::AwsS3(s) => s.download(from, cancel).await,
-            Self::AzureBlob(s) => s.download(from, cancel).await,
-            Self::Unreliable(s) => s.download(from, cancel).await,
+            Self::LocalFs(s) => s.download(from, opts, cancel).await,
+            Self::AwsS3(s) => s.download(from, opts, cancel).await,
+            Self::AzureBlob(s) => s.download(from, opts, cancel).await,
+            Self::Unreliable(s) => s.download(from, opts, cancel).await,
         }
     }
 
@@ -572,7 +583,7 @@ impl GenericRemoteStorage {
     ) -> Result<Download, DownloadError> {
         match byte_range {
             Some((start, end)) => self.download_byte_range(from, start, end, cancel).await,
-            None => self.download(from, cancel).await,
+            None => self.download(from, &DownloadOpts::default(), cancel).await,
         }
     }
 

libs/remote_storage/src/local_fs.rs

@@ -23,8 +23,8 @@ use tokio_util::{io::ReaderStream, sync::CancellationToken};
 use utils::crashsafe::path_with_suffix_extension;
 
 use crate::{
-    Download, DownloadError, Listing, ListingMode, ListingObject, RemotePath, TimeTravelError,
-    TimeoutOrCancel, REMOTE_STORAGE_PREFIX_SEPARATOR,
+    Download, DownloadError, DownloadOpts, Listing, ListingMode, ListingObject, RemotePath,
+    TimeTravelError, TimeoutOrCancel, REMOTE_STORAGE_PREFIX_SEPARATOR,
 };
 
 use super::{RemoteStorage, StorageMetadata};
@@ -494,11 +494,17 @@ impl RemoteStorage for LocalFs {
     async fn download(
         &self,
         from: &RemotePath,
+        opts: &DownloadOpts,
         cancel: &CancellationToken,
     ) -> Result<Download, DownloadError> {
         let target_path = from.with_base(&self.storage_root);
 
         let file_metadata = file_metadata(&target_path).await?;
+        let etag = mock_etag(&file_metadata);
+
+        if opts.etag.as_ref() == Some(&etag) {
+            return Err(DownloadError::Unmodified);
+        }
 
         let source = ReaderStream::new(
             fs::OpenOptions::new()
@@ -519,7 +525,6 @@ impl RemoteStorage for LocalFs {
         let cancel_or_timeout = crate::support::cancel_or_timeout(self.timeout, cancel.clone());
         let source = crate::support::DownloadStream::new(cancel_or_timeout, source);
 
-        let etag = mock_etag(&file_metadata);
         Ok(Download {
             metadata,
             last_modified: file_metadata
@@ -692,7 +697,7 @@ mod fs_tests {
     ) -> anyhow::Result<String> {
         let cancel = CancellationToken::new();
         let download = storage
-            .download(remote_storage_path, &cancel)
+            .download(remote_storage_path, &DownloadOpts::default(), &cancel)
             .await
             .map_err(|e| anyhow::anyhow!("Download failed: {e}"))?;
         ensure!(
@@ -773,8 +778,8 @@ mod fs_tests {
             "We should upload and download the same contents"
         );
 
-        let non_existing_path = "somewhere/else";
-        match storage.download(&RemotePath::new(Utf8Path::new(non_existing_path))?, &cancel).await {
+        let non_existing_path = RemotePath::new(Utf8Path::new("somewhere/else"))?;
+        match storage.download(&non_existing_path, &DownloadOpts::default(), &cancel).await {
             Err(DownloadError::NotFound) => {} // Should get NotFound for non existing keys
             other => panic!("Should get a NotFound error when downloading non-existing storage files, but got: {other:?}"),
         }
@@ -1101,7 +1106,13 @@ mod fs_tests {
             storage.upload(body, len, &path, None, &cancel).await?;
         }
 
-        let read = aggregate(storage.download(&path, &cancel).await?.download_stream).await?;
+        let read = aggregate(
+            storage
+                .download(&path, &DownloadOpts::default(), &cancel)
+                .await?
+                .download_stream,
+        )
+        .await?;
         assert_eq!(body, read);
 
         let shorter = Bytes::from_static(b"shorter body");
@@ -1112,7 +1123,13 @@ mod fs_tests {
             storage.upload(body, len, &path, None, &cancel).await?;
         }
 
-        let read = aggregate(storage.download(&path, &cancel).await?.download_stream).await?;
+        let read = aggregate(
+            storage
+                .download(&path, &DownloadOpts::default(), &cancel)
+                .await?
+                .download_stream,
+        )
+        .await?;
         assert_eq!(shorter, read);
         Ok(())
     }
@@ -1145,7 +1162,13 @@ mod fs_tests {
             storage.upload(body, len, &path, None, &cancel).await?;
         }
 
-        let read = aggregate(storage.download(&path, &cancel).await?.download_stream).await?;
+        let read = aggregate(
+            storage
+                .download(&path, &DownloadOpts::default(), &cancel)
+                .await?
+                .download_stream,
+        )
+        .await?;
         assert_eq!(body, read);
 
         Ok(())

libs/remote_storage/src/s3_bucket.rs

@@ -28,12 +28,13 @@ use aws_sdk_s3::{
     Client,
 };
 use aws_smithy_async::rt::sleep::TokioSleep;
+use http_types::StatusCode;
 
 use aws_smithy_types::{body::SdkBody, DateTime};
 use aws_smithy_types::{byte_stream::ByteStream, date_time::ConversionError};
 use bytes::Bytes;
 use futures::stream::Stream;
-use hyper::Body;
+use hyper0::Body;
 use scopeguard::ScopeGuard;
 use tokio_util::sync::CancellationToken;
 use utils::backoff;
@@ -44,8 +45,8 @@ use crate::{
     error::Cancelled,
     metrics::{start_counting_cancelled_wait, start_measuring_requests},
     support::PermitCarrying,
-    ConcurrencyLimiter, Download, DownloadError, Listing, ListingMode, ListingObject, RemotePath,
-    RemoteStorage, TimeTravelError, TimeoutOrCancel, MAX_KEYS_PER_DELETE,
+    ConcurrencyLimiter, Download, DownloadError, DownloadOpts, Listing, ListingMode, ListingObject,
+    RemotePath, RemoteStorage, TimeTravelError, TimeoutOrCancel, MAX_KEYS_PER_DELETE,
     REMOTE_STORAGE_PREFIX_SEPARATOR,
 };
 
@@ -67,6 +68,7 @@ pub struct S3Bucket {
 struct GetObjectRequest {
     bucket: String,
     key: String,
+    etag: Option<String>,
     range: Option<String>,
 }
 impl S3Bucket {
@@ -248,13 +250,18 @@ impl S3Bucket {
 
         let started_at = start_measuring_requests(kind);
 
-        let get_object = self
+        let mut builder = self
             .client
             .get_object()
             .bucket(request.bucket)
             .key(request.key)
-            .set_range(request.range)
-            .send();
+            .set_range(request.range);
+
+        if let Some(etag) = request.etag {
+            builder = builder.if_none_match(etag);
+        }
+
+        let get_object = builder.send();
 
         let get_object = tokio::select! {
             res = get_object => res,
@@ -277,6 +284,20 @@ impl S3Bucket {
                 );
                 return Err(DownloadError::NotFound);
             }
+            Err(SdkError::ServiceError(e))
+                // aws_smithy_runtime_api::http::response::StatusCode isn't
+                // re-exported by any aws crates, so just check the numeric
+                // status against http_types::StatusCode instead of pulling it.
+                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,
@@ -773,6 +794,7 @@ impl RemoteStorage for S3Bucket {
     async fn download(
         &self,
         from: &RemotePath,
+        opts: &DownloadOpts,
         cancel: &CancellationToken,
     ) -> Result<Download, DownloadError> {
         // if prefix is not none then download file `prefix/from`
@@ -781,6 +803,7 @@ impl RemoteStorage for S3Bucket {
             GetObjectRequest {
                 bucket: self.bucket_name.clone(),
                 key: self.relative_path_to_s3_object(from),
+                etag: opts.etag.as_ref().map(|e| e.to_string()),
                 range: None,
             },
             cancel,
@@ -807,6 +830,7 @@ impl RemoteStorage for S3Bucket {
             GetObjectRequest {
                 bucket: self.bucket_name.clone(),
                 key: self.relative_path_to_s3_object(from),
+                etag: None,
                 range,
             },
             cancel,

libs/remote_storage/src/simulate_failures.rs

@@ -12,8 +12,8 @@ use std::{collections::hash_map::Entry, sync::Arc};
 use tokio_util::sync::CancellationToken;
 
 use crate::{
-    Download, DownloadError, GenericRemoteStorage, Listing, ListingMode, RemotePath, RemoteStorage,
-    StorageMetadata, TimeTravelError,
+    Download, DownloadError, DownloadOpts, GenericRemoteStorage, Listing, ListingMode, RemotePath,
+    RemoteStorage, StorageMetadata, TimeTravelError,
 };
 
 pub struct UnreliableWrapper {
@@ -167,11 +167,12 @@ impl RemoteStorage for UnreliableWrapper {
     async fn download(
         &self,
         from: &RemotePath,
+        opts: &DownloadOpts,
         cancel: &CancellationToken,
     ) -> Result<Download, DownloadError> {
         self.attempt(RemoteOp::Download(from.clone()))
             .map_err(DownloadError::Other)?;
-        self.inner.download(from, cancel).await
+        self.inner.download(from, opts, cancel).await
     }
 
     async fn download_byte_range(

libs/remote_storage/tests/common/tests.rs

@@ -1,8 +1,7 @@
 use anyhow::Context;
 use camino::Utf8Path;
 use futures::StreamExt;
-use remote_storage::ListingMode;
-use remote_storage::RemotePath;
+use remote_storage::{DownloadError, DownloadOpts, ListingMode, ListingObject, RemotePath};
 use std::sync::Arc;
 use std::{collections::HashSet, num::NonZeroU32};
 use test_context::test_context;
@@ -284,7 +283,10 @@ async fn upload_download_works(ctx: &mut MaybeEnabledStorage) -> anyhow::Result<
     ctx.client.upload(data, len, &path, None, &cancel).await?;
 
     // Normal download request
-    let dl = ctx.client.download(&path, &cancel).await?;
+    let dl = ctx
+        .client
+        .download(&path, &DownloadOpts::default(), &cancel)
+        .await?;
     let buf = download_to_vec(dl).await?;
     assert_eq!(&buf, &orig);
 
@@ -337,6 +339,54 @@ async fn upload_download_works(ctx: &mut MaybeEnabledStorage) -> anyhow::Result<
     Ok(())
 }
 
+/// Tests that conditional downloads work properly, by returning
+/// DownloadError::Unmodified when the object ETag matches the given ETag.
+#[test_context(MaybeEnabledStorage)]
+#[tokio::test]
+async fn download_conditional(ctx: &mut MaybeEnabledStorage) -> anyhow::Result<()> {
+    let MaybeEnabledStorage::Enabled(ctx) = ctx else {
+        return Ok(());
+    };
+    let cancel = CancellationToken::new();
+
+    // Create a file.
+    let path = RemotePath::new(Utf8Path::new(format!("{}/file", ctx.base_prefix).as_str()))?;
+    let data = bytes::Bytes::from_static("foo".as_bytes());
+    let (stream, len) = wrap_stream(data);
+    ctx.client.upload(stream, len, &path, None, &cancel).await?;
+
+    // Download it to obtain its etag.
+    let mut opts = DownloadOpts::default();
+    let download = ctx.client.download(&path, &opts, &cancel).await?;
+
+    // Download with the etag yields DownloadError::Unmodified.
+    opts.etag = Some(download.etag);
+    let result = ctx.client.download(&path, &opts, &cancel).await;
+    assert!(
+        matches!(result, Err(DownloadError::Unmodified)),
+        "expected DownloadError::Unmodified, got {result:?}"
+    );
+
+    // Replace the file contents.
+    let data = bytes::Bytes::from_static("bar".as_bytes());
+    let (stream, len) = wrap_stream(data);
+    ctx.client.upload(stream, len, &path, None, &cancel).await?;
+
+    // A download with the old etag should yield the new file.
+    let download = ctx.client.download(&path, &opts, &cancel).await?;
+    assert_ne!(download.etag, opts.etag.unwrap(), "ETag did not change");
+
+    // A download with the new etag should yield Unmodified again.
+    opts.etag = Some(download.etag);
+    let result = ctx.client.download(&path, &opts, &cancel).await;
+    assert!(
+        matches!(result, Err(DownloadError::Unmodified)),
+        "expected DownloadError::Unmodified, got {result:?}"
+    );
+
+    Ok(())
+}
+
 #[test_context(MaybeEnabledStorage)]
 #[tokio::test]
 async fn copy_works(ctx: &mut MaybeEnabledStorage) -> anyhow::Result<()> {
@@ -364,7 +414,10 @@ async fn copy_works(ctx: &mut MaybeEnabledStorage) -> anyhow::Result<()> {
     // Normal download request
     ctx.client.copy_object(&path, &path_dest, &cancel).await?;
 
-    let dl = ctx.client.download(&path_dest, &cancel).await?;
+    let dl = ctx
+        .client
+        .download(&path_dest, &DownloadOpts::default(), &cancel)
+        .await?;
     let buf = download_to_vec(dl).await?;
     assert_eq!(&buf, &orig);
 
@@ -376,3 +429,56 @@ async fn copy_works(ctx: &mut MaybeEnabledStorage) -> anyhow::Result<()> {
 
     Ok(())
 }
+
+/// Tests that head_object works properly.
+#[test_context(MaybeEnabledStorage)]
+#[tokio::test]
+async fn head_object(ctx: &mut MaybeEnabledStorage) -> anyhow::Result<()> {
+    let MaybeEnabledStorage::Enabled(ctx) = ctx else {
+        return Ok(());
+    };
+    let cancel = CancellationToken::new();
+
+    let path = RemotePath::new(Utf8Path::new(format!("{}/file", ctx.base_prefix).as_str()))?;
+
+    // Errors on missing file.
+    let result = ctx.client.head_object(&path, &cancel).await;
+    assert!(
+        matches!(result, Err(DownloadError::NotFound)),
+        "expected NotFound, got {result:?}"
+    );
+
+    // Create the file.
+    let data = bytes::Bytes::from_static("foo".as_bytes());
+    let (stream, len) = wrap_stream(data);
+    ctx.client.upload(stream, len, &path, None, &cancel).await?;
+
+    // Fetch the head metadata.
+    let object = ctx.client.head_object(&path, &cancel).await?;
+    assert_eq!(
+        object,
+        ListingObject {
+            key: path.clone(),
+            last_modified: object.last_modified, // ignore
+            size: 3
+        }
+    );
+
+    // Wait for a couple of seconds, and then update the file to check the last
+    // modified timestamp.
+    tokio::time::sleep(std::time::Duration::from_secs(2)).await;
+
+    let data = bytes::Bytes::from_static("bar".as_bytes());
+    let (stream, len) = wrap_stream(data);
+    ctx.client.upload(stream, len, &path, None, &cancel).await?;
+    let new = ctx.client.head_object(&path, &cancel).await?;
+
+    assert!(
+        !new.last_modified
+            .duration_since(object.last_modified)?
+            .is_zero(),
+        "last_modified did not advance"
+    );
+
+    Ok(())
+}

libs/remote_storage/tests/test_real_s3.rs

@@ -12,8 +12,8 @@ use anyhow::Context;
 use camino::Utf8Path;
 use futures_util::StreamExt;
 use remote_storage::{
-    DownloadError, GenericRemoteStorage, ListingMode, RemotePath, RemoteStorageConfig,
-    RemoteStorageKind, S3Config,
+    DownloadError, DownloadOpts, GenericRemoteStorage, ListingMode, RemotePath,
+    RemoteStorageConfig, RemoteStorageKind, S3Config,
 };
 use test_context::test_context;
 use test_context::AsyncTestContext;
@@ -121,7 +121,8 @@ async fn s3_time_travel_recovery_works(ctx: &mut MaybeEnabledStorage) -> anyhow:
 
     // A little check to ensure that our clock is not too far off from the S3 clock
     {
-        let dl = retry(|| ctx.client.download(&path2, &cancel)).await?;
+        let opts = DownloadOpts::default();
+        let dl = retry(|| ctx.client.download(&path2, &opts, &cancel)).await?;
         let last_modified = dl.last_modified;
         let half_wt = WAIT_TIME.mul_f32(0.5);
         let t0_hwt = t0 + half_wt;
@@ -159,7 +160,12 @@ async fn s3_time_travel_recovery_works(ctx: &mut MaybeEnabledStorage) -> anyhow:
     let t2_files_recovered = list_files(&ctx.client, &cancel).await?;
     println!("after recovery to t2: {t2_files_recovered:?}");
     assert_eq!(t2_files, t2_files_recovered);
-    let path2_recovered_t2 = download_to_vec(ctx.client.download(&path2, &cancel).await?).await?;
+    let path2_recovered_t2 = download_to_vec(
+        ctx.client
+            .download(&path2, &DownloadOpts::default(), &cancel)
+            .await?,
+    )
+    .await?;
     assert_eq!(path2_recovered_t2, new_data.as_bytes());
 
     // after recovery to t1: path1 is back, path2 has the old content
@@ -170,7 +176,12 @@ async fn s3_time_travel_recovery_works(ctx: &mut MaybeEnabledStorage) -> anyhow:
     let t1_files_recovered = list_files(&ctx.client, &cancel).await?;
     println!("after recovery to t1: {t1_files_recovered:?}");
     assert_eq!(t1_files, t1_files_recovered);
-    let path2_recovered_t1 = download_to_vec(ctx.client.download(&path2, &cancel).await?).await?;
+    let path2_recovered_t1 = download_to_vec(
+        ctx.client
+            .download(&path2, &DownloadOpts::default(), &cancel)
+            .await?,
+    )
+    .await?;
     assert_eq!(path2_recovered_t1, old_data.as_bytes());
 
     // after recovery to t0: everything is gone except for path1
@@ -416,7 +427,7 @@ async fn download_is_timeouted(ctx: &mut MaybeEnabledStorage) {
     let started_at = std::time::Instant::now();
     let mut stream = ctx
         .client
-        .download(&path, &cancel)
+        .download(&path, &DownloadOpts::default(), &cancel)
         .await
         .expect("download succeeds")
         .download_stream;
@@ -491,7 +502,7 @@ async fn download_is_cancelled(ctx: &mut MaybeEnabledStorage) {
     {
         let stream = ctx
             .client
-            .download(&path, &cancel)
+            .download(&path, &DownloadOpts::default(), &cancel)
             .await
             .expect("download succeeds")
             .download_stream;

libs/tracing-utils/Cargo.toml

@@ -5,7 +5,7 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-hyper.workspace = true
+hyper0.workspace = true
 opentelemetry = { workspace = true, features = ["trace"] }
 opentelemetry_sdk = { workspace = true, features = ["rt-tokio"] }
 opentelemetry-otlp = { workspace = true, default-features = false, features = ["http-proto", "trace", "http", "reqwest-client"] }

libs/tracing-utils/src/http.rs

@@ -1,7 +1,7 @@
 //! Tracing wrapper for Hyper HTTP server
 
-use hyper::HeaderMap;
-use hyper::{Body, Request, Response};
+use hyper0::HeaderMap;
+use hyper0::{Body, Request, Response};
 use std::future::Future;
 use tracing::Instrument;
 use tracing_opentelemetry::OpenTelemetrySpanExt;

libs/utils/Cargo.toml

@@ -22,7 +22,7 @@ chrono.workspace = true
 git-version.workspace = true
 hex = { workspace = true, features = ["serde"] }
 humantime.workspace = true
-hyper = { workspace = true, features = ["full"] }
+hyper0 = { workspace = true, features = ["full"] }
 fail.workspace = true
 futures = { workspace = true}
 jsonwebtoken.workspace = true

libs/utils/src/lib.rs

@@ -2,6 +2,8 @@
 //! between other crates in this repository.
 #![deny(clippy::undocumented_unsafe_blocks)]
 
+extern crate hyper0 as hyper;
+
 pub mod backoff;
 
 /// `Lsn` type implements common tasks on Log Sequence Numbers

libs/vm_monitor/src/lib.rs

@@ -7,11 +7,13 @@ use axum::{
     extract::{ws::WebSocket, State, WebSocketUpgrade},
     response::Response,
 };
-use axum::{routing::get, Router, Server};
+use axum::{routing::get, Router};
 use clap::Parser;
 use futures::Future;
+use std::net::SocketAddr;
 use std::{fmt::Debug, time::Duration};
 use sysinfo::{RefreshKind, System, SystemExt};
+use tokio::net::TcpListener;
 use tokio::{sync::broadcast, task::JoinHandle};
 use tokio_util::sync::CancellationToken;
 use tracing::{error, info};
@@ -132,14 +134,14 @@ pub async fn start(args: &'static Args, token: CancellationToken) -> anyhow::Res
             args,
         });
 
-    let addr = args.addr();
-    let bound = Server::try_bind(&addr.parse().expect("parsing address should not fail"))
+    let addr_str = args.addr();
+    let addr: SocketAddr = addr_str.parse().expect("parsing address should not fail");
+
+    let listener = TcpListener::bind(&addr)
+        .await
         .with_context(|| format!("failed to bind to {addr}"))?;
-
-    info!(addr, "server bound");
-
-    bound
-        .serve(app.into_make_service())
+    info!(addr_str, "server bound");
+    axum::serve(listener, app.into_make_service())
         .await
         .context("server exited")?;
 

pageserver/Cargo.toml

@@ -30,7 +30,7 @@ futures.workspace = true
 hex.workspace = true
 humantime.workspace = true
 humantime-serde.workspace = true
-hyper.workspace = true
+hyper0.workspace = true
 itertools.workspace = true
 md5.workspace = true
 nix.workspace = true

pageserver/src/bin/pageserver.rs

@@ -575,7 +575,7 @@ fn start_pageserver(
             .build()
             .map_err(|err| anyhow!(err))?;
         let service = utils::http::RouterService::new(router).unwrap();
-        let server = hyper::Server::from_tcp(http_listener)?
+        let server = hyper0::Server::from_tcp(http_listener)?
             .serve(service)
             .with_graceful_shutdown({
                 let cancel = cancel.clone();

pageserver/src/http/routes.rs

@@ -1742,6 +1742,10 @@ async fn timeline_compact_handler(
     let state = get_state(&request);
 
     let mut flags = EnumSet::empty();
+
+    if Some(true) == parse_query_param::<_, bool>(&request, "force_l0_compaction")? {
+        flags |= CompactFlags::ForceL0Compaction;
+    }
     if Some(true) == parse_query_param::<_, bool>(&request, "force_repartition")? {
         flags |= CompactFlags::ForceRepartition;
     }
@@ -1788,6 +1792,9 @@ async fn timeline_checkpoint_handler(
     let state = get_state(&request);
 
     let mut flags = EnumSet::empty();
+    if Some(true) == parse_query_param::<_, bool>(&request, "force_l0_compaction")? {
+        flags |= CompactFlags::ForceL0Compaction;
+    }
     if Some(true) == parse_query_param::<_, bool>(&request, "force_repartition")? {
         flags |= CompactFlags::ForceRepartition;
     }

pageserver/src/lib.rs

@@ -13,6 +13,8 @@ pub mod http;
 pub mod import_datadir;
 pub mod l0_flush;
 
+extern crate hyper0 as hyper;
+
 use futures::{stream::FuturesUnordered, StreamExt};
 pub use pageserver_api::keyspace;
 use tokio_util::sync::CancellationToken;

pageserver/src/tenant.rs

@@ -97,6 +97,7 @@ use crate::tenant::remote_timeline_client::MaybeDeletedIndexPart;
 use crate::tenant::remote_timeline_client::INITDB_PATH;
 use crate::tenant::storage_layer::DeltaLayer;
 use crate::tenant::storage_layer::ImageLayer;
+use crate::walingest::WalLagCooldown;
 use crate::walredo;
 use crate::InitializationOrder;
 use std::collections::hash_map::Entry;
@@ -319,6 +320,9 @@ pub struct Tenant {
     /// background warmup.
     pub(crate) activate_now_sem: tokio::sync::Semaphore,
 
+    /// Time it took for the tenant to activate. Zero if not active yet.
+    attach_wal_lag_cooldown: Arc<std::sync::OnceLock<WalLagCooldown>>,
+
     // Cancellation token fires when we have entered shutdown().  This is a parent of
     // Timelines' cancellation token.
     pub(crate) cancel: CancellationToken,
@@ -1000,11 +1004,15 @@ impl Tenant {
                 // Remote preload is complete.
                 drop(remote_load_completion);
 
+
                 // We will time the duration of the attach phase unless this is a creation (attach will do no work)
+                let attach_start = std::time::Instant::now();
                 let attached = {
                     let _attach_timer = Some(TENANT.attach.start_timer());
                     tenant_clone.attach(preload, &ctx).await
                 };
+                let attach_duration = attach_start.elapsed();
+                _ = tenant_clone.attach_wal_lag_cooldown.set(WalLagCooldown::new(attach_start, attach_duration));
 
                 match attached {
                     Ok(()) => {
@@ -2754,6 +2762,7 @@ impl Tenant {
             pg_version,
             state,
             last_aux_file_policy,
+            self.attach_wal_lag_cooldown.clone(),
             self.cancel.child_token(),
         );
 
@@ -2860,6 +2869,7 @@ impl Tenant {
                 Some(Duration::from_secs(3600 * 24)),
             )),
             activate_now_sem: tokio::sync::Semaphore::new(0),
+            attach_wal_lag_cooldown: Arc::new(std::sync::OnceLock::new()),
             cancel: CancellationToken::default(),
             gate: Gate::default(),
             timeline_get_throttle: Arc::new(throttle::Throttle::new(

pageserver/src/tenant/remote_timeline_client/download.rs

@@ -27,7 +27,7 @@ use crate::tenant::Generation;
 use crate::virtual_file::owned_buffers_io::io_buf_ext::IoBufExt;
 use crate::virtual_file::{on_fatal_io_error, MaybeFatalIo, VirtualFile};
 use crate::TEMP_FILE_SUFFIX;
-use remote_storage::{DownloadError, GenericRemoteStorage, ListingMode, RemotePath};
+use remote_storage::{DownloadError, DownloadOpts, GenericRemoteStorage, ListingMode, RemotePath};
 use utils::crashsafe::path_with_suffix_extension;
 use utils::id::{TenantId, TimelineId};
 use utils::pausable_failpoint;
@@ -153,7 +153,9 @@ async fn download_object<'a>(
                     .with_context(|| format!("create a destination file for layer '{dst_path}'"))
                     .map_err(DownloadError::Other)?;
 
-                let download = storage.download(src_path, cancel).await?;
+                let download = storage
+                    .download(src_path, &DownloadOpts::default(), cancel)
+                    .await?;
 
                 pausable_failpoint!("before-downloading-layer-stream-pausable");
 
@@ -204,7 +206,9 @@ async fn download_object<'a>(
                     .with_context(|| format!("create a destination file for layer '{dst_path}'"))
                     .map_err(DownloadError::Other)?;
 
-                let mut download = storage.download(src_path, cancel).await?;
+                let mut download = storage
+                    .download(src_path, &DownloadOpts::default(), cancel)
+                    .await?;
 
                 pausable_failpoint!("before-downloading-layer-stream-pausable");
 
@@ -344,7 +348,9 @@ async fn do_download_index_part(
 
     let index_part_bytes = download_retry_forever(
         || async {
-            let download = storage.download(&remote_path, cancel).await?;
+            let download = storage
+                .download(&remote_path, &DownloadOpts::default(), cancel)
+                .await?;
 
             let mut bytes = Vec::new();
 
@@ -526,10 +532,15 @@ pub(crate) async fn download_initdb_tar_zst(
                 .with_context(|| format!("tempfile creation {temp_path}"))
                 .map_err(DownloadError::Other)?;
 
-            let download = match storage.download(&remote_path, cancel).await {
+            let download = match storage
+                .download(&remote_path, &DownloadOpts::default(), cancel)
+                .await
+            {
                 Ok(dl) => dl,
                 Err(DownloadError::NotFound) => {
-                    storage.download(&remote_preserved_path, cancel).await?
+                    storage
+                        .download(&remote_preserved_path, &DownloadOpts::default(), cancel)
+                        .await?
                 }
                 Err(other) => Err(other)?,
             };

pageserver/src/tenant/secondary/downloader.rs

@@ -49,7 +49,7 @@ use futures::Future;
 use metrics::UIntGauge;
 use pageserver_api::models::SecondaryProgress;
 use pageserver_api::shard::TenantShardId;
-use remote_storage::{DownloadError, Etag, GenericRemoteStorage};
+use remote_storage::{DownloadError, DownloadOpts, Etag, GenericRemoteStorage};
 
 use tokio_util::sync::CancellationToken;
 use tracing::{info_span, instrument, warn, Instrument};
@@ -944,36 +944,34 @@ impl<'a> TenantDownloader<'a> {
     ) -> Result<HeatMapDownload, UpdateError> {
         debug_assert_current_span_has_tenant_id();
         let tenant_shard_id = self.secondary_state.get_tenant_shard_id();
-        // TODO: pull up etag check into the request, to do a conditional GET rather than
-        // issuing a GET and then maybe ignoring the response body
-        // (https://github.com/neondatabase/neon/issues/6199)
         tracing::debug!("Downloading heatmap for secondary tenant",);
 
         let heatmap_path = remote_heatmap_path(tenant_shard_id);
         let cancel = &self.secondary_state.cancel;
+        let opts = DownloadOpts {
+            etag: prev_etag.cloned(),
+        };
 
         backoff::retry(
             || async {
-                let download = self
+                let download = match self
                     .remote_storage
-                    .download(&heatmap_path, cancel)
+                    .download(&heatmap_path, &opts, cancel)
                     .await
-                    .map_err(UpdateError::from)?;
+                {
+                    Ok(download) => download,
+                    Err(DownloadError::Unmodified) => return Ok(HeatMapDownload::Unmodified),
+                    Err(err) => return Err(err.into()),
+                };
 
-                SECONDARY_MODE.download_heatmap.inc();
-
-                if Some(&download.etag) == prev_etag {
-                    Ok(HeatMapDownload::Unmodified)
-                } else {
-                    let mut heatmap_bytes = Vec::new();
-                    let mut body = tokio_util::io::StreamReader::new(download.download_stream);
-                    let _size = tokio::io::copy_buf(&mut body, &mut heatmap_bytes).await?;
-                    Ok(HeatMapDownload::Modified(HeatMapModified {
-                        etag: download.etag,
-                        last_modified: download.last_modified,
-                        bytes: heatmap_bytes,
-                    }))
-                }
+                let mut heatmap_bytes = Vec::new();
+                let mut body = tokio_util::io::StreamReader::new(download.download_stream);
+                let _size = tokio::io::copy_buf(&mut body, &mut heatmap_bytes).await?;
+                Ok(HeatMapDownload::Modified(HeatMapModified {
+                    etag: download.etag,
+                    last_modified: download.last_modified,
+                    bytes: heatmap_bytes,
+                }))
             },
             |e| matches!(e, UpdateError::NoData | UpdateError::Cancelled),
             FAILED_DOWNLOAD_WARN_THRESHOLD,
@@ -984,6 +982,7 @@ impl<'a> TenantDownloader<'a> {
         .await
         .ok_or_else(|| UpdateError::Cancelled)
         .and_then(|x| x)
+        .inspect(|_| SECONDARY_MODE.download_heatmap.inc())
     }
 
     /// Download heatmap layers that are not present on local disk, or update their

pageserver/src/tenant/storage_layer/delta_layer.rs

@@ -53,6 +53,7 @@ use camino::{Utf8Path, Utf8PathBuf};
 use futures::StreamExt;
 use itertools::Itertools;
 use pageserver_api::config::MaxVectoredReadBytes;
+use pageserver_api::key::DBDIR_KEY;
 use pageserver_api::keyspace::KeySpace;
 use pageserver_api::models::ImageCompressionAlgorithm;
 use pageserver_api::shard::TenantShardId;
@@ -963,14 +964,25 @@ impl DeltaLayerInner {
                 .blobs_at
                 .as_slice()
                 .iter()
-                .map(|(_, blob_meta)| format!("{}@{}", blob_meta.key, blob_meta.lsn))
+                .filter_map(|(_, blob_meta)| {
+                    if blob_meta.key.is_rel_dir_key() || blob_meta.key == DBDIR_KEY {
+                        // The size of values for these keys is unbounded and can
+                        // grow very large in pathological cases.
+                        None
+                    } else {
+                        Some(format!("{}@{}", blob_meta.key, blob_meta.lsn))
+                    }
+                })
                 .join(", ");
-            tracing::warn!(
-                "Oversized vectored read ({} > {}) for keys {}",
-                largest_read_size,
-                read_size_soft_max,
-                offenders
-            );
+
+            if !offenders.is_empty() {
+                tracing::warn!(
+                    "Oversized vectored read ({} > {}) for keys {}",
+                    largest_read_size,
+                    read_size_soft_max,
+                    offenders
+                );
+            }
         }
 
         largest_read_size

pageserver/src/tenant/storage_layer/image_layer.rs

@@ -49,6 +49,7 @@ use camino::{Utf8Path, Utf8PathBuf};
 use hex;
 use itertools::Itertools;
 use pageserver_api::config::MaxVectoredReadBytes;
+use pageserver_api::key::DBDIR_KEY;
 use pageserver_api::keyspace::KeySpace;
 use pageserver_api::shard::{ShardIdentity, TenantShardId};
 use rand::{distributions::Alphanumeric, Rng};
@@ -587,14 +588,25 @@ impl ImageLayerInner {
                     .blobs_at
                     .as_slice()
                     .iter()
-                    .map(|(_, blob_meta)| format!("{}@{}", blob_meta.key, blob_meta.lsn))
+                    .filter_map(|(_, blob_meta)| {
+                        if blob_meta.key.is_rel_dir_key() || blob_meta.key == DBDIR_KEY {
+                            // The size of values for these keys is unbounded and can
+                            // grow very large in pathological cases.
+                            None
+                        } else {
+                            Some(format!("{}@{}", blob_meta.key, blob_meta.lsn))
+                        }
+                    })
                     .join(", ");
-                tracing::warn!(
-                    "Oversized vectored read ({} > {}) for keys {}",
-                    buf_size,
-                    max_vectored_read_bytes,
-                    offenders
-                );
+
+                if !offenders.is_empty() {
+                    tracing::warn!(
+                        "Oversized vectored read ({} > {}) for keys {}",
+                        buf_size,
+                        max_vectored_read_bytes,
+                        offenders
+                    );
+                }
             }
 
             let buf = BytesMut::with_capacity(buf_size);

pageserver/src/tenant/storage_layer/layer.rs

@@ -442,11 +442,13 @@ impl Layer {
             // Visibility was modified to Visible: maybe log about this
             match ctx.task_kind() {
                 TaskKind::CalculateSyntheticSize
+                | TaskKind::OndemandLogicalSizeCalculation
                 | TaskKind::GarbageCollector
                 | TaskKind::MgmtRequest => {
                     // This situation is expected in code paths do binary searches of the LSN space to resolve
                     // an LSN to a timestamp, which happens during GC, during GC cutoff calculations in synthetic size,
-                    // and on-demand for certain HTTP API requests.
+                    // and on-demand for certain HTTP API requests. On-demand logical size calculation is also included
+                    // because it is run as a sub-task of synthetic size.
                 }
                 _ => {
                     // In all other contexts, it is unusual to do I/O involving layers which are not visible at
@@ -456,8 +458,8 @@ impl Layer {
                     // This case is legal in brief time windows: for example an in-flight getpage request can hold on to a layer object
                     // which was covered by a concurrent compaction.
                     tracing::info!(
-                        "Layer {} became visible as a result of access",
-                        self.0.desc.key()
+                        layer=%self,
+                        "became visible as a result of access",
                     );
                 }
             }
@@ -686,7 +688,9 @@ impl Drop for LayerInner {
             // and we could be delaying shutdown for nothing.
         }
 
-        if let Some(timeline) = self.timeline.upgrade() {
+        let timeline = self.timeline.upgrade();
+
+        if let Some(timeline) = timeline.as_ref() {
             // Only need to decrement metrics if the timeline still exists: otherwise
             // it will have already de-registered these metrics via TimelineMetrics::shutdown
             if self.desc.is_delta() {
@@ -717,7 +721,6 @@ 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 timeline = self.timeline.clone();
         let meta = self.metadata();
         let status = self.status.take();
 
@@ -727,7 +730,7 @@ impl Drop for LayerInner {
             // carry this until we are finished for [`Layer::wait_drop`] support
             let _status = status;
 
-            let Some(timeline) = timeline.upgrade() else {
+            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);

pageserver/src/tenant/timeline.rs

@@ -48,7 +48,6 @@ use utils::{
     sync::gate::{Gate, GateGuard},
 };
 
-use std::pin::pin;
 use std::sync::atomic::Ordering as AtomicOrdering;
 use std::sync::{Arc, Mutex, RwLock, Weak};
 use std::time::{Duration, Instant, SystemTime};
@@ -62,6 +61,7 @@ use std::{
     collections::btree_map::Entry,
     ops::{Deref, Range},
 };
+use std::{pin::pin, sync::OnceLock};
 
 use crate::{
     aux_file::AuxFileSizeEstimator,
@@ -71,6 +71,7 @@ use crate::{
         metadata::TimelineMetadata,
         storage_layer::{inmemory_layer::IndexEntry, PersistentLayerDesc},
     },
+    walingest::WalLagCooldown,
     walredo,
 };
 use crate::{
@@ -429,6 +430,8 @@ pub struct Timeline {
     pub(crate) l0_flush_global_state: L0FlushGlobalState,
 
     pub(crate) handles: handle::PerTimelineState<crate::page_service::TenantManagerTypes>,
+
+    pub(crate) attach_wal_lag_cooldown: Arc<OnceLock<WalLagCooldown>>,
 }
 
 pub struct WalReceiverInfo {
@@ -737,6 +740,7 @@ pub enum GetLogicalSizePriority {
 pub(crate) enum CompactFlags {
     ForceRepartition,
     ForceImageLayerCreation,
+    ForceL0Compaction,
     EnhancedGcBottomMostCompaction,
     DryRun,
 }
@@ -2130,6 +2134,7 @@ impl Timeline {
         pg_version: u32,
         state: TimelineState,
         aux_file_policy: Option<AuxFilePolicy>,
+        attach_wal_lag_cooldown: Arc<OnceLock<WalLagCooldown>>,
         cancel: CancellationToken,
     ) -> Arc<Self> {
         let disk_consistent_lsn = metadata.disk_consistent_lsn();
@@ -2271,6 +2276,8 @@ impl Timeline {
                 l0_flush_global_state: resources.l0_flush_global_state,
 
                 handles: Default::default(),
+
+                attach_wal_lag_cooldown,
             };
 
             if aux_file_policy == Some(AuxFilePolicy::V1) {

pageserver/src/tenant/timeline/analysis.rs

@@ -11,6 +11,7 @@ pub(crate) struct RangeAnalysis {
     has_image: bool,
     num_of_deltas_above_image: usize,
     total_num_of_deltas: usize,
+    num_of_l0: usize,
 }
 
 impl Timeline {
@@ -20,8 +21,10 @@ impl Timeline {
         let mut delta_ranges = Vec::new();
         let mut image_ranges = Vec::new();
 
+        let num_of_l0;
         let all_layer_files = {
             let guard = self.layers.read().await;
+            num_of_l0 = guard.layer_map().unwrap().level0_deltas().len();
             guard.all_persistent_layers()
         };
         let lsn = self.get_last_record_lsn();
@@ -82,6 +85,7 @@ impl Timeline {
                 has_image: image_layer.is_some(),
                 num_of_deltas_above_image: maybe_delta_layers.len(),
                 total_num_of_deltas: pitr_delta_layers.len(),
+                num_of_l0,
             });
         }
 

pageserver/src/tenant/timeline/compaction.rs

@@ -353,7 +353,13 @@ impl Timeline {
 
                 // 2. Compact
                 let timer = self.metrics.compact_time_histo.start_timer();
-                let fully_compacted = self.compact_level0(target_file_size, ctx).await?;
+                let fully_compacted = self
+                    .compact_level0(
+                        target_file_size,
+                        flags.contains(CompactFlags::ForceL0Compaction),
+                        ctx,
+                    )
+                    .await?;
                 timer.stop_and_record();
 
                 let mut partitioning = dense_partitioning;
@@ -658,6 +664,7 @@ impl Timeline {
     async fn compact_level0(
         self: &Arc<Self>,
         target_file_size: u64,
+        force_compaction_ignore_threshold: bool,
         ctx: &RequestContext,
     ) -> Result<bool, CompactionError> {
         let CompactLevel0Phase1Result {
@@ -679,9 +686,15 @@ impl Timeline {
             let now = tokio::time::Instant::now();
             stats.read_lock_acquisition_micros =
                 DurationRecorder::Recorded(RecordedDuration(now - begin), now);
-            self.compact_level0_phase1(phase1_layers_locked, stats, target_file_size, &ctx)
-                .instrument(phase1_span)
-                .await?
+            self.compact_level0_phase1(
+                phase1_layers_locked,
+                stats,
+                target_file_size,
+                force_compaction_ignore_threshold,
+                &ctx,
+            )
+            .instrument(phase1_span)
+            .await?
         };
 
         if new_layers.is_empty() && deltas_to_compact.is_empty() {
@@ -700,6 +713,7 @@ impl Timeline {
         guard: tokio::sync::RwLockReadGuard<'a, LayerManager>,
         mut stats: CompactLevel0Phase1StatsBuilder,
         target_file_size: u64,
+        force_compaction_ignore_threshold: bool,
         ctx: &RequestContext,
     ) -> Result<CompactLevel0Phase1Result, CompactionError> {
         stats.read_lock_held_spawn_blocking_startup_micros =
@@ -711,11 +725,26 @@ impl Timeline {
         // Only compact if enough layers have accumulated.
         let threshold = self.get_compaction_threshold();
         if level0_deltas.is_empty() || level0_deltas.len() < threshold {
-            debug!(
-                level0_deltas = level0_deltas.len(),
-                threshold, "too few deltas to compact"
-            );
-            return Ok(CompactLevel0Phase1Result::default());
+            if force_compaction_ignore_threshold {
+                if !level0_deltas.is_empty() {
+                    info!(
+                        level0_deltas = level0_deltas.len(),
+                        threshold, "too few deltas to compact, but forcing compaction"
+                    );
+                } else {
+                    info!(
+                        level0_deltas = level0_deltas.len(),
+                        threshold, "too few deltas to compact, cannot force compaction"
+                    );
+                    return Ok(CompactLevel0Phase1Result::default());
+                }
+            } else {
+                debug!(
+                    level0_deltas = level0_deltas.len(),
+                    threshold, "too few deltas to compact"
+                );
+                return Ok(CompactLevel0Phase1Result::default());
+            }
         }
 
         let mut level0_deltas = level0_deltas

pageserver/src/walingest.rs

@@ -21,7 +21,10 @@
 //! redo Postgres process, but some records it can handle directly with
 //! bespoken Rust code.
 
+use std::sync::Arc;
+use std::sync::OnceLock;
 use std::time::Duration;
+use std::time::Instant;
 use std::time::SystemTime;
 
 use pageserver_api::shard::ShardIdentity;
@@ -69,7 +72,29 @@ impl CheckPoint {
     }
 }
 
+/// Temporary limitation of WAL lag warnings after attach
+///
+/// After tenant attach, we want to limit WAL lag warnings because
+/// we don't look at the WAL until the attach is complete, which
+/// might take a while.
+pub struct WalLagCooldown {
+    /// Until when should this limitation apply at all
+    active_until: std::time::Instant,
+    /// The maximum lag to suppress. Lags above this limit get reported anyways.
+    max_lag: Duration,
+}
+
+impl WalLagCooldown {
+    pub fn new(attach_start: Instant, attach_duration: Duration) -> Self {
+        Self {
+            active_until: attach_start + attach_duration * 3 + Duration::from_secs(120),
+            max_lag: attach_duration * 2 + Duration::from_secs(60),
+        }
+    }
+}
+
 pub struct WalIngest {
+    attach_wal_lag_cooldown: Arc<OnceLock<WalLagCooldown>>,
     shard: ShardIdentity,
     checkpoint: CheckPoint,
     checkpoint_modified: bool,
@@ -103,6 +128,7 @@ impl WalIngest {
             shard: *timeline.get_shard_identity(),
             checkpoint,
             checkpoint_modified: false,
+            attach_wal_lag_cooldown: timeline.attach_wal_lag_cooldown.clone(),
             warn_ingest_lag: WarnIngestLag {
                 lag_msg_ratelimit: RateLimit::new(std::time::Duration::from_secs(10)),
                 future_lsn_msg_ratelimit: RateLimit::new(std::time::Duration::from_secs(10)),
@@ -1429,6 +1455,13 @@ impl WalIngest {
                     Ok(lag) => {
                         if lag > conf.wait_lsn_timeout {
                             rate_limits.lag_msg_ratelimit.call2(|rate_limit_stats| {
+                                if let Some(cooldown) = self.attach_wal_lag_cooldown.get() {
+                                    if std::time::Instant::now() < cooldown.active_until && lag <= cooldown.max_lag {
+                                        return;
+                                    }
+                                } else {
+                                    // Still loading? We shouldn't be here
+                                }
                                 let lag = humantime::format_duration(lag);
                                 warn!(%rate_limit_stats, %lag, "ingesting record with timestamp lagging more than wait_lsn_timeout");
                             })

pgxn/neon/file_cache.c

@@ -42,6 +42,7 @@
 
 #include "hll.h"
 #include "bitmap.h"
+#include "neon.h"
 
 #define CriticalAssert(cond) do if (!(cond)) elog(PANIC, "Assertion %s failed at %s:%d: ", #cond, __FILE__, __LINE__); while (0)
 
@@ -173,7 +174,9 @@ lfc_disable(char const *op)
 			 * If the reason of error is ENOSPC, then truncation of file may
 			 * help to reclaim some space
 			 */
+			pgstat_report_wait_start(WAIT_EVENT_NEON_LFC_TRUNCATE);
 			int			rc = ftruncate(lfc_desc, 0);
+			pgstat_report_wait_end();
 
 			if (rc < 0)
 				elog(WARNING, "Failed to truncate local file cache %s: %m", lfc_path);
@@ -769,8 +772,10 @@ lfc_readv_select(NRelFileInfo rinfo, ForkNumber forkNum, BlockNumber blkno,
 
 		if (iteration_hits != 0)
 		{
+			pgstat_report_wait_start(WAIT_EVENT_NEON_LFC_READ);
 			rc = preadv(lfc_desc, iov, blocks_in_chunk,
 						((off_t) entry_offset * BLOCKS_PER_CHUNK + chunk_offs) * BLCKSZ);
+			pgstat_report_wait_end();
 
 			if (rc != (BLCKSZ * blocks_in_chunk))
 			{
@@ -944,8 +949,11 @@ lfc_writev(NRelFileInfo rinfo, ForkNumber forkNum, BlockNumber blkno,
 		lfc_ctl->writes += blocks_in_chunk;
 		LWLockRelease(lfc_lock);
 
+		pgstat_report_wait_start(WAIT_EVENT_NEON_LFC_WRITE);
 		rc = pwritev(lfc_desc, iov, blocks_in_chunk,
 					 ((off_t) entry_offset * BLOCKS_PER_CHUNK + chunk_offs) * BLCKSZ);
+		pgstat_report_wait_end();
+
 		if (rc != BLCKSZ * blocks_in_chunk)
 		{
 			lfc_disable("write");

pgxn/neon/libpagestore.c

@@ -490,7 +490,7 @@ pageserver_connect(shardno_t shard_no, int elevel)
 											   WL_EXIT_ON_PM_DEATH | WL_LATCH_SET | WL_SOCKET_READABLE,
 											   PQsocket(shard->conn),
 											   0,
-											   PG_WAIT_EXTENSION);
+											   WAIT_EVENT_NEON_PS_STARTING);
 					elog(DEBUG5, "PGRES_POLLING_READING=>%d", rc);
 					if (rc & WL_LATCH_SET)
 					{
@@ -512,7 +512,7 @@ pageserver_connect(shardno_t shard_no, int elevel)
 											   WL_EXIT_ON_PM_DEATH | WL_LATCH_SET | WL_SOCKET_WRITEABLE,
 											   PQsocket(shard->conn),
 											   0,
-											   PG_WAIT_EXTENSION);
+											   WAIT_EVENT_NEON_PS_STARTING);
 					elog(DEBUG5, "PGRES_POLLING_WRITING=>%d", rc);
 					if (rc & WL_LATCH_SET)
 					{
@@ -608,7 +608,8 @@ pageserver_connect(shardno_t shard_no, int elevel)
 			WaitEvent	event;
 
 			/* Sleep until there's something to do */
-			(void) WaitEventSetWait(shard->wes_read, -1L, &event, 1, PG_WAIT_EXTENSION);
+			(void) WaitEventSetWait(shard->wes_read, -1L, &event, 1,
+									WAIT_EVENT_NEON_PS_CONFIGURING);
 			ResetLatch(MyLatch);
 
 			CHECK_FOR_INTERRUPTS();
@@ -656,7 +657,8 @@ static int
 call_PQgetCopyData(shardno_t shard_no, char **buffer)
 {
 	int			ret;
-	PGconn	   *pageserver_conn = page_servers[shard_no].conn;
+	PageServer *shard = &page_servers[shard_no];
+	PGconn	   *pageserver_conn = shard->conn;
 
 retry:
 	ret = PQgetCopyData(pageserver_conn, buffer, 1 /* async */ );
@@ -666,7 +668,8 @@ retry:
 		WaitEvent	event;
 
 		/* Sleep until there's something to do */
-		(void) WaitEventSetWait(page_servers[shard_no].wes_read, -1L, &event, 1, PG_WAIT_EXTENSION);
+		(void) WaitEventSetWait(shard->wes_read, -1L, &event, 1,
+								WAIT_EVENT_NEON_PS_READ);
 		ResetLatch(MyLatch);
 
 		CHECK_FOR_INTERRUPTS();

pgxn/neon/neon.c

@@ -41,6 +41,9 @@
 #include "pagestore_client.h"
 #include "control_plane_connector.h"
 #include "walsender_hooks.h"
+#if PG_MAJORVERSION_NUM >= 16
+#include "storage/ipc.h"
+#endif
 
 PG_MODULE_MAGIC;
 void		_PG_init(void);
@@ -49,6 +52,23 @@ static int	logical_replication_max_snap_files = 300;
 
 static int  running_xacts_overflow_policy;
 
+#if PG_MAJORVERSION_NUM >= 16
+static shmem_startup_hook_type prev_shmem_startup_hook;
+
+static void neon_shmem_startup_hook(void);
+#endif
+#if PG_MAJORVERSION_NUM >= 17
+uint32		WAIT_EVENT_NEON_LFC_MAINTENANCE;
+uint32		WAIT_EVENT_NEON_LFC_READ;
+uint32		WAIT_EVENT_NEON_LFC_TRUNCATE;
+uint32		WAIT_EVENT_NEON_LFC_WRITE;
+uint32		WAIT_EVENT_NEON_PS_STARTING;
+uint32		WAIT_EVENT_NEON_PS_CONFIGURING;
+uint32		WAIT_EVENT_NEON_PS_SEND;
+uint32		WAIT_EVENT_NEON_PS_READ;
+uint32		WAIT_EVENT_NEON_WAL_DL;
+#endif
+
 enum RunningXactsOverflowPolicies {
 	OP_IGNORE,
 	OP_SKIP,
@@ -635,6 +655,9 @@ _PG_init(void)
 	 */
 #if PG_VERSION_NUM >= 160000
 	load_file("$libdir/neon_rmgr", false);
+
+	prev_shmem_startup_hook = shmem_startup_hook;
+	shmem_startup_hook = neon_shmem_startup_hook;
 #endif
 
 	pg_init_libpagestore();
@@ -721,3 +744,25 @@ backpressure_throttling_time(PG_FUNCTION_ARGS)
 {
 	PG_RETURN_UINT64(BackpressureThrottlingTime());
 }
+
+#if PG_MAJORVERSION_NUM >= 16
+static void
+neon_shmem_startup_hook(void)
+{
+	/* Initialize */
+	if (prev_shmem_startup_hook)
+		prev_shmem_startup_hook();
+
+#if PG_PG_MAJORVERSION_NUM >= 17
+	WAIT_EVENT_NEON_LFC_MAINTENANCE = WaitEventExtensionNew("Neon/FileCache_Maintenance");
+	WAIT_EVENT_NEON_LFC_READ = WaitEventExtensionNew("Neon/FileCache_Read");
+	WAIT_EVENT_NEON_LFC_TRUNCATE = WaitEventExtensionNew("Neon/FileCache_Truncate");
+	WAIT_EVENT_NEON_LFC_WRITE = WaitEventExtensionNew("Neon/FileCache_Write");
+	WAIT_EVENT_NEON_PS_STARTING = WaitEventExtensionNew("Neon/PS_Starting");
+	WAIT_EVENT_NEON_PS_CONFIGURING = WaitEventExtensionNew("Neon/PS_Configuring");
+	WAIT_EVENT_NEON_PS_SEND = WaitEventExtensionNew("Neon/PS_SendIO");
+	WAIT_EVENT_NEON_PS_READ = WaitEventExtensionNew("Neon/PS_ReadIO");
+	WAIT_EVENT_NEON_WAL_DL = WaitEventExtensionNew("Neon/WAL_Download");
+#endif
+}
+#endif

pgxn/neon/neon.h

@@ -12,6 +12,7 @@
 #ifndef NEON_H
 #define NEON_H
 #include "access/xlogreader.h"
+#include "utils/wait_event.h"
 
 /* GUCs */
 extern char *neon_auth_token;
@@ -22,6 +23,28 @@ extern char *wal_acceptors_list;
 extern int	wal_acceptor_reconnect_timeout;
 extern int	wal_acceptor_connection_timeout;
 
+#if PG_MAJORVERSION_NUM >= 17
+extern uint32		WAIT_EVENT_NEON_LFC_MAINTENANCE;
+extern uint32		WAIT_EVENT_NEON_LFC_READ;
+extern uint32		WAIT_EVENT_NEON_LFC_TRUNCATE;
+extern uint32		WAIT_EVENT_NEON_LFC_WRITE;
+extern uint32		WAIT_EVENT_NEON_PS_STARTING;
+extern uint32		WAIT_EVENT_NEON_PS_CONFIGURING;
+extern uint32		WAIT_EVENT_NEON_PS_SEND;
+extern uint32		WAIT_EVENT_NEON_PS_READ;
+extern uint32		WAIT_EVENT_NEON_WAL_DL;
+#else
+#define WAIT_EVENT_NEON_LFC_MAINTENANCE	PG_WAIT_EXTENSION
+#define WAIT_EVENT_NEON_LFC_READ		WAIT_EVENT_BUFFILE_READ
+#define WAIT_EVENT_NEON_LFC_TRUNCATE	WAIT_EVENT_BUFFILE_TRUNCATE
+#define WAIT_EVENT_NEON_LFC_WRITE		WAIT_EVENT_BUFFILE_WRITE
+#define WAIT_EVENT_NEON_PS_STARTING		PG_WAIT_EXTENSION
+#define WAIT_EVENT_NEON_PS_CONFIGURING	PG_WAIT_EXTENSION
+#define WAIT_EVENT_NEON_PS_SEND			PG_WAIT_EXTENSION
+#define WAIT_EVENT_NEON_PS_READ			PG_WAIT_EXTENSION
+#define WAIT_EVENT_NEON_WAL_DL			WAIT_EVENT_WAL_READ
+#endif
+
 extern void pg_init_libpagestore(void);
 extern void pg_init_walproposer(void);
 

pgxn/neon/walproposer.c

@@ -213,7 +213,7 @@ WalProposerPoll(WalProposer *wp)
 		rc = wp->api.wait_event_set(wp, timeout, &sk, &events);
 
 		/* Exit loop if latch is set (we got new WAL) */
-		if ((rc == 1 && events & WL_LATCH_SET))
+		if (rc == 1 && (events & WL_LATCH_SET))
 			break;
 
 		/*

pgxn/neon/walproposer_pg.c

@@ -1814,7 +1814,7 @@ walprop_pg_wait_event_set(WalProposer *wp, long timeout, Safekeeper **sk, uint32
 	 * If wait is terminated by latch set (walsenders' latch is set on each
 	 * wal flush). (no need for pm death check due to WL_EXIT_ON_PM_DEATH)
 	 */
-	if ((rc == 1 && event.events & WL_LATCH_SET) || late_cv_trigger)
+	if ((rc == 1 && (event.events & WL_LATCH_SET)) || late_cv_trigger)
 	{
 		/* Reset our latch */
 		ResetLatch(MyLatch);
@@ -1826,7 +1826,7 @@ walprop_pg_wait_event_set(WalProposer *wp, long timeout, Safekeeper **sk, uint32
 	 * If the event contains something about the socket, it means we got an
 	 * event from a safekeeper socket.
 	 */
-	if (rc == 1 && (event.events & (WL_SOCKET_MASK)))
+	if (rc == 1 && (event.events & WL_SOCKET_MASK))
 	{
 		*sk = (Safekeeper *) event.user_data;
 		*events = event.events;

pgxn/neon/walsender_hooks.c

@@ -160,7 +160,7 @@ NeonWALPageRead(
 							  WL_LATCH_SET | WL_EXIT_ON_PM_DEATH | reader_events,
 							  sock,
 							  timeout_ms,
-							  WAIT_EVENT_WAL_SENDER_MAIN);
+							  WAIT_EVENT_NEON_WAL_DL);
 		}
 	}
 }
@@ -191,13 +191,14 @@ NeonOnDemandXLogReaderRoutines(XLogReaderRoutine *xlr)
 
 	if (!wal_reader)
 	{
-		XLogRecPtr	epochStartLsn = pg_atomic_read_u64(&GetWalpropShmemState()->propEpochStartLsn);
+		XLogRecPtr	basebackupLsn = GetRedoStartLsn();
 
-		if (epochStartLsn == 0)
+		/* should never happen */
+		if (basebackupLsn == 0)
 		{
-			elog(ERROR, "Unable to start walsender when propEpochStartLsn is 0!");
+			elog(ERROR, "unable to start walsender when basebackupLsn is 0");
 		}
-		wal_reader = NeonWALReaderAllocate(wal_segment_size, epochStartLsn, "[walsender] ");
+		wal_reader = NeonWALReaderAllocate(wal_segment_size, basebackupLsn, "[walsender] ");
 	}
 	xlr->page_read = NeonWALPageRead;
 	xlr->segment_open = NeonWALReadSegmentOpen;

poetry.lock

@@ -2064,73 +2064,80 @@ test = ["enum34", "ipaddress", "mock", "pywin32", "wmi"]
 
 [[package]]
 name = "psycopg2-binary"
-version = "2.9.6"
+version = "2.9.9"
 description = "psycopg2 - Python-PostgreSQL Database Adapter"
 optional = false
-python-versions = ">=3.6"
+python-versions = ">=3.7"
 files = [
-    {file = "psycopg2-binary-2.9.6.tar.gz", hash = "sha256:1f64dcfb8f6e0c014c7f55e51c9759f024f70ea572fbdef123f85318c297947c"},
-    {file = "psycopg2_binary-2.9.6-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:d26e0342183c762de3276cca7a530d574d4e25121ca7d6e4a98e4f05cb8e4df7"},
-    {file = "psycopg2_binary-2.9.6-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:c48d8f2db17f27d41fb0e2ecd703ea41984ee19362cbce52c097963b3a1b4365"},
-    {file = "psycopg2_binary-2.9.6-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:ffe9dc0a884a8848075e576c1de0290d85a533a9f6e9c4e564f19adf8f6e54a7"},
-    {file = "psycopg2_binary-2.9.6-cp310-cp310-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:8a76e027f87753f9bd1ab5f7c9cb8c7628d1077ef927f5e2446477153a602f2c"},
-    {file = "psycopg2_binary-2.9.6-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:6460c7a99fc939b849431f1e73e013d54aa54293f30f1109019c56a0b2b2ec2f"},
-    {file = "psycopg2_binary-2.9.6-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ae102a98c547ee2288637af07393dd33f440c25e5cd79556b04e3fca13325e5f"},
-    {file = "psycopg2_binary-2.9.6-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:9972aad21f965599ed0106f65334230ce826e5ae69fda7cbd688d24fa922415e"},
-    {file = "psycopg2_binary-2.9.6-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:7a40c00dbe17c0af5bdd55aafd6ff6679f94a9be9513a4c7e071baf3d7d22a70"},
-    {file = "psycopg2_binary-2.9.6-cp310-cp310-musllinux_1_1_ppc64le.whl", hash = "sha256:cacbdc5839bdff804dfebc058fe25684cae322987f7a38b0168bc1b2df703fb1"},
-    {file = "psycopg2_binary-2.9.6-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:7f0438fa20fb6c7e202863e0d5ab02c246d35efb1d164e052f2f3bfe2b152bd0"},
-    {file = "psycopg2_binary-2.9.6-cp310-cp310-win32.whl", hash = "sha256:b6c8288bb8a84b47e07013bb4850f50538aa913d487579e1921724631d02ea1b"},
-    {file = "psycopg2_binary-2.9.6-cp310-cp310-win_amd64.whl", hash = "sha256:61b047a0537bbc3afae10f134dc6393823882eb263088c271331602b672e52e9"},
-    {file = "psycopg2_binary-2.9.6-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:964b4dfb7c1c1965ac4c1978b0f755cc4bd698e8aa2b7667c575fb5f04ebe06b"},
-    {file = "psycopg2_binary-2.9.6-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:afe64e9b8ea66866a771996f6ff14447e8082ea26e675a295ad3bdbffdd72afb"},
-    {file = "psycopg2_binary-2.9.6-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:15e2ee79e7cf29582ef770de7dab3d286431b01c3bb598f8e05e09601b890081"},
-    {file = "psycopg2_binary-2.9.6-cp311-cp311-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:dfa74c903a3c1f0d9b1c7e7b53ed2d929a4910e272add6700c38f365a6002820"},
-    {file = "psycopg2_binary-2.9.6-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:b83456c2d4979e08ff56180a76429263ea254c3f6552cd14ada95cff1dec9bb8"},
-    {file = "psycopg2_binary-2.9.6-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:0645376d399bfd64da57148694d78e1f431b1e1ee1054872a5713125681cf1be"},
-    {file = "psycopg2_binary-2.9.6-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:e99e34c82309dd78959ba3c1590975b5d3c862d6f279f843d47d26ff89d7d7e1"},
-    {file = "psycopg2_binary-2.9.6-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:4ea29fc3ad9d91162c52b578f211ff1c931d8a38e1f58e684c45aa470adf19e2"},
-    {file = "psycopg2_binary-2.9.6-cp311-cp311-musllinux_1_1_ppc64le.whl", hash = "sha256:4ac30da8b4f57187dbf449294d23b808f8f53cad6b1fc3623fa8a6c11d176dd0"},
-    {file = "psycopg2_binary-2.9.6-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:e78e6e2a00c223e164c417628572a90093c031ed724492c763721c2e0bc2a8df"},
-    {file = "psycopg2_binary-2.9.6-cp311-cp311-win32.whl", hash = "sha256:1876843d8e31c89c399e31b97d4b9725a3575bb9c2af92038464231ec40f9edb"},
-    {file = "psycopg2_binary-2.9.6-cp311-cp311-win_amd64.whl", hash = "sha256:b4b24f75d16a89cc6b4cdff0eb6a910a966ecd476d1e73f7ce5985ff1328e9a6"},
-    {file = "psycopg2_binary-2.9.6-cp36-cp36m-win32.whl", hash = "sha256:498807b927ca2510baea1b05cc91d7da4718a0f53cb766c154c417a39f1820a0"},
-    {file = "psycopg2_binary-2.9.6-cp36-cp36m-win_amd64.whl", hash = "sha256:0d236c2825fa656a2d98bbb0e52370a2e852e5a0ec45fc4f402977313329174d"},
-    {file = "psycopg2_binary-2.9.6-cp37-cp37m-macosx_10_9_x86_64.whl", hash = "sha256:34b9ccdf210cbbb1303c7c4db2905fa0319391bd5904d32689e6dd5c963d2ea8"},
-    {file = "psycopg2_binary-2.9.6-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:84d2222e61f313c4848ff05353653bf5f5cf6ce34df540e4274516880d9c3763"},
-    {file = "psycopg2_binary-2.9.6-cp37-cp37m-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:30637a20623e2a2eacc420059be11527f4458ef54352d870b8181a4c3020ae6b"},
-    {file = "psycopg2_binary-2.9.6-cp37-cp37m-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8122cfc7cae0da9a3077216528b8bb3629c43b25053284cc868744bfe71eb141"},
-    {file = "psycopg2_binary-2.9.6-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:38601cbbfe600362c43714482f43b7c110b20cb0f8172422c616b09b85a750c5"},
-    {file = "psycopg2_binary-2.9.6-cp37-cp37m-musllinux_1_1_aarch64.whl", hash = "sha256:c7e62ab8b332147a7593a385d4f368874d5fe4ad4e341770d4983442d89603e3"},
-    {file = "psycopg2_binary-2.9.6-cp37-cp37m-musllinux_1_1_i686.whl", hash = "sha256:2ab652e729ff4ad76d400df2624d223d6e265ef81bb8aa17fbd63607878ecbee"},
-    {file = "psycopg2_binary-2.9.6-cp37-cp37m-musllinux_1_1_ppc64le.whl", hash = "sha256:c83a74b68270028dc8ee74d38ecfaf9c90eed23c8959fca95bd703d25b82c88e"},
-    {file = "psycopg2_binary-2.9.6-cp37-cp37m-musllinux_1_1_x86_64.whl", hash = "sha256:d4e6036decf4b72d6425d5b29bbd3e8f0ff1059cda7ac7b96d6ac5ed34ffbacd"},
-    {file = "psycopg2_binary-2.9.6-cp37-cp37m-win32.whl", hash = "sha256:a8c28fd40a4226b4a84bdf2d2b5b37d2c7bd49486b5adcc200e8c7ec991dfa7e"},
-    {file = "psycopg2_binary-2.9.6-cp37-cp37m-win_amd64.whl", hash = "sha256:51537e3d299be0db9137b321dfb6a5022caaab275775680e0c3d281feefaca6b"},
-    {file = "psycopg2_binary-2.9.6-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:cf4499e0a83b7b7edcb8dabecbd8501d0d3a5ef66457200f77bde3d210d5debb"},
-    {file = "psycopg2_binary-2.9.6-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:7e13a5a2c01151f1208d5207e42f33ba86d561b7a89fca67c700b9486a06d0e2"},
-    {file = "psycopg2_binary-2.9.6-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:0e0f754d27fddcfd74006455b6e04e6705d6c31a612ec69ddc040a5468e44b4e"},
-    {file = "psycopg2_binary-2.9.6-cp38-cp38-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:d57c3fd55d9058645d26ae37d76e61156a27722097229d32a9e73ed54819982a"},
-    {file = "psycopg2_binary-2.9.6-cp38-cp38-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:71f14375d6f73b62800530b581aed3ada394039877818b2d5f7fc77e3bb6894d"},
-    {file = "psycopg2_binary-2.9.6-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:441cc2f8869a4f0f4bb408475e5ae0ee1f3b55b33f350406150277f7f35384fc"},
-    {file = "psycopg2_binary-2.9.6-cp38-cp38-musllinux_1_1_aarch64.whl", hash = "sha256:65bee1e49fa6f9cf327ce0e01c4c10f39165ee76d35c846ade7cb0ec6683e303"},
-    {file = "psycopg2_binary-2.9.6-cp38-cp38-musllinux_1_1_i686.whl", hash = "sha256:af335bac6b666cc6aea16f11d486c3b794029d9df029967f9938a4bed59b6a19"},
-    {file = "psycopg2_binary-2.9.6-cp38-cp38-musllinux_1_1_ppc64le.whl", hash = "sha256:cfec476887aa231b8548ece2e06d28edc87c1397ebd83922299af2e051cf2827"},
-    {file = "psycopg2_binary-2.9.6-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:65c07febd1936d63bfde78948b76cd4c2a411572a44ac50719ead41947d0f26b"},
-    {file = "psycopg2_binary-2.9.6-cp38-cp38-win32.whl", hash = "sha256:4dfb4be774c4436a4526d0c554af0cc2e02082c38303852a36f6456ece7b3503"},
-    {file = "psycopg2_binary-2.9.6-cp38-cp38-win_amd64.whl", hash = "sha256:02c6e3cf3439e213e4ee930308dc122d6fb4d4bea9aef4a12535fbd605d1a2fe"},
-    {file = "psycopg2_binary-2.9.6-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:e9182eb20f41417ea1dd8e8f7888c4d7c6e805f8a7c98c1081778a3da2bee3e4"},
-    {file = "psycopg2_binary-2.9.6-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:8a6979cf527e2603d349a91060f428bcb135aea2be3201dff794813256c274f1"},
-    {file = "psycopg2_binary-2.9.6-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:8338a271cb71d8da40b023a35d9c1e919eba6cbd8fa20a54b748a332c355d896"},
-    {file = "psycopg2_binary-2.9.6-cp39-cp39-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:e3ed340d2b858d6e6fb5083f87c09996506af483227735de6964a6100b4e6a54"},
-    {file = "psycopg2_binary-2.9.6-cp39-cp39-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:f81e65376e52f03422e1fb475c9514185669943798ed019ac50410fb4c4df232"},
-    {file = "psycopg2_binary-2.9.6-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bfb13af3c5dd3a9588000910178de17010ebcccd37b4f9794b00595e3a8ddad3"},
-    {file = "psycopg2_binary-2.9.6-cp39-cp39-musllinux_1_1_aarch64.whl", hash = "sha256:4c727b597c6444a16e9119386b59388f8a424223302d0c06c676ec8b4bc1f963"},
-    {file = "psycopg2_binary-2.9.6-cp39-cp39-musllinux_1_1_i686.whl", hash = "sha256:4d67fbdaf177da06374473ef6f7ed8cc0a9dc640b01abfe9e8a2ccb1b1402c1f"},
-    {file = "psycopg2_binary-2.9.6-cp39-cp39-musllinux_1_1_ppc64le.whl", hash = "sha256:0892ef645c2fabb0c75ec32d79f4252542d0caec1d5d949630e7d242ca4681a3"},
-    {file = "psycopg2_binary-2.9.6-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:02c0f3757a4300cf379eb49f543fb7ac527fb00144d39246ee40e1df684ab514"},
-    {file = "psycopg2_binary-2.9.6-cp39-cp39-win32.whl", hash = "sha256:c3dba7dab16709a33a847e5cd756767271697041fbe3fe97c215b1fc1f5c9848"},
-    {file = "psycopg2_binary-2.9.6-cp39-cp39-win_amd64.whl", hash = "sha256:f6a88f384335bb27812293fdb11ac6aee2ca3f51d3c7820fe03de0a304ab6249"},
+    {file = "psycopg2-binary-2.9.9.tar.gz", hash = "sha256:7f01846810177d829c7692f1f5ada8096762d9172af1b1a28d4ab5b77c923c1c"},
+    {file = "psycopg2_binary-2.9.9-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:c2470da5418b76232f02a2fcd2229537bb2d5a7096674ce61859c3229f2eb202"},
+    {file = "psycopg2_binary-2.9.9-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:c6af2a6d4b7ee9615cbb162b0738f6e1fd1f5c3eda7e5da17861eacf4c717ea7"},
+    {file = "psycopg2_binary-2.9.9-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:75723c3c0fbbf34350b46a3199eb50638ab22a0228f93fb472ef4d9becc2382b"},
+    {file = "psycopg2_binary-2.9.9-cp310-cp310-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:83791a65b51ad6ee6cf0845634859d69a038ea9b03d7b26e703f94c7e93dbcf9"},
+    {file = "psycopg2_binary-2.9.9-cp310-cp310-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:0ef4854e82c09e84cc63084a9e4ccd6d9b154f1dbdd283efb92ecd0b5e2b8c84"},
+    {file = "psycopg2_binary-2.9.9-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ed1184ab8f113e8d660ce49a56390ca181f2981066acc27cf637d5c1e10ce46e"},
+    {file = "psycopg2_binary-2.9.9-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:d2997c458c690ec2bc6b0b7ecbafd02b029b7b4283078d3b32a852a7ce3ddd98"},
+    {file = "psycopg2_binary-2.9.9-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:b58b4710c7f4161b5e9dcbe73bb7c62d65670a87df7bcce9e1faaad43e715245"},
+    {file = "psycopg2_binary-2.9.9-cp310-cp310-musllinux_1_1_ppc64le.whl", hash = "sha256:0c009475ee389757e6e34611d75f6e4f05f0cf5ebb76c6037508318e1a1e0d7e"},
+    {file = "psycopg2_binary-2.9.9-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:8dbf6d1bc73f1d04ec1734bae3b4fb0ee3cb2a493d35ede9badbeb901fb40f6f"},
+    {file = "psycopg2_binary-2.9.9-cp310-cp310-win32.whl", hash = "sha256:3f78fd71c4f43a13d342be74ebbc0666fe1f555b8837eb113cb7416856c79682"},
+    {file = "psycopg2_binary-2.9.9-cp310-cp310-win_amd64.whl", hash = "sha256:876801744b0dee379e4e3c38b76fc89f88834bb15bf92ee07d94acd06ec890a0"},
+    {file = "psycopg2_binary-2.9.9-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:ee825e70b1a209475622f7f7b776785bd68f34af6e7a46e2e42f27b659b5bc26"},
+    {file = "psycopg2_binary-2.9.9-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:1ea665f8ce695bcc37a90ee52de7a7980be5161375d42a0b6c6abedbf0d81f0f"},
+    {file = "psycopg2_binary-2.9.9-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:143072318f793f53819048fdfe30c321890af0c3ec7cb1dfc9cc87aa88241de2"},
+    {file = "psycopg2_binary-2.9.9-cp311-cp311-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:c332c8d69fb64979ebf76613c66b985414927a40f8defa16cf1bc028b7b0a7b0"},
+    {file = "psycopg2_binary-2.9.9-cp311-cp311-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:f7fc5a5acafb7d6ccca13bfa8c90f8c51f13d8fb87d95656d3950f0158d3ce53"},
+    {file = "psycopg2_binary-2.9.9-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:977646e05232579d2e7b9c59e21dbe5261f403a88417f6a6512e70d3f8a046be"},
+    {file = "psycopg2_binary-2.9.9-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:b6356793b84728d9d50ead16ab43c187673831e9d4019013f1402c41b1db9b27"},
+    {file = "psycopg2_binary-2.9.9-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:bc7bb56d04601d443f24094e9e31ae6deec9ccb23581f75343feebaf30423359"},
+    {file = "psycopg2_binary-2.9.9-cp311-cp311-musllinux_1_1_ppc64le.whl", hash = "sha256:77853062a2c45be16fd6b8d6de2a99278ee1d985a7bd8b103e97e41c034006d2"},
+    {file = "psycopg2_binary-2.9.9-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:78151aa3ec21dccd5cdef6c74c3e73386dcdfaf19bced944169697d7ac7482fc"},
+    {file = "psycopg2_binary-2.9.9-cp311-cp311-win32.whl", hash = "sha256:dc4926288b2a3e9fd7b50dc6a1909a13bbdadfc67d93f3374d984e56f885579d"},
+    {file = "psycopg2_binary-2.9.9-cp311-cp311-win_amd64.whl", hash = "sha256:b76bedd166805480ab069612119ea636f5ab8f8771e640ae103e05a4aae3e417"},
+    {file = "psycopg2_binary-2.9.9-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:8532fd6e6e2dc57bcb3bc90b079c60de896d2128c5d9d6f24a63875a95a088cf"},
+    {file = "psycopg2_binary-2.9.9-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:8f8544b092a29a6ddd72f3556a9fcf249ec412e10ad28be6a0c0d948924f2212"},
+    {file = "psycopg2_binary-2.9.9-cp312-cp312-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:2d423c8d8a3c82d08fe8af900ad5b613ce3632a1249fd6a223941d0735fce493"},
+    {file = "psycopg2_binary-2.9.9-cp312-cp312-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:2e5afae772c00980525f6d6ecf7cbca55676296b580c0e6abb407f15f3706996"},
+    {file = "psycopg2_binary-2.9.9-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6e6f98446430fdf41bd36d4faa6cb409f5140c1c2cf58ce0bbdaf16af7d3f119"},
+    {file = "psycopg2_binary-2.9.9-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:c77e3d1862452565875eb31bdb45ac62502feabbd53429fdc39a1cc341d681ba"},
+    {file = "psycopg2_binary-2.9.9-cp312-cp312-musllinux_1_1_i686.whl", hash = "sha256:cb16c65dcb648d0a43a2521f2f0a2300f40639f6f8c1ecbc662141e4e3e1ee07"},
+    {file = "psycopg2_binary-2.9.9-cp312-cp312-musllinux_1_1_ppc64le.whl", hash = "sha256:911dda9c487075abd54e644ccdf5e5c16773470a6a5d3826fda76699410066fb"},
+    {file = "psycopg2_binary-2.9.9-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:57fede879f08d23c85140a360c6a77709113efd1c993923c59fde17aa27599fe"},
+    {file = "psycopg2_binary-2.9.9-cp37-cp37m-macosx_10_9_x86_64.whl", hash = "sha256:2293b001e319ab0d869d660a704942c9e2cce19745262a8aba2115ef41a0a42a"},
+    {file = "psycopg2_binary-2.9.9-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:03ef7df18daf2c4c07e2695e8cfd5ee7f748a1d54d802330985a78d2a5a6dca9"},
+    {file = "psycopg2_binary-2.9.9-cp37-cp37m-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:0a602ea5aff39bb9fac6308e9c9d82b9a35c2bf288e184a816002c9fae930b77"},
+    {file = "psycopg2_binary-2.9.9-cp37-cp37m-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:8359bf4791968c5a78c56103702000105501adb557f3cf772b2c207284273984"},
+    {file = "psycopg2_binary-2.9.9-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:275ff571376626195ab95a746e6a04c7df8ea34638b99fc11160de91f2fef503"},
+    {file = "psycopg2_binary-2.9.9-cp37-cp37m-musllinux_1_1_aarch64.whl", hash = "sha256:f9b5571d33660d5009a8b3c25dc1db560206e2d2f89d3df1cb32d72c0d117d52"},
+    {file = "psycopg2_binary-2.9.9-cp37-cp37m-musllinux_1_1_i686.whl", hash = "sha256:420f9bbf47a02616e8554e825208cb947969451978dceb77f95ad09c37791dae"},
+    {file = "psycopg2_binary-2.9.9-cp37-cp37m-musllinux_1_1_ppc64le.whl", hash = "sha256:4154ad09dac630a0f13f37b583eae260c6aa885d67dfbccb5b02c33f31a6d420"},
+    {file = "psycopg2_binary-2.9.9-cp37-cp37m-musllinux_1_1_x86_64.whl", hash = "sha256:a148c5d507bb9b4f2030a2025c545fccb0e1ef317393eaba42e7eabd28eb6041"},
+    {file = "psycopg2_binary-2.9.9-cp37-cp37m-win32.whl", hash = "sha256:68fc1f1ba168724771e38bee37d940d2865cb0f562380a1fb1ffb428b75cb692"},
+    {file = "psycopg2_binary-2.9.9-cp37-cp37m-win_amd64.whl", hash = "sha256:281309265596e388ef483250db3640e5f414168c5a67e9c665cafce9492eda2f"},
+    {file = "psycopg2_binary-2.9.9-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:60989127da422b74a04345096c10d416c2b41bd7bf2a380eb541059e4e999980"},
+    {file = "psycopg2_binary-2.9.9-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:246b123cc54bb5361588acc54218c8c9fb73068bf227a4a531d8ed56fa3ca7d6"},
+    {file = "psycopg2_binary-2.9.9-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:34eccd14566f8fe14b2b95bb13b11572f7c7d5c36da61caf414d23b91fcc5d94"},
+    {file = "psycopg2_binary-2.9.9-cp38-cp38-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:18d0ef97766055fec15b5de2c06dd8e7654705ce3e5e5eed3b6651a1d2a9a152"},
+    {file = "psycopg2_binary-2.9.9-cp38-cp38-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:d3f82c171b4ccd83bbaf35aa05e44e690113bd4f3b7b6cc54d2219b132f3ae55"},
+    {file = "psycopg2_binary-2.9.9-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ead20f7913a9c1e894aebe47cccf9dc834e1618b7aa96155d2091a626e59c972"},
+    {file = "psycopg2_binary-2.9.9-cp38-cp38-musllinux_1_1_aarch64.whl", hash = "sha256:ca49a8119c6cbd77375ae303b0cfd8c11f011abbbd64601167ecca18a87e7cdd"},
+    {file = "psycopg2_binary-2.9.9-cp38-cp38-musllinux_1_1_i686.whl", hash = "sha256:323ba25b92454adb36fa425dc5cf6f8f19f78948cbad2e7bc6cdf7b0d7982e59"},
+    {file = "psycopg2_binary-2.9.9-cp38-cp38-musllinux_1_1_ppc64le.whl", hash = "sha256:1236ed0952fbd919c100bc839eaa4a39ebc397ed1c08a97fc45fee2a595aa1b3"},
+    {file = "psycopg2_binary-2.9.9-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:729177eaf0aefca0994ce4cffe96ad3c75e377c7b6f4efa59ebf003b6d398716"},
+    {file = "psycopg2_binary-2.9.9-cp38-cp38-win32.whl", hash = "sha256:804d99b24ad523a1fe18cc707bf741670332f7c7412e9d49cb5eab67e886b9b5"},
+    {file = "psycopg2_binary-2.9.9-cp38-cp38-win_amd64.whl", hash = "sha256:a6cdcc3ede532f4a4b96000b6362099591ab4a3e913d70bcbac2b56c872446f7"},
+    {file = "psycopg2_binary-2.9.9-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:72dffbd8b4194858d0941062a9766f8297e8868e1dd07a7b36212aaa90f49472"},
+    {file = "psycopg2_binary-2.9.9-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:30dcc86377618a4c8f3b72418df92e77be4254d8f89f14b8e8f57d6d43603c0f"},
+    {file = "psycopg2_binary-2.9.9-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:31a34c508c003a4347d389a9e6fcc2307cc2150eb516462a7a17512130de109e"},
+    {file = "psycopg2_binary-2.9.9-cp39-cp39-manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:15208be1c50b99203fe88d15695f22a5bed95ab3f84354c494bcb1d08557df67"},
+    {file = "psycopg2_binary-2.9.9-cp39-cp39-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:1873aade94b74715be2246321c8650cabf5a0d098a95bab81145ffffa4c13876"},
+    {file = "psycopg2_binary-2.9.9-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:3a58c98a7e9c021f357348867f537017057c2ed7f77337fd914d0bedb35dace7"},
+    {file = "psycopg2_binary-2.9.9-cp39-cp39-musllinux_1_1_aarch64.whl", hash = "sha256:4686818798f9194d03c9129a4d9a702d9e113a89cb03bffe08c6cf799e053291"},
+    {file = "psycopg2_binary-2.9.9-cp39-cp39-musllinux_1_1_i686.whl", hash = "sha256:ebdc36bea43063116f0486869652cb2ed7032dbc59fbcb4445c4862b5c1ecf7f"},
+    {file = "psycopg2_binary-2.9.9-cp39-cp39-musllinux_1_1_ppc64le.whl", hash = "sha256:ca08decd2697fdea0aea364b370b1249d47336aec935f87b8bbfd7da5b2ee9c1"},
+    {file = "psycopg2_binary-2.9.9-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:ac05fb791acf5e1a3e39402641827780fe44d27e72567a000412c648a85ba860"},
+    {file = "psycopg2_binary-2.9.9-cp39-cp39-win32.whl", hash = "sha256:9dba73be7305b399924709b91682299794887cbbd88e38226ed9f6712eabee90"},
+    {file = "psycopg2_binary-2.9.9-cp39-cp39-win_amd64.whl", hash = "sha256:f7ae5d65ccfbebdfa761585228eb4d0df3a8b15cfb53bd953e713e09fbb12957"},
 ]
 
 [[package]]
@@ -2577,7 +2584,6 @@ files = [
     {file = "PyYAML-6.0.1-cp311-cp311-win_amd64.whl", hash = "sha256:bf07ee2fef7014951eeb99f56f39c9bb4af143d8aa3c21b1677805985307da34"},
     {file = "PyYAML-6.0.1-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:855fb52b0dc35af121542a76b9a84f8d1cd886ea97c84703eaa6d88e37a2ad28"},
     {file = "PyYAML-6.0.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:40df9b996c2b73138957fe23a16a4f0ba614f4c0efce1e9406a184b6d07fa3a9"},
-    {file = "PyYAML-6.0.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a08c6f0fe150303c1c6b71ebcd7213c2858041a7e01975da3a99aed1e7a378ef"},
     {file = "PyYAML-6.0.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6c22bec3fbe2524cde73d7ada88f6566758a8f7227bfbf93a408a9d86bcc12a0"},
     {file = "PyYAML-6.0.1-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:8d4e9c88387b0f5c7d5f281e55304de64cf7f9c0021a3525bd3b1c542da3b0e4"},
     {file = "PyYAML-6.0.1-cp312-cp312-win32.whl", hash = "sha256:d483d2cdf104e7c9fa60c544d92981f12ad66a457afae824d146093b8c294c54"},
@@ -2702,13 +2708,13 @@ files = [
 
 [[package]]
 name = "requests"
-version = "2.32.0"
+version = "2.32.3"
 description = "Python HTTP for Humans."
 optional = false
 python-versions = ">=3.8"
 files = [
-    {file = "requests-2.32.0-py3-none-any.whl", hash = "sha256:f2c3881dddb70d056c5bd7600a4fae312b2a300e39be6a118d30b90bd27262b5"},
-    {file = "requests-2.32.0.tar.gz", hash = "sha256:fa5490319474c82ef1d2c9bc459d3652e3ae4ef4c4ebdd18a21145a47ca4b6b8"},
+    {file = "requests-2.32.3-py3-none-any.whl", hash = "sha256:70761cfe03c773ceb22aa2f671b4757976145175cdfca038c02654d061d6dcc6"},
+    {file = "requests-2.32.3.tar.gz", hash = "sha256:55365417734eb18255590a9ff9eb97e9e1da868d4ccd6402399eaf68af20a760"},
 ]
 
 [package.dependencies]
@@ -3131,16 +3137,6 @@ files = [
     {file = "wrapt-1.14.1-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:8ad85f7f4e20964db4daadcab70b47ab05c7c1cf2a7c1e51087bfaa83831854c"},
     {file = "wrapt-1.14.1-cp310-cp310-win32.whl", hash = "sha256:a9a52172be0b5aae932bef82a79ec0a0ce87288c7d132946d645eba03f0ad8a8"},
     {file = "wrapt-1.14.1-cp310-cp310-win_amd64.whl", hash = "sha256:6d323e1554b3d22cfc03cd3243b5bb815a51f5249fdcbb86fda4bf62bab9e164"},
-    {file = "wrapt-1.14.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:ecee4132c6cd2ce5308e21672015ddfed1ff975ad0ac8d27168ea82e71413f55"},
-    {file = "wrapt-1.14.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:2020f391008ef874c6d9e208b24f28e31bcb85ccff4f335f15a3251d222b92d9"},
-    {file = "wrapt-1.14.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2feecf86e1f7a86517cab34ae6c2f081fd2d0dac860cb0c0ded96d799d20b335"},
-    {file = "wrapt-1.14.1-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:240b1686f38ae665d1b15475966fe0472f78e71b1b4903c143a842659c8e4cb9"},
-    {file = "wrapt-1.14.1-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a9008dad07d71f68487c91e96579c8567c98ca4c3881b9b113bc7b33e9fd78b8"},
-    {file = "wrapt-1.14.1-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:6447e9f3ba72f8e2b985a1da758767698efa72723d5b59accefd716e9e8272bf"},
-    {file = "wrapt-1.14.1-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:acae32e13a4153809db37405f5eba5bac5fbe2e2ba61ab227926a22901051c0a"},
-    {file = "wrapt-1.14.1-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:49ef582b7a1152ae2766557f0550a9fcbf7bbd76f43fbdc94dd3bf07cc7168be"},
-    {file = "wrapt-1.14.1-cp311-cp311-win32.whl", hash = "sha256:358fe87cc899c6bb0ddc185bf3dbfa4ba646f05b1b0b9b5a27c2cb92c2cea204"},
-    {file = "wrapt-1.14.1-cp311-cp311-win_amd64.whl", hash = "sha256:26046cd03936ae745a502abf44dac702a5e6880b2b01c29aea8ddf3353b68224"},
     {file = "wrapt-1.14.1-cp35-cp35m-manylinux1_i686.whl", hash = "sha256:43ca3bbbe97af00f49efb06e352eae40434ca9d915906f77def219b88e85d907"},
     {file = "wrapt-1.14.1-cp35-cp35m-manylinux1_x86_64.whl", hash = "sha256:6b1a564e6cb69922c7fe3a678b9f9a3c54e72b469875aa8018f18b4d1dd1adf3"},
     {file = "wrapt-1.14.1-cp35-cp35m-manylinux2010_i686.whl", hash = "sha256:00b6d4ea20a906c0ca56d84f93065b398ab74b927a7a3dbd470f6fc503f95dc3"},
@@ -3378,4 +3374,4 @@ cffi = ["cffi (>=1.11)"]
 [metadata]
 lock-version = "2.0"
 python-versions = "^3.9"
-content-hash = "c09bcb333ab550958b33dbf4fec968c500d8e701fd4c96402cddbd9bb8048055"
+content-hash = "9055b73352f1534f664cd8af6ebf8d93cf3bf857f115756f312ff2e3ae1bbbc1"

proxy/Cargo.toml

@@ -38,7 +38,7 @@ hostname.workspace = true
 http.workspace = true
 humantime.workspace = true
 humantime-serde.workspace = true
-hyper.workspace = true
+hyper0.workspace = true
 hyper1 = { package = "hyper", version = "1.2", features = ["server"] }
 hyper-util = { version = "0.1", features = ["server", "http1", "http2", "tokio"] }
 http-body-util = { version = "0.1" }

proxy/src/auth.rs

@@ -73,6 +73,9 @@ pub(crate) enum AuthErrorImpl {
 
     #[error("Authentication timed out")]
     UserTimeout(Elapsed),
+
+    #[error("Disconnected due to inactivity after {0}.")]
+    ConfirmationTimeout(humantime::Duration),
 }
 
 #[derive(Debug, Error)]
@@ -103,6 +106,10 @@ impl AuthError {
     pub(crate) fn user_timeout(elapsed: Elapsed) -> Self {
         AuthErrorImpl::UserTimeout(elapsed).into()
     }
+
+    pub(crate) fn confirmation_timeout(timeout: humantime::Duration) -> Self {
+        AuthErrorImpl::ConfirmationTimeout(timeout).into()
+    }
 }
 
 impl<E: Into<AuthErrorImpl>> From<E> for AuthError {
@@ -125,6 +132,7 @@ impl UserFacingError for AuthError {
             AuthErrorImpl::IpAddressNotAllowed(_) => self.to_string(),
             AuthErrorImpl::TooManyConnections => self.to_string(),
             AuthErrorImpl::UserTimeout(_) => self.to_string(),
+            AuthErrorImpl::ConfirmationTimeout(_) => self.to_string(),
         }
     }
 }
@@ -143,6 +151,7 @@ impl ReportableError for AuthError {
             AuthErrorImpl::IpAddressNotAllowed(_) => crate::error::ErrorKind::User,
             AuthErrorImpl::TooManyConnections => crate::error::ErrorKind::RateLimit,
             AuthErrorImpl::UserTimeout(_) => crate::error::ErrorKind::User,
+            AuthErrorImpl::ConfirmationTimeout(_) => crate::error::ErrorKind::User,
         }
     }
 }

proxy/src/auth/backend.rs

@@ -620,6 +620,7 @@ mod tests {
         ip_allowlist_check_enabled: true,
         is_auth_broker: false,
         accept_jwts: false,
+        webauth_confirmation_timeout: std::time::Duration::from_secs(5),
     });
 
     async fn read_message(r: &mut (impl AsyncRead + Unpin), b: &mut BytesMut) -> PgMessage {

proxy/src/auth/backend/web.rs

@@ -89,7 +89,12 @@ pub(super) async fn authenticate(
 
     // Wait for web console response (see `mgmt`).
     info!(parent: &span, "waiting for console's reply...");
-    let db_info = waiter.await.map_err(WebAuthError::from)?;
+    let db_info = tokio::time::timeout(auth_config.webauth_confirmation_timeout, waiter)
+        .await
+        .map_err(|_elapsed| {
+            auth::AuthError::confirmation_timeout(auth_config.webauth_confirmation_timeout.into())
+        })?
+        .map_err(WebAuthError::from)?;
 
     if auth_config.ip_allowlist_check_enabled {
         if let Some(allowed_ips) = &db_info.allowed_ips {

proxy/src/bin/local_proxy.rs

@@ -77,10 +77,10 @@ struct LocalProxyCliArgs {
     #[clap(long, default_value = "127.0.0.1:5432")]
     compute: SocketAddr,
     /// Path of the local proxy config file
-    #[clap(long, default_value = "./localproxy.json")]
+    #[clap(long, default_value = "./local_proxy.json")]
     config_path: Utf8PathBuf,
     /// Path of the local proxy PID file
-    #[clap(long, default_value = "./localproxy.pid")]
+    #[clap(long, default_value = "./local_proxy.pid")]
     pid_path: Utf8PathBuf,
 }
 
@@ -109,7 +109,7 @@ struct SqlOverHttpArgs {
 
 #[tokio::main]
 async fn main() -> anyhow::Result<()> {
-    let _logging_guard = proxy::logging::init().await?;
+    let _logging_guard = proxy::logging::init_local_proxy()?;
     let _panic_hook_guard = utils::logging::replace_panic_hook_with_tracing_panic_hook();
     let _sentry_guard = init_sentry(Some(GIT_VERSION.into()), &[]);
 
@@ -138,7 +138,7 @@ async fn main() -> anyhow::Result<()> {
     // in order to trigger the appropriate SIGHUP on config change.
     //
     // This also claims a "lock" that makes sure only one instance
-    // of local-proxy runs at a time.
+    // of local_proxy runs at a time.
     let _process_guard = loop {
         match pid_file::claim_for_current_process(&args.pid_path) {
             Ok(guard) => break guard,
@@ -164,12 +164,6 @@ async fn main() -> anyhow::Result<()> {
         16,
     ));
 
-    // write the process ID to a file so that compute-ctl can find our process later
-    // in order to trigger the appropriate SIGHUP on config change.
-    let pid = std::process::id();
-    info!("process running in PID {pid}");
-    std::fs::write(args.pid_path, format!("{pid}\n")).context("writing PID to file")?;
-
     let mut maintenance_tasks = JoinSet::new();
 
     let refresh_config_notify = Arc::new(Notify::new());
@@ -182,9 +176,9 @@ async fn main() -> anyhow::Result<()> {
 
     // trigger the first config load **after** setting up the signal hook
     // to avoid the race condition where:
-    // 1. No config file registered when local-proxy starts up
+    // 1. No config file registered when local_proxy starts up
     // 2. The config file is written but the signal hook is not yet received
-    // 3. local-proxy completes startup but has no config loaded, despite there being a registerd config.
+    // 3. local_proxy completes startup but has no config loaded, despite there being a registerd config.
     refresh_config_notify.notify_one();
     tokio::spawn(refresh_config_loop(args.config_path, refresh_config_notify));
 
@@ -279,6 +273,7 @@ fn build_config(args: &LocalProxyCliArgs) -> anyhow::Result<&'static ProxyConfig
             ip_allowlist_check_enabled: true,
             is_auth_broker: false,
             accept_jwts: true,
+            webauth_confirmation_timeout: Duration::ZERO,
         },
         proxy_protocol_v2: config::ProxyProtocolV2::Rejected,
         handshake_timeout: Duration::from_secs(10),
@@ -310,7 +305,7 @@ async fn refresh_config_inner(path: &Utf8Path) -> anyhow::Result<()> {
 
     let mut jwks_set = vec![];
 
-    for jwks in data.jwks {
+    for jwks in data.jwks.into_iter().flatten() {
         let mut jwks_url = url::Url::from_str(&jwks.jwks_url).context("parsing JWKS url")?;
 
         ensure!(

proxy/src/bin/proxy.rs

@@ -236,6 +236,10 @@ struct ProxyCliArgs {
     // TODO(conradludgate): switch default to rejected or required once we've updated all deployments
     #[clap(value_enum, long, default_value_t = ProxyProtocolV2::Supported)]
     proxy_protocol_v2: ProxyProtocolV2,
+
+    /// Time the proxy waits for the webauth session to be confirmed by the control plane.
+    #[clap(long, default_value = "2m", value_parser = humantime::parse_duration)]
+    webauth_confirmation_timeout: std::time::Duration,
 }
 
 #[derive(clap::Args, Clone, Copy, Debug)]
@@ -719,6 +723,7 @@ fn build_config(args: &ProxyCliArgs) -> anyhow::Result<&'static ProxyConfig> {
         ip_allowlist_check_enabled: !args.is_private_access_proxy,
         is_auth_broker: args.is_auth_broker,
         accept_jwts: args.is_auth_broker,
+        webauth_confirmation_timeout: args.webauth_confirmation_timeout,
     };
 
     let config = Box::leak(Box::new(ProxyConfig {

proxy/src/compute.rs

@@ -20,7 +20,7 @@ use tokio_postgres::tls::MakeTlsConnect;
 use tokio_postgres_rustls::MakeRustlsConnect;
 use tracing::{error, info, warn};
 
-const COULD_NOT_CONNECT: &str = "Couldn't connect to compute node";
+pub const COULD_NOT_CONNECT: &str = "Couldn't connect to compute node";
 
 #[derive(Debug, Error)]
 pub(crate) enum ConnectionError {

proxy/src/config.rs

@@ -84,6 +84,7 @@ pub struct AuthenticationConfig {
     pub jwks_cache: JwkCache,
     pub is_auth_broker: bool,
     pub accept_jwts: bool,
+    pub webauth_confirmation_timeout: tokio::time::Duration,
 }
 
 impl TlsConfig {

proxy/src/lib.rs

@@ -90,6 +90,8 @@ use tokio::task::JoinError;
 use tokio_util::sync::CancellationToken;
 use tracing::warn;
 
+extern crate hyper0 as hyper;
+
 pub mod auth;
 pub mod cache;
 pub mod cancellation;

proxy/src/logging.rs

@@ -1,6 +1,13 @@
+use tracing::Subscriber;
 use tracing_subscriber::{
     filter::{EnvFilter, LevelFilter},
+    fmt::{
+        format::{Format, Full},
+        time::SystemTime,
+        FormatEvent, FormatFields,
+    },
     prelude::*,
+    registry::LookupSpan,
 };
 
 /// Initialize logging and OpenTelemetry tracing and exporter.
@@ -33,6 +40,45 @@ pub async fn init() -> anyhow::Result<LoggingGuard> {
     Ok(LoggingGuard)
 }
 
+/// Initialize logging for local_proxy with log prefix and no opentelemetry.
+///
+/// Logging can be configured using `RUST_LOG` environment variable.
+pub fn init_local_proxy() -> anyhow::Result<LoggingGuard> {
+    let env_filter = EnvFilter::builder()
+        .with_default_directive(LevelFilter::INFO.into())
+        .from_env_lossy();
+
+    let fmt_layer = tracing_subscriber::fmt::layer()
+        .with_ansi(false)
+        .with_writer(std::io::stderr)
+        .event_format(LocalProxyFormatter(Format::default().with_target(false)));
+
+    tracing_subscriber::registry()
+        .with(env_filter)
+        .with(fmt_layer)
+        .try_init()?;
+
+    Ok(LoggingGuard)
+}
+
+pub struct LocalProxyFormatter(Format<Full, SystemTime>);
+
+impl<S, N> FormatEvent<S, N> for LocalProxyFormatter
+where
+    S: Subscriber + for<'a> LookupSpan<'a>,
+    N: for<'a> FormatFields<'a> + 'static,
+{
+    fn format_event(
+        &self,
+        ctx: &tracing_subscriber::fmt::FmtContext<'_, S, N>,
+        mut writer: tracing_subscriber::fmt::format::Writer<'_>,
+        event: &tracing::Event<'_>,
+    ) -> std::fmt::Result {
+        writer.write_str("[local_proxy] ")?;
+        self.0.format_event(ctx, writer, event)
+    }
+}
+
 pub struct LoggingGuard;
 
 impl Drop for LoggingGuard {

proxy/src/proxy/connect_compute.rs

@@ -1,5 +1,6 @@
 use crate::{
     auth::backend::ComputeCredentialKeys,
+    compute::COULD_NOT_CONNECT,
     compute::{self, PostgresConnection},
     config::RetryConfig,
     console::{self, errors::WakeComputeError, locks::ApiLocks, CachedNodeInfo, NodeInfo},
@@ -15,7 +16,7 @@ use crate::{
 use async_trait::async_trait;
 use pq_proto::StartupMessageParams;
 use tokio::time;
-use tracing::{error, info, warn};
+use tracing::{debug, info, warn};
 
 use super::retry::ShouldRetryWakeCompute;
 
@@ -116,7 +117,6 @@ where
 
     node_info.set_keys(user_info.get_keys());
     node_info.allow_self_signed_compute = allow_self_signed_compute;
-    // let mut node_info = credentials.get_node_info(ctx, user_info).await?;
     mechanism.update_connect_config(&mut node_info.config);
     let retry_type = RetryType::ConnectToCompute;
 
@@ -139,10 +139,10 @@ where
         Err(e) => e,
     };
 
-    error!(error = ?err, "could not connect to compute node");
+    debug!(error = ?err, COULD_NOT_CONNECT);
 
     let node_info = if !node_info.cached() || !err.should_retry_wake_compute() {
-        // If we just recieved this from cplane and dodn't get it from cache, we shouldn't retry.
+        // If we just recieved this from cplane and didn't get it from cache, we shouldn't retry.
         // Do not need to retrieve a new node_info, just return the old one.
         if should_retry(&err, num_retries, connect_to_compute_retry_config) {
             Metrics::get().proxy.retries_metric.observe(
@@ -191,7 +191,7 @@ where
             }
             Err(e) => {
                 if !should_retry(&e, num_retries, connect_to_compute_retry_config) {
-                    error!(error = ?e, num_retries, retriable = false, "couldn't connect to compute node");
+                    // Don't log an error here, caller will print the error
                     Metrics::get().proxy.retries_metric.observe(
                         RetriesMetricGroup {
                             outcome: ConnectOutcome::Failed,
@@ -202,7 +202,7 @@ where
                     return Err(e.into());
                 }
 
-                warn!(error = ?e, num_retries, retriable = true, "couldn't connect to compute node");
+                warn!(error = ?e, num_retries, retriable = true, COULD_NOT_CONNECT);
             }
         };
 

pyproject.toml

@@ -6,10 +6,10 @@ package-mode = false
 [tool.poetry.dependencies]
 python = "^3.9"
 pytest = "^7.4.4"
-psycopg2-binary = "^2.9.6"
+psycopg2-binary = "^2.9.9"
 typing-extensions = "^4.6.1"
 PyJWT = {version = "^2.1.0", extras = ["crypto"]}
-requests = "^2.32.0"
+requests = "^2.32.3"
 pytest-xdist = "^3.3.1"
 asyncpg = "^0.29.0"
 aiopg = "^1.4.0"

safekeeper/Cargo.toml

@@ -23,7 +23,7 @@ crc32c.workspace = true
 fail.workspace = true
 hex.workspace = true
 humantime.workspace = true
-hyper.workspace = true
+hyper0.workspace = true
 futures.workspace = true
 once_cell.workspace = true
 parking_lot.workspace = true

safekeeper/src/debug_dump.rs

@@ -253,6 +253,13 @@ pub async fn build(args: Args) -> Result<Response> {
         });
     }
 
+    // Tokio forbids to drop runtime in async context, so this is a stupid way
+    // to drop it in non async context.
+    tokio::task::spawn_blocking(move || {
+        let _r = runtime;
+    })
+    .await?;
+
     Ok(Response {
         start_time,
         finish_time: Utc::now(),

safekeeper/src/lib.rs

@@ -1,4 +1,7 @@
 #![deny(clippy::undocumented_unsafe_blocks)]
+
+extern crate hyper0 as hyper;
+
 use camino::Utf8PathBuf;
 use once_cell::sync::Lazy;
 use remote_storage::RemoteStorageConfig;

safekeeper/src/remove_wal.rs

@@ -2,21 +2,29 @@ use utils::lsn::Lsn;
 
 use crate::timeline_manager::StateSnapshot;
 
-/// Get oldest LSN we still need to keep. We hold WAL till it is consumed
-/// by all of 1) pageserver (remote_consistent_lsn) 2) peers 3) s3
-/// offloading.
-/// While it is safe to use inmem values for determining horizon,
-/// we use persistent to make possible normal states less surprising.
-/// All segments covering LSNs before horizon_lsn can be removed.
+/// Get oldest LSN we still need to keep.
+///
+/// We hold WAL till it is consumed by
+/// 1) pageserver (remote_consistent_lsn)
+/// 2) s3 offloading.
+/// 3) Additionally we must store WAL since last local commit_lsn because
+///    that's where we start looking for last WAL record on start.
+///
+/// If some peer safekeeper misses data it will fetch it from the remote
+/// storage. While it is safe to use inmem values for determining horizon, we
+/// use persistent to make possible normal states less surprising. All segments
+/// covering LSNs before horizon_lsn can be removed.
 pub(crate) fn calc_horizon_lsn(state: &StateSnapshot, extra_horizon_lsn: Option<Lsn>) -> Lsn {
     use std::cmp::min;
 
-    let mut horizon_lsn = min(
-        state.cfile_remote_consistent_lsn,
-        state.cfile_peer_horizon_lsn,
-    );
+    let mut horizon_lsn = state.cfile_remote_consistent_lsn;
     // we don't want to remove WAL that is not yet offloaded to s3
     horizon_lsn = min(horizon_lsn, state.cfile_backup_lsn);
+    // Min by local commit_lsn to be able to begin reading WAL from somewhere on
+    // sk start. Technically we don't allow local commit_lsn to be higher than
+    // flush_lsn, but let's be double safe by including it as well.
+    horizon_lsn = min(horizon_lsn, state.cfile_commit_lsn);
+    horizon_lsn = min(horizon_lsn, state.flush_lsn);
     if let Some(extra_horizon_lsn) = extra_horizon_lsn {
         horizon_lsn = min(horizon_lsn, extra_horizon_lsn);
     }

safekeeper/src/timeline_manager.rs

@@ -47,7 +47,7 @@ pub(crate) struct StateSnapshot {
     pub(crate) remote_consistent_lsn: Lsn,
 
     // persistent control file values
-    pub(crate) cfile_peer_horizon_lsn: Lsn,
+    pub(crate) cfile_commit_lsn: Lsn,
     pub(crate) cfile_remote_consistent_lsn: Lsn,
     pub(crate) cfile_backup_lsn: Lsn,
 
@@ -70,7 +70,7 @@ impl StateSnapshot {
             commit_lsn: state.inmem.commit_lsn,
             backup_lsn: state.inmem.backup_lsn,
             remote_consistent_lsn: state.inmem.remote_consistent_lsn,
-            cfile_peer_horizon_lsn: state.peer_horizon_lsn,
+            cfile_commit_lsn: state.commit_lsn,
             cfile_remote_consistent_lsn: state.remote_consistent_lsn,
             cfile_backup_lsn: state.backup_lsn,
             flush_lsn: read_guard.sk.flush_lsn(),

safekeeper/tests/walproposer_sim/safekeeper.rs

@@ -13,7 +13,7 @@ use desim::{
     node_os::NodeOs,
     proto::{AnyMessage, NetEvent, NodeEvent},
 };
-use hyper::Uri;
+use hyper0::Uri;
 use safekeeper::{
     safekeeper::{ProposerAcceptorMessage, SafeKeeper, ServerInfo, UNKNOWN_SERVER_VERSION},
     state::{TimelinePersistentState, TimelineState},

storage_broker/Cargo.toml

@@ -16,7 +16,7 @@ futures.workspace = true
 futures-core.workspace = true
 futures-util.workspace = true
 humantime.workspace = true
-hyper = { workspace = true, features = ["full"] }
+hyper0 = { workspace = true, features = ["full"] }
 once_cell.workspace = true
 parking_lot.workspace = true
 prost.workspace = true

storage_broker/src/bin/storage_broker.rs

@@ -10,6 +10,9 @@
 //!
 //! Only safekeeper message is supported, but it is not hard to add something
 //! else with generics.
+
+extern crate hyper0 as hyper;
+
 use clap::{command, Parser};
 use futures_core::Stream;
 use futures_util::StreamExt;

storage_broker/src/lib.rs

@@ -1,3 +1,5 @@
+extern crate hyper0 as hyper;
+
 use hyper::body::HttpBody;
 use std::pin::Pin;
 use std::task::{Context, Poll};

storage_controller/Cargo.toml

@@ -21,7 +21,7 @@ clap.workspace = true
 fail.workspace = true
 futures.workspace = true
 hex.workspace = true
-hyper.workspace = true
+hyper0.workspace = true
 humantime.workspace = true
 itertools.workspace = true
 lasso.workspace = true

storage_controller/src/lib.rs

@@ -1,6 +1,8 @@
 use serde::Serialize;
 use utils::seqwait::MonotonicCounter;
 
+extern crate hyper0 as hyper;
+
 mod auth;
 mod background_node_operations;
 mod compute_hook;

storage_controller/src/main.rs

@@ -1,6 +1,6 @@
 use anyhow::{anyhow, Context};
 use clap::Parser;
-use hyper::Uri;
+use hyper0::Uri;
 use metrics::launch_timestamp::LaunchTimestamp;
 use metrics::BuildInfo;
 use std::path::PathBuf;
@@ -11,8 +11,8 @@ use storage_controller::metrics::preinitialize_metrics;
 use storage_controller::persistence::Persistence;
 use storage_controller::service::chaos_injector::ChaosInjector;
 use storage_controller::service::{
-    Config, Service, HEARTBEAT_INTERVAL_DEFAULT, MAX_OFFLINE_INTERVAL_DEFAULT,
-    MAX_WARMING_UP_INTERVAL_DEFAULT, RECONCILER_CONCURRENCY_DEFAULT,
+    Config, Service, HEARTBEAT_INTERVAL_DEFAULT, LONG_RECONCILE_THRESHOLD_DEFAULT,
+    MAX_OFFLINE_INTERVAL_DEFAULT, MAX_WARMING_UP_INTERVAL_DEFAULT, RECONCILER_CONCURRENCY_DEFAULT,
 };
 use tokio::signal::unix::SignalKind;
 use tokio_util::sync::CancellationToken;
@@ -108,6 +108,9 @@ struct Cli {
     // Period with which to send heartbeats to registered nodes
     #[arg(long)]
     heartbeat_interval: Option<humantime::Duration>,
+
+    #[arg(long)]
+    long_reconcile_threshold: Option<humantime::Duration>,
 }
 
 enum StrictMode {
@@ -293,6 +296,10 @@ async fn async_main() -> anyhow::Result<()> {
             .heartbeat_interval
             .map(humantime::Duration::into)
             .unwrap_or(HEARTBEAT_INTERVAL_DEFAULT),
+        long_reconcile_threshold: args
+            .long_reconcile_threshold
+            .map(humantime::Duration::into)
+            .unwrap_or(LONG_RECONCILE_THRESHOLD_DEFAULT),
         address_for_peers: args.address_for_peers,
         start_as_candidate: args.start_as_candidate,
         http_service_port: args.listen.port() as i32,
@@ -317,7 +324,7 @@ async fn async_main() -> anyhow::Result<()> {
 
     // Start HTTP server
     let server_shutdown = CancellationToken::new();
-    let server = hyper::Server::from_tcp(http_listener)?
+    let server = hyper0::Server::from_tcp(http_listener)?
         .serve(router_service)
         .with_graceful_shutdown({
             let server_shutdown = server_shutdown.clone();

storage_controller/src/metrics.rs

@@ -87,6 +87,10 @@ pub(crate) struct StorageControllerMetricGroup {
         measured::HistogramVec<DatabaseQueryLatencyLabelGroupSet, 5>,
 
     pub(crate) storage_controller_leadership_status: measured::GaugeVec<LeadershipStatusGroupSet>,
+
+    /// HTTP request status counters for handled requests
+    pub(crate) storage_controller_reconcile_long_running:
+        measured::CounterVec<ReconcileLongRunningLabelGroupSet>,
 }
 
 impl StorageControllerMetrics {
@@ -168,6 +172,17 @@ pub(crate) struct LeadershipStatusGroup {
     pub(crate) status: LeadershipStatus,
 }
 
+#[derive(measured::LabelGroup, Clone)]
+#[label(set = ReconcileLongRunningLabelGroupSet)]
+pub(crate) struct ReconcileLongRunningLabelGroup<'a> {
+    #[label(dynamic_with = lasso::ThreadedRodeo, default)]
+    pub(crate) tenant_id: &'a str,
+    #[label(dynamic_with = lasso::ThreadedRodeo, default)]
+    pub(crate) shard_number: &'a str,
+    #[label(dynamic_with = lasso::ThreadedRodeo, default)]
+    pub(crate) sequence: &'a str,
+}
+
 #[derive(FixedCardinalityLabel, Clone, Copy)]
 pub(crate) enum ReconcileOutcome {
     #[label(rename = "ok")]

storage_controller/src/reconciler.rs

@@ -454,7 +454,7 @@ impl Reconciler {
                 Ok(l) => l,
                 Err(e) => {
                     tracing::info!("🕑 Can't get LSNs on node {node} yet, waiting ({e})",);
-                    std::thread::sleep(Duration::from_millis(500));
+                    tokio::time::sleep(Duration::from_millis(500)).await;
                     continue;
                 }
             };
@@ -469,10 +469,7 @@ impl Reconciler {
                         }
                     }
                     None => {
-                        // Expected timeline isn't yet visible on migration destination.
-                        // (IRL we would have to account for timeline deletion, but this
-                        //  is just test helper)
-                        any_behind = true;
+                        // Timeline was deleted in the meantime - ignore it
                     }
                 }
             }
@@ -481,7 +478,7 @@ impl Reconciler {
                 tracing::info!("✅ LSN caught up.  Proceeding...");
                 break;
             } else {
-                std::thread::sleep(Duration::from_millis(500));
+                tokio::time::sleep(Duration::from_millis(500)).await;
             }
         }
 
@@ -562,6 +559,8 @@ impl Reconciler {
         self.location_config(&dest_ps, dest_conf, None, false)
             .await?;
 
+        pausable_failpoint!("reconciler-live-migrate-pre-await-lsn");
+
         if let Some(baseline) = baseline_lsns {
             tracing::info!("🕑 Waiting for LSN to catch up...");
             self.await_lsn(self.tenant_shard_id, &dest_ps, baseline)

storage_controller/src/service.rs

@@ -128,6 +128,9 @@ pub const MAX_WARMING_UP_INTERVAL_DEFAULT: Duration = Duration::from_secs(300);
 /// How often to send heartbeats to registered nodes?
 pub const HEARTBEAT_INTERVAL_DEFAULT: Duration = Duration::from_secs(5);
 
+/// How long is too long for a reconciliation?
+pub const LONG_RECONCILE_THRESHOLD_DEFAULT: Duration = Duration::from_secs(120);
+
 #[derive(Clone, strum_macros::Display)]
 enum TenantOperations {
     Create,
@@ -348,6 +351,8 @@ pub struct Config {
     pub start_as_candidate: bool,
 
     pub http_service_port: i32,
+
+    pub long_reconcile_threshold: Duration,
 }
 
 impl From<DatabaseError> for ApiError {
@@ -521,6 +526,21 @@ pub(crate) enum ReconcileResultRequest {
     Stop,
 }
 
+#[derive(Clone)]
+struct MutationLocation {
+    node: Node,
+    generation: Generation,
+}
+
+#[derive(Clone)]
+struct ShardMutationLocations {
+    latest: MutationLocation,
+    other: Vec<MutationLocation>,
+}
+
+#[derive(Default, Clone)]
+struct TenantMutationLocations(BTreeMap<TenantShardId, ShardMutationLocations>);
+
 impl Service {
     pub fn get_config(&self) -> &Config {
         &self.config
@@ -2982,38 +3002,83 @@ impl Service {
         failpoint_support::sleep_millis_async!("tenant-create-timeline-shared-lock");
 
         self.tenant_remote_mutation(tenant_id, move |mut targets| async move {
-            if targets.is_empty() {
+            if targets.0.is_empty() {
                 return Err(ApiError::NotFound(
                     anyhow::anyhow!("Tenant not found").into(),
                 ));
             };
-            let shard_zero = targets.remove(0);
+
+            let (shard_zero_tid, shard_zero_locations) =
+                targets.0.pop_first().expect("Must have at least one shard");
+            assert!(shard_zero_tid.is_shard_zero());
 
             async fn create_one(
                 tenant_shard_id: TenantShardId,
-                node: Node,
+                locations: ShardMutationLocations,
                 jwt: Option<String>,
                 create_req: TimelineCreateRequest,
             ) -> Result<TimelineInfo, ApiError> {
+                let latest = locations.latest.node;
+
                 tracing::info!(
-                    "Creating timeline on shard {}/{}, attached to node {node}",
+                    "Creating timeline on shard {}/{}, attached to node {latest} in generation {:?}",
                     tenant_shard_id,
                     create_req.new_timeline_id,
+                    locations.latest.generation
                 );
-                let client = PageserverClient::new(node.get_id(), node.base_url(), jwt.as_deref());
 
-                client
+                let client =
+                    PageserverClient::new(latest.get_id(), latest.base_url(), jwt.as_deref());
+
+                let timeline_info = client
                     .timeline_create(tenant_shard_id, &create_req)
                     .await
-                    .map_err(|e| passthrough_api_error(&node, e))
+                    .map_err(|e| passthrough_api_error(&latest, e))?;
+
+                // We propagate timeline creations to all attached locations such that a compute
+                // for the new timeline is able to start regardless of the current state of the
+                // tenant shard reconciliation.
+                for location in locations.other {
+                    tracing::info!(
+                        "Creating timeline on shard {}/{}, stale attached to node {} in generation {:?}",
+                        tenant_shard_id,
+                        create_req.new_timeline_id,
+                        location.node,
+                        location.generation
+                    );
+
+                    let client = PageserverClient::new(
+                        location.node.get_id(),
+                        location.node.base_url(),
+                        jwt.as_deref(),
+                    );
+
+                    let res = client
+                        .timeline_create(tenant_shard_id, &create_req)
+                        .await;
+
+                    if let Err(e) = res {
+                        match e {
+                            mgmt_api::Error::ApiError(StatusCode::NOT_FOUND, _) => {
+                                // Tenant might have been detached from the stale location,
+                                // so ignore 404s.
+                            },
+                            _ => {
+                                return Err(passthrough_api_error(&location.node, e));
+                            }
+                        }
+                    }
+                }
+
+                Ok(timeline_info)
             }
 
             // Because the caller might not provide an explicit LSN, we must do the creation first on a single shard, and then
             // use whatever LSN that shard picked when creating on subsequent shards.  We arbitrarily use shard zero as the shard
             // that will get the first creation request, and propagate the LSN to all the >0 shards.
             let timeline_info = create_one(
-                shard_zero.0,
-                shard_zero.1,
+                shard_zero_tid,
+                shard_zero_locations,
                 self.config.jwt_token.clone(),
                 create_req.clone(),
             )
@@ -3026,14 +3091,24 @@ impl Service {
             }
 
             // Create timeline on remaining shards with number >0
-            if !targets.is_empty() {
+            if !targets.0.is_empty() {
                 // If we had multiple shards, issue requests for the remainder now.
                 let jwt = &self.config.jwt_token;
                 self.tenant_for_shards(
-                    targets.iter().map(|t| (t.0, t.1.clone())).collect(),
-                    |tenant_shard_id: TenantShardId, node: Node| {
+                    targets
+                        .0
+                        .iter()
+                        .map(|t| (*t.0, t.1.latest.node.clone()))
+                        .collect(),
+                    |tenant_shard_id: TenantShardId, _node: Node| {
                         let create_req = create_req.clone();
-                        Box::pin(create_one(tenant_shard_id, node, jwt.clone(), create_req))
+                        let mutation_locations = targets.0.remove(&tenant_shard_id).unwrap();
+                        Box::pin(create_one(
+                            tenant_shard_id,
+                            mutation_locations,
+                            jwt.clone(),
+                            create_req,
+                        ))
                     },
                 )
                 .await?;
@@ -3063,7 +3138,7 @@ impl Service {
         .await;
 
         self.tenant_remote_mutation(tenant_id, move |targets| async move {
-            if targets.is_empty() {
+            if targets.0.is_empty() {
                 return Err(ApiError::NotFound(
                     anyhow::anyhow!("Tenant not found").into(),
                 ));
@@ -3094,8 +3169,9 @@ impl Service {
 
             // no shard needs to go first/last; the operation should be idempotent
             // TODO: it would be great to ensure that all shards return the same error
+            let locations = targets.0.iter().map(|t| (*t.0, t.1.latest.node.clone())).collect();
             let results = self
-                .tenant_for_shards(targets, |tenant_shard_id, node| {
+                .tenant_for_shards(locations, |tenant_shard_id, node| {
                     futures::FutureExt::boxed(config_one(
                         tenant_shard_id,
                         timeline_id,
@@ -3126,7 +3202,7 @@ impl Service {
         .await;
 
         self.tenant_remote_mutation(tenant_id, move |targets| async move {
-            if targets.is_empty() {
+            if targets.0.is_empty() {
                 return Err(ApiError::NotFound(
                     anyhow::anyhow!("Tenant not found").into(),
                 ));
@@ -3174,8 +3250,9 @@ impl Service {
             }
 
             // no shard needs to go first/last; the operation should be idempotent
+            let locations = targets.0.iter().map(|t| (*t.0, t.1.latest.node.clone())).collect();
             let mut results = self
-                .tenant_for_shards(targets, |tenant_shard_id, node| {
+                .tenant_for_shards(locations, |tenant_shard_id, node| {
                     futures::FutureExt::boxed(detach_one(
                         tenant_shard_id,
                         timeline_id,
@@ -3222,7 +3299,7 @@ impl Service {
         .await;
 
         self.tenant_remote_mutation(tenant_id, move |targets| async move {
-            if targets.is_empty() {
+            if targets.0.is_empty() {
                 return Err(ApiError::NotFound(
                     anyhow::anyhow!("Tenant not found").into(),
                 ));
@@ -3244,7 +3321,12 @@ impl Service {
             }
 
             // no shard needs to go first/last; the operation should be idempotent
-            self.tenant_for_shards(targets, |tenant_shard_id, node| {
+            let locations = targets
+                .0
+                .iter()
+                .map(|t| (*t.0, t.1.latest.node.clone()))
+                .collect();
+            self.tenant_for_shards(locations, |tenant_shard_id, node| {
                 futures::FutureExt::boxed(do_one(
                     tenant_shard_id,
                     timeline_id,
@@ -3339,11 +3421,11 @@ impl Service {
         op: O,
     ) -> Result<R, ApiError>
     where
-        O: FnOnce(Vec<(TenantShardId, Node)>) -> F,
+        O: FnOnce(TenantMutationLocations) -> F,
         F: std::future::Future<Output = R>,
     {
-        let target_gens = {
-            let mut targets = Vec::new();
+        let mutation_locations = {
+            let mut locations = TenantMutationLocations::default();
 
             // Load the currently attached pageservers for the latest generation of each shard.  This can
             // run concurrently with reconciliations, and it is not guaranteed that the node we find here
@@ -3394,14 +3476,50 @@ impl Service {
                     .ok_or(ApiError::Conflict(format!(
                         "Raced with removal of node {node_id}"
                     )))?;
-                targets.push((tenant_shard_id, node.clone(), generation));
+                let generation = generation.expect("Checked above");
+
+                let tenant = locked.tenants.get(&tenant_shard_id);
+
+                // TODO(vlad): Abstract the logic that finds stale attached locations
+                // from observed state into a [`Service`] method.
+                let other_locations = match tenant {
+                    Some(tenant) => {
+                        let mut other = tenant.attached_locations();
+                        let latest_location_index =
+                            other.iter().position(|&l| l == (node.get_id(), generation));
+                        if let Some(idx) = latest_location_index {
+                            other.remove(idx);
+                        }
+
+                        other
+                    }
+                    None => Vec::default(),
+                };
+
+                let location = ShardMutationLocations {
+                    latest: MutationLocation {
+                        node: node.clone(),
+                        generation,
+                    },
+                    other: other_locations
+                        .into_iter()
+                        .filter_map(|(node_id, generation)| {
+                            let node = locked.nodes.get(&node_id)?;
+
+                            Some(MutationLocation {
+                                node: node.clone(),
+                                generation,
+                            })
+                        })
+                        .collect(),
+                };
+                locations.0.insert(tenant_shard_id, location);
             }
 
-            targets
+            locations
         };
 
-        let targets = target_gens.iter().map(|t| (t.0, t.1.clone())).collect();
-        let result = op(targets).await;
+        let result = op(mutation_locations.clone()).await;
 
         // Post-check: are all the generations of all the shards the same as they were initially?  This proves that
         // our remote operation executed on the latest generation and is therefore persistent.
@@ -3417,9 +3535,10 @@ impl Service {
                      }| (tenant_shard_id, generation),
                 )
                 .collect::<Vec<_>>()
-                != target_gens
+                != mutation_locations
+                    .0
                     .into_iter()
-                    .map(|i| (i.0, i.2))
+                    .map(|i| (i.0, Some(i.1.latest.generation)))
                     .collect::<Vec<_>>()
             {
                 // We raced with something that incremented the generation, and therefore cannot be
@@ -3449,12 +3568,14 @@ impl Service {
         .await;
 
         self.tenant_remote_mutation(tenant_id, move |mut targets| async move {
-            if targets.is_empty() {
+            if targets.0.is_empty() {
                 return Err(ApiError::NotFound(
                     anyhow::anyhow!("Tenant not found").into(),
                 ));
             }
-            let shard_zero = targets.remove(0);
+
+            let (shard_zero_tid, shard_zero_locations) = targets.0.pop_first().expect("Must have at least one shard");
+            assert!(shard_zero_tid.is_shard_zero());
 
             async fn delete_one(
                 tenant_shard_id: TenantShardId,
@@ -3477,8 +3598,9 @@ impl Service {
                     })
             }
 
+            let locations = targets.0.iter().map(|t| (*t.0, t.1.latest.node.clone())).collect();
             let statuses = self
-                .tenant_for_shards(targets, |tenant_shard_id: TenantShardId, node: Node| {
+                .tenant_for_shards(locations, |tenant_shard_id: TenantShardId, node: Node| {
                     Box::pin(delete_one(
                         tenant_shard_id,
                         timeline_id,
@@ -3496,9 +3618,9 @@ impl Service {
             // Delete shard zero last: this is not strictly necessary, but since a caller's GET on a timeline will be routed
             // to shard zero, it gives a more obvious behavior that a GET returns 404 once the deletion is done.
             let shard_zero_status = delete_one(
-                shard_zero.0,
+                shard_zero_tid,
                 timeline_id,
-                shard_zero.1,
+                shard_zero_locations.latest.node,
                 self.config.jwt_token.clone(),
             )
             .await?;

storage_controller/src/tenant_shard.rs

@@ -5,7 +5,9 @@ use std::{
 };
 
 use crate::{
-    metrics::{self, ReconcileCompleteLabelGroup, ReconcileOutcome},
+    metrics::{
+        self, ReconcileCompleteLabelGroup, ReconcileLongRunningLabelGroup, ReconcileOutcome,
+    },
     persistence::TenantShardPersistence,
     reconciler::{ReconcileUnits, ReconcilerConfig},
     scheduler::{
@@ -14,6 +16,8 @@ use crate::{
     },
     service::ReconcileResultRequest,
 };
+use futures::future::{self, Either};
+use itertools::Itertools;
 use pageserver_api::controller_api::{
     AvailabilityZone, NodeSchedulingPolicy, PlacementPolicy, ShardSchedulingPolicy,
 };
@@ -1083,6 +1087,47 @@ impl TenantShard {
         }
     }
 
+    async fn reconcile(
+        sequence: Sequence,
+        mut reconciler: Reconciler,
+        must_notify: bool,
+    ) -> ReconcileResult {
+        // Attempt to make observed state match intent state
+        let result = reconciler.reconcile().await;
+
+        // If we know we had a pending compute notification from some previous action, send a notification irrespective
+        // of whether the above reconcile() did any work
+        if result.is_ok() && must_notify {
+            // If this fails we will send the need to retry in [`ReconcileResult::pending_compute_notification`]
+            reconciler.compute_notify().await.ok();
+        }
+
+        // Update result counter
+        let outcome_label = match &result {
+            Ok(_) => ReconcileOutcome::Success,
+            Err(ReconcileError::Cancel) => ReconcileOutcome::Cancel,
+            Err(_) => ReconcileOutcome::Error,
+        };
+
+        metrics::METRICS_REGISTRY
+            .metrics_group
+            .storage_controller_reconcile_complete
+            .inc(ReconcileCompleteLabelGroup {
+                status: outcome_label,
+            });
+
+        // Constructing result implicitly drops Reconciler, freeing any ReconcileUnits before the Service might
+        // try and schedule more work in response to our result.
+        ReconcileResult {
+            sequence,
+            result,
+            tenant_shard_id: reconciler.tenant_shard_id,
+            generation: reconciler.generation,
+            observed: reconciler.observed,
+            pending_compute_notification: reconciler.compute_notify_failure,
+        }
+    }
+
     #[allow(clippy::too_many_arguments)]
     #[instrument(skip_all, fields(tenant_id=%self.tenant_shard_id.tenant_id, shard_id=%self.tenant_shard_id.shard_slug()))]
     pub(crate) fn spawn_reconciler(
@@ -1122,7 +1167,7 @@ impl TenantShard {
 
         let reconciler_cancel = cancel.child_token();
         let reconciler_intent = TargetState::from_intent(pageservers, &self.intent);
-        let mut reconciler = Reconciler {
+        let reconciler = Reconciler {
             tenant_shard_id: self.tenant_shard_id,
             shard: self.shard,
             placement_policy: self.policy.clone(),
@@ -1142,6 +1187,7 @@ impl TenantShard {
         };
 
         let reconcile_seq = self.sequence;
+        let long_reconcile_threshold = service_config.long_reconcile_threshold;
 
         tracing::info!(seq=%reconcile_seq, "Spawning Reconciler for sequence {}", self.sequence);
         let must_notify = self.pending_compute_notification;
@@ -1178,41 +1224,55 @@ impl TenantShard {
                     return;
                 }
 
-                // Attempt to make observed state match intent state
-                let result = reconciler.reconcile().await;
+                let (tenant_id_label, shard_number_label, sequence_label) = {
+                    (
+                        reconciler.tenant_shard_id.tenant_id.to_string(),
+                        reconciler.tenant_shard_id.shard_number.0.to_string(),
+                        reconcile_seq.to_string(),
+                    )
+                };
 
-                // If we know we had a pending compute notification from some previous action, send a notification irrespective
-                // of whether the above reconcile() did any work
-                if result.is_ok() && must_notify {
-                    // If this fails we will send the need to retry in [`ReconcileResult::pending_compute_notification`]
-                    reconciler.compute_notify().await.ok();
+                let label_group = ReconcileLongRunningLabelGroup {
+                    tenant_id: &tenant_id_label,
+                    shard_number: &shard_number_label,
+                    sequence: &sequence_label,
+                };
+
+                let reconcile_fut = Self::reconcile(reconcile_seq, reconciler, must_notify);
+                let long_reconcile_fut = {
+                    let label_group = label_group.clone();
+                    async move {
+                        tokio::time::sleep(long_reconcile_threshold).await;
+
+                        tracing::warn!("Reconcile passed the long running threshold of {long_reconcile_threshold:?}");
+
+                        metrics::METRICS_REGISTRY
+                            .metrics_group
+                            .storage_controller_reconcile_long_running
+                            .inc(label_group);
+                    }
+                };
+
+                let reconcile_fut = std::pin::pin!(reconcile_fut);
+                let long_reconcile_fut = std::pin::pin!(long_reconcile_fut);
+
+                let (was_long, result) =
+                    match future::select(reconcile_fut, long_reconcile_fut).await {
+                        Either::Left((reconcile_result, _)) => (false, reconcile_result),
+                        Either::Right((_, reconcile_fut)) => (true, reconcile_fut.await),
+                    };
+
+                if was_long {
+                    let id = metrics::METRICS_REGISTRY
+                        .metrics_group
+                        .storage_controller_reconcile_long_running
+                        .with_labels(label_group);
+                    metrics::METRICS_REGISTRY
+                        .metrics_group
+                        .storage_controller_reconcile_long_running
+                        .remove_metric(id);
                 }
 
-                // Update result counter
-                let outcome_label = match &result {
-                    Ok(_) => ReconcileOutcome::Success,
-                    Err(ReconcileError::Cancel) => ReconcileOutcome::Cancel,
-                    Err(_) => ReconcileOutcome::Error,
-                };
-
-                metrics::METRICS_REGISTRY
-                    .metrics_group
-                    .storage_controller_reconcile_complete
-                    .inc(ReconcileCompleteLabelGroup {
-                        status: outcome_label,
-                    });
-
-                // Constructing result implicitly drops Reconciler, freeing any ReconcileUnits before the Service might
-                // try and schedule more work in response to our result.
-                let result = ReconcileResult {
-                    sequence: reconcile_seq,
-                    result,
-                    tenant_shard_id: reconciler.tenant_shard_id,
-                    generation: reconciler.generation,
-                    observed: reconciler.observed,
-                    pending_compute_notification: reconciler.compute_notify_failure,
-                };
-
                 result_tx
                     .send(ReconcileResultRequest::ReconcileResult(result))
                     .ok();
@@ -1351,6 +1411,32 @@ impl TenantShard {
     pub(crate) fn set_preferred_az(&mut self, preferred_az_id: AvailabilityZone) {
         self.preferred_az_id = Some(preferred_az_id);
     }
+
+    /// Returns all the nodes to which this tenant shard is attached according to the
+    /// observed state and the generations. Return vector is sorted from latest generation
+    /// to earliest.
+    pub(crate) fn attached_locations(&self) -> Vec<(NodeId, Generation)> {
+        self.observed
+            .locations
+            .iter()
+            .filter_map(|(node_id, observed)| {
+                use LocationConfigMode::{AttachedMulti, AttachedSingle, AttachedStale};
+
+                let conf = observed.conf.as_ref()?;
+
+                match (conf.generation, conf.mode) {
+                    (Some(gen), AttachedMulti | AttachedSingle | AttachedStale) => {
+                        Some((*node_id, gen))
+                    }
+                    _ => None,
+                }
+            })
+            .sorted_by(|(_lhs_node_id, lhs_gen), (_rhs_node_id, rhs_gen)| {
+                lhs_gen.cmp(rhs_gen).reverse()
+            })
+            .map(|(node_id, gen)| (node_id, Generation::new(gen)))
+            .collect()
+    }
 }
 
 #[cfg(test)]

storage_scrubber/Cargo.toml

@@ -5,6 +5,7 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
+aws-config.workspace = true
 aws-sdk-s3.workspace = true
 either.workspace = true
 anyhow.workspace = true
@@ -31,7 +32,6 @@ storage_controller_client.workspace = true
 tokio = { workspace = true, features = ["macros", "rt-multi-thread"] }
 chrono = { workspace = true, default-features = false, features = ["clock", "serde"] }
 reqwest = { workspace = true, default-features = false, features = ["rustls-tls", "json"] }
-aws-config = { workspace = true, default-features = false, features = ["rustls", "sso"] }
 
 pageserver = { path = "../pageserver" }
 pageserver_api = { path = "../libs/pageserver_api" }

storage_scrubber/src/lib.rs

@@ -28,8 +28,9 @@ use pageserver::tenant::remote_timeline_client::{remote_tenant_path, remote_time
 use pageserver::tenant::TENANTS_SEGMENT_NAME;
 use pageserver_api::shard::TenantShardId;
 use remote_storage::{
-    GenericRemoteStorage, Listing, ListingMode, RemotePath, RemoteStorageConfig, RemoteStorageKind,
-    S3Config, DEFAULT_MAX_KEYS_PER_LIST_RESPONSE, DEFAULT_REMOTE_STORAGE_S3_CONCURRENCY_LIMIT,
+    DownloadOpts, GenericRemoteStorage, Listing, ListingMode, RemotePath, RemoteStorageConfig,
+    RemoteStorageKind, S3Config, DEFAULT_MAX_KEYS_PER_LIST_RESPONSE,
+    DEFAULT_REMOTE_STORAGE_S3_CONCURRENCY_LIMIT,
 };
 use reqwest::Url;
 use serde::{Deserialize, Serialize};
@@ -488,7 +489,10 @@ async fn download_object_with_retries(
     let cancel = CancellationToken::new();
     for trial in 0..MAX_RETRIES {
         let mut buf = Vec::new();
-        let download = match remote_client.download(key, &cancel).await {
+        let download = match remote_client
+            .download(key, &DownloadOpts::default(), &cancel)
+            .await
+        {
             Ok(response) => response,
             Err(e) => {
                 error!("Failed to download object for key {key}: {e}");

test_runner/fixtures/compare_fixtures.py

@@ -340,23 +340,27 @@ def neon_with_baseline(request: FixtureRequest) -> PgCompare:
 
 
 @pytest.fixture(scope="function", autouse=True)
-def sync_after_each_test():
-    # The fixture calls `sync(2)` after each test if `SYNC_AFTER_EACH_TEST` env var is `true`
+def sync_between_tests():
+    # The fixture calls `sync(2)` after each test if `SYNC_BETWEEN_TESTS` env var is `true`
     #
-    # In CI, `SYNC_AFTER_EACH_TEST` is set to `true` only for benchmarks (`test_runner/performance`)
+    # In CI, `SYNC_BETWEEN_TESTS` is set to `true` only for benchmarks (`test_runner/performance`)
     # that are run on self-hosted runners because some of these tests are pretty write-heavy
     # and create issues to start the processes within 10s
-    key = "SYNC_AFTER_EACH_TEST"
+    key = "SYNC_BETWEEN_TESTS"
     enabled = os.environ.get(key) == "true"
 
+    if enabled:
+        start = time.time()
+        # we only run benches on unices, the method might not exist on windows
+        os.sync()
+        elapsed = time.time() - start
+        log.info(f"called sync before test {elapsed=}")
+
     yield
 
-    if not enabled:
-        # regress test, or running locally
-        return
-
-    start = time.time()
-    # we only run benches on unices, the method might not exist on windows
-    os.sync()
-    elapsed = time.time() - start
-    log.info(f"called sync after test {elapsed=}")
+    if enabled:
+        start = time.time()
+        # we only run benches on unices, the method might not exist on windows
+        os.sync()
+        elapsed = time.time() - start
+        log.info(f"called sync after test {elapsed=}")

test_runner/fixtures/neon_cli.py

@@ -0,0 +1,662 @@
+from __future__ import annotations
+
+import abc
+import json
+import os
+import re
+import subprocess
+import tempfile
+import textwrap
+from itertools import chain, product
+from pathlib import Path
+from typing import (
+    Any,
+    Dict,
+    List,
+    Optional,
+    Tuple,
+    TypeVar,
+    cast,
+)
+
+import toml
+
+from fixtures.common_types import Lsn, TenantId, TimelineId
+from fixtures.log_helper import log
+from fixtures.pageserver.common_types import IndexPartDump
+from fixtures.pg_version import PgVersion
+from fixtures.utils import AuxFileStore
+
+T = TypeVar("T")
+
+
+class AbstractNeonCli(abc.ABC):
+    """
+    A typed wrapper around an arbitrary Neon CLI tool.
+    Supports a way to run arbitrary command directly via CLI.
+    Do not use directly, use specific subclasses instead.
+    """
+
+    def __init__(self, extra_env: Optional[Dict[str, str]], binpath: Path):
+        self.extra_env = extra_env
+        self.binpath = binpath
+
+    COMMAND: str = cast(str, None)  # To be overwritten by the derived class.
+
+    def raw_cli(
+        self,
+        arguments: List[str],
+        extra_env_vars: Optional[Dict[str, str]] = None,
+        check_return_code=True,
+        timeout=None,
+    ) -> "subprocess.CompletedProcess[str]":
+        """
+        Run the command with the specified arguments.
+
+        Arguments must be in list form, e.g. ['endpoint', 'create']
+
+        Return both stdout and stderr, which can be accessed as
+
+        >>> result = env.neon_cli.raw_cli(...)
+        >>> assert result.stderr == ""
+        >>> log.info(result.stdout)
+
+        If `check_return_code`, on non-zero exit code logs failure and raises.
+        """
+
+        assert isinstance(arguments, list)
+        assert isinstance(self.COMMAND, str)
+
+        command_path = str(self.binpath / self.COMMAND)
+
+        args = [command_path] + arguments
+        log.info('Running command "{}"'.format(" ".join(args)))
+
+        env_vars = os.environ.copy()
+
+        # extra env
+        for extra_env_key, extra_env_value in (self.extra_env or {}).items():
+            env_vars[extra_env_key] = extra_env_value
+        for extra_env_key, extra_env_value in (extra_env_vars or {}).items():
+            env_vars[extra_env_key] = extra_env_value
+
+        # Pass through coverage settings
+        var = "LLVM_PROFILE_FILE"
+        val = os.environ.get(var)
+        if val:
+            env_vars[var] = val
+
+        # Intercept CalledProcessError and print more info
+        try:
+            res = subprocess.run(
+                args,
+                env=env_vars,
+                check=False,
+                universal_newlines=True,
+                stdout=subprocess.PIPE,
+                stderr=subprocess.PIPE,
+                timeout=timeout,
+            )
+        except subprocess.TimeoutExpired as e:
+            if e.stderr:
+                stderr = e.stderr.decode(errors="replace")
+            else:
+                stderr = ""
+
+            if e.stdout:
+                stdout = e.stdout.decode(errors="replace")
+            else:
+                stdout = ""
+
+            log.warn(f"CLI timeout: stderr={stderr}, stdout={stdout}")
+            raise
+
+        indent = "  "
+        if not res.returncode:
+            stripped = res.stdout.strip()
+            lines = stripped.splitlines()
+            if len(lines) < 2:
+                log.debug(f"Run {res.args} success: {stripped}")
+            else:
+                log.debug("Run %s success:\n%s" % (res.args, textwrap.indent(stripped, indent)))
+        elif check_return_code:
+            # this way command output will be in recorded and shown in CI in failure message
+            indent = indent * 2
+            msg = textwrap.dedent(
+                """\
+            Run %s failed:
+              stdout:
+            %s
+              stderr:
+            %s
+            """
+            )
+            msg = msg % (
+                res.args,
+                textwrap.indent(res.stdout.strip(), indent),
+                textwrap.indent(res.stderr.strip(), indent),
+            )
+            log.info(msg)
+            raise RuntimeError(msg) from subprocess.CalledProcessError(
+                res.returncode, res.args, res.stdout, res.stderr
+            )
+        return res
+
+
+class NeonLocalCli(AbstractNeonCli):
+    """A typed wrapper around the `neon_local` CLI tool.
+    Supports main commands via typed methods and a way to run arbitrary command directly via CLI.
+
+    Note: The methods in this class are supposed to be faithful wrappers of the underlying
+    'neon_local' commands. If you're tempted to add any logic here, please consider putting it
+    in the caller instead!
+
+    There are a few exceptions where these wrapper methods intentionally differ from the
+    underlying commands, however:
+    - Many 'neon_local' commands take an optional 'tenant_id' argument and use the default from
+      the config file if it's omitted. The corresponding wrappers require an explicit 'tenant_id'
+      argument. The idea is that we don't want to rely on the config file's default in tests,
+      because NeonEnv has its own 'initial_tenant'. They are currently always the same, but we
+      want to rely on the Neonenv's default instead of the config file default in tests.
+
+    - Similarly, --pg_version argument is always required in the wrappers, even when it's
+      optional in the 'neon_local' command. The default in 'neon_local' is a specific
+      hardcoded version, but in tests, we never want to accidentally rely on that;, we
+      always want to use the version from the test fixtures.
+
+    - Wrappers for commands that create a new tenant or timeline ID require the new tenant
+      or timeline ID to be passed by the caller, while the 'neon_local' commands will
+      generate a random ID if it's not specified. This is because we don't want to have to
+      parse the ID from the 'neon_local' output. Making it required ensures that the
+      caller has to generate it.
+    """
+
+    COMMAND = "neon_local"
+
+    def __init__(
+        self,
+        extra_env: Optional[Dict[str, str]],
+        binpath: Path,
+        repo_dir: Path,
+        pg_distrib_dir: Path,
+    ):
+        if extra_env is None:
+            env_vars = {}
+        else:
+            env_vars = extra_env.copy()
+        env_vars["NEON_REPO_DIR"] = str(repo_dir)
+        env_vars["POSTGRES_DISTRIB_DIR"] = str(pg_distrib_dir)
+
+        super().__init__(env_vars, binpath)
+
+    def raw_cli(self, *args, **kwargs) -> subprocess.CompletedProcess[str]:
+        return super().raw_cli(*args, **kwargs)
+
+    def tenant_create(
+        self,
+        tenant_id: TenantId,
+        timeline_id: TimelineId,
+        pg_version: PgVersion,
+        conf: Optional[Dict[str, Any]] = None,
+        shard_count: Optional[int] = None,
+        shard_stripe_size: Optional[int] = None,
+        placement_policy: Optional[str] = None,
+        set_default: bool = False,
+        aux_file_policy: Optional[AuxFileStore] = None,
+    ):
+        """
+        Creates a new tenant, returns its id and its initial timeline's id.
+        """
+        args = [
+            "tenant",
+            "create",
+            "--tenant-id",
+            str(tenant_id),
+            "--timeline-id",
+            str(timeline_id),
+            "--pg-version",
+            pg_version,
+        ]
+        if conf is not None:
+            args.extend(
+                chain.from_iterable(
+                    product(["-c"], (f"{key}:{value}" for key, value in conf.items()))
+                )
+            )
+
+        if aux_file_policy is AuxFileStore.V2:
+            args.extend(["-c", "switch_aux_file_policy:v2"])
+        elif aux_file_policy is AuxFileStore.V1:
+            args.extend(["-c", "switch_aux_file_policy:v1"])
+        elif aux_file_policy is AuxFileStore.CrossValidation:
+            args.extend(["-c", "switch_aux_file_policy:cross-validation"])
+
+        if set_default:
+            args.append("--set-default")
+
+        if shard_count is not None:
+            args.extend(["--shard-count", str(shard_count)])
+
+        if shard_stripe_size is not None:
+            args.extend(["--shard-stripe-size", str(shard_stripe_size)])
+
+        if placement_policy is not None:
+            args.extend(["--placement-policy", str(placement_policy)])
+
+        res = self.raw_cli(args)
+        res.check_returncode()
+
+    def tenant_import(self, tenant_id: TenantId):
+        args = ["tenant", "import", "--tenant-id", str(tenant_id)]
+        res = self.raw_cli(args)
+        res.check_returncode()
+
+    def tenant_set_default(self, tenant_id: TenantId):
+        """
+        Update default tenant for future operations that require tenant_id.
+        """
+        res = self.raw_cli(["tenant", "set-default", "--tenant-id", str(tenant_id)])
+        res.check_returncode()
+
+    def tenant_config(self, tenant_id: TenantId, conf: Dict[str, str]):
+        """
+        Update tenant config.
+        """
+
+        args = ["tenant", "config", "--tenant-id", str(tenant_id)]
+        if conf is not None:
+            args.extend(
+                chain.from_iterable(
+                    product(["-c"], (f"{key}:{value}" for key, value in conf.items()))
+                )
+            )
+
+        res = self.raw_cli(args)
+        res.check_returncode()
+
+    def tenant_list(self) -> "subprocess.CompletedProcess[str]":
+        res = self.raw_cli(["tenant", "list"])
+        res.check_returncode()
+        return res
+
+    def timeline_create(
+        self,
+        new_branch_name: str,
+        tenant_id: TenantId,
+        timeline_id: TimelineId,
+        pg_version: PgVersion,
+    ) -> TimelineId:
+        if timeline_id is None:
+            timeline_id = TimelineId.generate()
+
+        cmd = [
+            "timeline",
+            "create",
+            "--branch-name",
+            new_branch_name,
+            "--tenant-id",
+            str(tenant_id),
+            "--timeline-id",
+            str(timeline_id),
+            "--pg-version",
+            pg_version,
+        ]
+
+        res = self.raw_cli(cmd)
+        res.check_returncode()
+
+        return timeline_id
+
+    def timeline_branch(
+        self,
+        tenant_id: TenantId,
+        timeline_id: TimelineId,
+        new_branch_name,
+        ancestor_branch_name: Optional[str] = None,
+        ancestor_start_lsn: Optional[Lsn] = None,
+    ):
+        cmd = [
+            "timeline",
+            "branch",
+            "--branch-name",
+            new_branch_name,
+            "--timeline-id",
+            str(timeline_id),
+            "--tenant-id",
+            str(tenant_id),
+        ]
+        if ancestor_branch_name is not None:
+            cmd.extend(["--ancestor-branch-name", ancestor_branch_name])
+        if ancestor_start_lsn is not None:
+            cmd.extend(["--ancestor-start-lsn", str(ancestor_start_lsn)])
+
+        res = self.raw_cli(cmd)
+        res.check_returncode()
+
+    def timeline_import(
+        self,
+        tenant_id: TenantId,
+        timeline_id: TimelineId,
+        new_branch_name: str,
+        base_lsn: Lsn,
+        base_tarfile: Path,
+        pg_version: PgVersion,
+        end_lsn: Optional[Lsn] = None,
+        wal_tarfile: Optional[Path] = None,
+    ):
+        cmd = [
+            "timeline",
+            "import",
+            "--tenant-id",
+            str(tenant_id),
+            "--timeline-id",
+            str(timeline_id),
+            "--pg-version",
+            pg_version,
+            "--branch-name",
+            new_branch_name,
+            "--base-lsn",
+            str(base_lsn),
+            "--base-tarfile",
+            str(base_tarfile),
+        ]
+        if end_lsn is not None:
+            cmd.extend(["--end-lsn", str(end_lsn)])
+        if wal_tarfile is not None:
+            cmd.extend(["--wal-tarfile", str(wal_tarfile)])
+
+        res = self.raw_cli(cmd)
+        res.check_returncode()
+
+    def timeline_list(self, tenant_id: TenantId) -> List[Tuple[str, TimelineId]]:
+        """
+        Returns a list of (branch_name, timeline_id) tuples out of parsed `neon timeline list` CLI output.
+        """
+
+        # main [b49f7954224a0ad25cc0013ea107b54b]
+        # ┣━ @0/16B5A50: test_cli_branch_list_main [20f98c79111b9015d84452258b7d5540]
+        TIMELINE_DATA_EXTRACTOR: re.Pattern = re.compile(  # type: ignore[type-arg]
+            r"\s?(?P<branch_name>[^\s]+)\s\[(?P<timeline_id>[^\]]+)\]", re.MULTILINE
+        )
+        res = self.raw_cli(["timeline", "list", "--tenant-id", str(tenant_id)])
+        timelines_cli = sorted(
+            map(
+                lambda branch_and_id: (branch_and_id[0], TimelineId(branch_and_id[1])),
+                TIMELINE_DATA_EXTRACTOR.findall(res.stdout),
+            )
+        )
+        return timelines_cli
+
+    def init(
+        self,
+        init_config: Dict[str, Any],
+        force: Optional[str] = None,
+    ) -> "subprocess.CompletedProcess[str]":
+        with tempfile.NamedTemporaryFile(mode="w+") as init_config_tmpfile:
+            init_config_tmpfile.write(toml.dumps(init_config))
+            init_config_tmpfile.flush()
+
+            cmd = [
+                "init",
+                f"--config={init_config_tmpfile.name}",
+            ]
+
+            if force is not None:
+                cmd.extend(["--force", force])
+
+            res = self.raw_cli(cmd)
+            res.check_returncode()
+        return res
+
+    def storage_controller_start(
+        self,
+        timeout_in_seconds: Optional[int] = None,
+        instance_id: Optional[int] = None,
+        base_port: Optional[int] = None,
+    ):
+        cmd = ["storage_controller", "start"]
+        if timeout_in_seconds is not None:
+            cmd.append(f"--start-timeout={timeout_in_seconds}s")
+        if instance_id is not None:
+            cmd.append(f"--instance-id={instance_id}")
+        if base_port is not None:
+            cmd.append(f"--base-port={base_port}")
+        return self.raw_cli(cmd)
+
+    def storage_controller_stop(self, immediate: bool, instance_id: Optional[int] = None):
+        cmd = ["storage_controller", "stop"]
+        if immediate:
+            cmd.extend(["-m", "immediate"])
+        if instance_id is not None:
+            cmd.append(f"--instance-id={instance_id}")
+        return self.raw_cli(cmd)
+
+    def pageserver_start(
+        self,
+        id: int,
+        extra_env_vars: Optional[Dict[str, str]] = None,
+        timeout_in_seconds: Optional[int] = None,
+    ) -> "subprocess.CompletedProcess[str]":
+        start_args = ["pageserver", "start", f"--id={id}"]
+        if timeout_in_seconds is not None:
+            start_args.append(f"--start-timeout={timeout_in_seconds}s")
+        return self.raw_cli(start_args, extra_env_vars=extra_env_vars)
+
+    def pageserver_stop(self, id: int, immediate=False) -> "subprocess.CompletedProcess[str]":
+        cmd = ["pageserver", "stop", f"--id={id}"]
+        if immediate:
+            cmd.extend(["-m", "immediate"])
+
+        log.info(f"Stopping pageserver with {cmd}")
+        return self.raw_cli(cmd)
+
+    def safekeeper_start(
+        self,
+        id: int,
+        extra_opts: Optional[List[str]] = None,
+        extra_env_vars: Optional[Dict[str, str]] = None,
+        timeout_in_seconds: Optional[int] = None,
+    ) -> "subprocess.CompletedProcess[str]":
+        if extra_opts is not None:
+            extra_opts = [f"-e={opt}" for opt in extra_opts]
+        else:
+            extra_opts = []
+        if timeout_in_seconds is not None:
+            extra_opts.append(f"--start-timeout={timeout_in_seconds}s")
+        return self.raw_cli(
+            ["safekeeper", "start", str(id), *extra_opts], extra_env_vars=extra_env_vars
+        )
+
+    def safekeeper_stop(
+        self, id: Optional[int] = None, immediate=False
+    ) -> "subprocess.CompletedProcess[str]":
+        args = ["safekeeper", "stop"]
+        if id is not None:
+            args.append(str(id))
+        if immediate:
+            args.extend(["-m", "immediate"])
+        return self.raw_cli(args)
+
+    def storage_broker_start(
+        self, timeout_in_seconds: Optional[int] = None
+    ) -> "subprocess.CompletedProcess[str]":
+        cmd = ["storage_broker", "start"]
+        if timeout_in_seconds is not None:
+            cmd.append(f"--start-timeout={timeout_in_seconds}s")
+        return self.raw_cli(cmd)
+
+    def storage_broker_stop(self) -> "subprocess.CompletedProcess[str]":
+        cmd = ["storage_broker", "stop"]
+        return self.raw_cli(cmd)
+
+    def endpoint_create(
+        self,
+        branch_name: str,
+        pg_port: int,
+        http_port: int,
+        tenant_id: TenantId,
+        pg_version: PgVersion,
+        endpoint_id: Optional[str] = None,
+        hot_standby: bool = False,
+        lsn: Optional[Lsn] = None,
+        pageserver_id: Optional[int] = None,
+        allow_multiple=False,
+    ) -> "subprocess.CompletedProcess[str]":
+        args = [
+            "endpoint",
+            "create",
+            "--tenant-id",
+            str(tenant_id),
+            "--branch-name",
+            branch_name,
+            "--pg-version",
+            pg_version,
+        ]
+        if lsn is not None:
+            args.extend(["--lsn", str(lsn)])
+        if pg_port is not None:
+            args.extend(["--pg-port", str(pg_port)])
+        if http_port is not None:
+            args.extend(["--http-port", str(http_port)])
+        if endpoint_id is not None:
+            args.append(endpoint_id)
+        if hot_standby:
+            args.extend(["--hot-standby", "true"])
+        if pageserver_id is not None:
+            args.extend(["--pageserver-id", str(pageserver_id)])
+        if allow_multiple:
+            args.extend(["--allow-multiple"])
+
+        res = self.raw_cli(args)
+        res.check_returncode()
+        return res
+
+    def endpoint_start(
+        self,
+        endpoint_id: str,
+        safekeepers: Optional[List[int]] = None,
+        remote_ext_config: Optional[str] = None,
+        pageserver_id: Optional[int] = None,
+        allow_multiple=False,
+        basebackup_request_tries: Optional[int] = None,
+    ) -> "subprocess.CompletedProcess[str]":
+        args = [
+            "endpoint",
+            "start",
+        ]
+        extra_env_vars = {}
+        if basebackup_request_tries is not None:
+            extra_env_vars["NEON_COMPUTE_TESTING_BASEBACKUP_TRIES"] = str(basebackup_request_tries)
+        if remote_ext_config is not None:
+            args.extend(["--remote-ext-config", remote_ext_config])
+
+        if safekeepers is not None:
+            args.extend(["--safekeepers", (",".join(map(str, safekeepers)))])
+        if endpoint_id is not None:
+            args.append(endpoint_id)
+        if pageserver_id is not None:
+            args.extend(["--pageserver-id", str(pageserver_id)])
+        if allow_multiple:
+            args.extend(["--allow-multiple"])
+
+        res = self.raw_cli(args, extra_env_vars)
+        res.check_returncode()
+        return res
+
+    def endpoint_reconfigure(
+        self,
+        endpoint_id: str,
+        tenant_id: Optional[TenantId] = None,
+        pageserver_id: Optional[int] = None,
+        safekeepers: Optional[List[int]] = None,
+        check_return_code=True,
+    ) -> "subprocess.CompletedProcess[str]":
+        args = ["endpoint", "reconfigure", endpoint_id]
+        if tenant_id is not None:
+            args.extend(["--tenant-id", str(tenant_id)])
+        if pageserver_id is not None:
+            args.extend(["--pageserver-id", str(pageserver_id)])
+        if safekeepers is not None:
+            args.extend(["--safekeepers", (",".join(map(str, safekeepers)))])
+        return self.raw_cli(args, check_return_code=check_return_code)
+
+    def endpoint_stop(
+        self,
+        endpoint_id: str,
+        destroy=False,
+        check_return_code=True,
+        mode: Optional[str] = None,
+    ) -> "subprocess.CompletedProcess[str]":
+        args = [
+            "endpoint",
+            "stop",
+        ]
+        if destroy:
+            args.append("--destroy")
+        if mode is not None:
+            args.append(f"--mode={mode}")
+        if endpoint_id is not None:
+            args.append(endpoint_id)
+
+        return self.raw_cli(args, check_return_code=check_return_code)
+
+    def mappings_map_branch(
+        self, name: str, tenant_id: TenantId, timeline_id: TimelineId
+    ) -> "subprocess.CompletedProcess[str]":
+        """
+        Map tenant id and timeline id to a neon_local branch name. They do not have to exist.
+        Usually needed when creating branches via PageserverHttpClient and not neon_local.
+
+        After creating a name mapping, you can use EndpointFactory.create_start
+        with this registered branch name.
+        """
+        args = [
+            "mappings",
+            "map",
+            "--branch-name",
+            name,
+            "--tenant-id",
+            str(tenant_id),
+            "--timeline-id",
+            str(timeline_id),
+        ]
+
+        return self.raw_cli(args, check_return_code=True)
+
+    def start(self, check_return_code=True) -> "subprocess.CompletedProcess[str]":
+        return self.raw_cli(["start"], check_return_code=check_return_code)
+
+    def stop(self, check_return_code=True) -> "subprocess.CompletedProcess[str]":
+        return self.raw_cli(["stop"], check_return_code=check_return_code)
+
+
+class WalCraft(AbstractNeonCli):
+    """
+    A typed wrapper around the `wal_craft` CLI tool.
+    Supports main commands via typed methods and a way to run arbitrary command directly via CLI.
+    """
+
+    COMMAND = "wal_craft"
+
+    def postgres_config(self) -> List[str]:
+        res = self.raw_cli(["print-postgres-config"])
+        res.check_returncode()
+        return res.stdout.split("\n")
+
+    def in_existing(self, type: str, connection: str) -> None:
+        res = self.raw_cli(["in-existing", type, connection])
+        res.check_returncode()
+
+
+class Pagectl(AbstractNeonCli):
+    """
+    A typed wrapper around the `pagectl` utility CLI tool.
+    """
+
+    COMMAND = "pagectl"
+
+    def dump_index_part(self, path: Path) -> IndexPartDump:
+        res = self.raw_cli(["index-part", "dump", str(path)])
+        res.check_returncode()
+        parsed = json.loads(res.stdout)
+        return IndexPartDump.from_json(parsed)

test_runner/fixtures/neon_fixtures.py

@@ -9,8 +9,6 @@ import os
 import re
 import shutil
 import subprocess
-import tempfile
-import textwrap
 import threading
 import time
 import uuid
@@ -21,7 +19,6 @@ from datetime import datetime
 from enum import Enum
 from fcntl import LOCK_EX, LOCK_UN, flock
 from functools import cached_property
-from itertools import chain, product
 from pathlib import Path
 from types import TracebackType
 from typing import (
@@ -64,11 +61,12 @@ from fixtures.common_types import Lsn, NodeId, TenantId, TenantShardId, Timeline
 from fixtures.endpoint.http import EndpointHttpClient
 from fixtures.log_helper import log
 from fixtures.metrics import Metrics, MetricsGetter, parse_metrics
+from fixtures.neon_cli import NeonLocalCli, Pagectl
 from fixtures.pageserver.allowed_errors import (
     DEFAULT_PAGESERVER_ALLOWED_ERRORS,
     DEFAULT_STORAGE_CONTROLLER_ALLOWED_ERRORS,
 )
-from fixtures.pageserver.common_types import IndexPartDump, LayerName, parse_layer_file_name
+from fixtures.pageserver.common_types import LayerName, parse_layer_file_name
 from fixtures.pageserver.http import PageserverHttpClient
 from fixtures.pageserver.utils import (
     wait_for_last_record_lsn,
@@ -491,7 +489,7 @@ class NeonEnvBuilder:
         log.debug(
             f"Services started, creating initial tenant {env.initial_tenant} and its initial timeline"
         )
-        initial_tenant, initial_timeline = env.neon_cli.create_tenant(
+        initial_tenant, initial_timeline = env.create_tenant(
             tenant_id=env.initial_tenant,
             conf=initial_tenant_conf,
             timeline_id=env.initial_timeline,
@@ -952,10 +950,16 @@ class NeonEnv:
 
     initial_tenant - tenant ID of the initial tenant created in the repository
 
-    neon_cli - can be used to run the 'neon' CLI tool
+    neon_cli - can be used to run the 'neon_local' CLI tool
 
-    create_tenant() - initializes a new tenant in the page server, returns
-        the tenant id
+    create_tenant() - initializes a new tenant and an initial empty timeline on it,
+        returns the tenant and timeline id
+
+    create_branch() - branch a new timeline from an existing one, returns
+        the new timeline id
+
+    create_timeline() - initializes a new timeline by running initdb, returns
+        the new timeline id
     """
 
     BASE_PAGESERVER_ID = 1
@@ -966,8 +970,6 @@ class NeonEnv:
         self.rust_log_override = config.rust_log_override
         self.port_distributor = config.port_distributor
         self.s3_mock_server = config.mock_s3_server
-        self.neon_cli = NeonCli(env=self)
-        self.pagectl = Pagectl(env=self)
         self.endpoints = EndpointFactory(self)
         self.safekeepers: List[Safekeeper] = []
         self.pageservers: List[NeonPageserver] = []
@@ -987,6 +989,21 @@ class NeonEnv:
         self.initial_tenant = config.initial_tenant
         self.initial_timeline = config.initial_timeline
 
+        neon_local_env_vars = {}
+        if self.rust_log_override is not None:
+            neon_local_env_vars["RUST_LOG"] = self.rust_log_override
+        self.neon_cli = NeonLocalCli(
+            extra_env=neon_local_env_vars,
+            binpath=self.neon_local_binpath,
+            repo_dir=self.repo_dir,
+            pg_distrib_dir=self.pg_distrib_dir,
+        )
+
+        pagectl_env_vars = {}
+        if self.rust_log_override is not None:
+            pagectl_env_vars["RUST_LOG"] = self.rust_log_override
+        self.pagectl = Pagectl(extra_env=pagectl_env_vars, binpath=self.neon_binpath)
+
         # The URL for the pageserver to use as its control_plane_api config
         if config.storage_controller_port_override is not None:
             log.info(
@@ -1310,6 +1327,74 @@ class NeonEnv:
         self.endpoint_counter += 1
         return "ep-" + str(self.endpoint_counter)
 
+    def create_tenant(
+        self,
+        tenant_id: Optional[TenantId] = None,
+        timeline_id: Optional[TimelineId] = None,
+        conf: Optional[Dict[str, Any]] = None,
+        shard_count: Optional[int] = None,
+        shard_stripe_size: Optional[int] = None,
+        placement_policy: Optional[str] = None,
+        set_default: bool = False,
+        aux_file_policy: Optional[AuxFileStore] = None,
+    ) -> Tuple[TenantId, TimelineId]:
+        """
+        Creates a new tenant, returns its id and its initial timeline's id.
+        """
+        tenant_id = tenant_id or TenantId.generate()
+        timeline_id = timeline_id or TimelineId.generate()
+
+        self.neon_cli.tenant_create(
+            tenant_id=tenant_id,
+            timeline_id=timeline_id,
+            pg_version=self.pg_version,
+            conf=conf,
+            shard_count=shard_count,
+            shard_stripe_size=shard_stripe_size,
+            placement_policy=placement_policy,
+            set_default=set_default,
+            aux_file_policy=aux_file_policy,
+        )
+
+        return tenant_id, timeline_id
+
+    def config_tenant(self, tenant_id: Optional[TenantId], conf: Dict[str, str]):
+        """
+        Update tenant config.
+        """
+        tenant_id = tenant_id or self.initial_tenant
+        self.neon_cli.tenant_config(tenant_id, conf)
+
+    def create_branch(
+        self,
+        new_branch_name: str = DEFAULT_BRANCH_NAME,
+        tenant_id: Optional[TenantId] = None,
+        ancestor_branch_name: Optional[str] = None,
+        ancestor_start_lsn: Optional[Lsn] = None,
+        new_timeline_id: Optional[TimelineId] = None,
+    ) -> TimelineId:
+        new_timeline_id = new_timeline_id or TimelineId.generate()
+        tenant_id = tenant_id or self.initial_tenant
+
+        self.neon_cli.timeline_branch(
+            tenant_id, new_timeline_id, new_branch_name, ancestor_branch_name, ancestor_start_lsn
+        )
+
+        return new_timeline_id
+
+    def create_timeline(
+        self,
+        new_branch_name: str,
+        tenant_id: Optional[TenantId] = None,
+        timeline_id: Optional[TimelineId] = None,
+    ) -> TimelineId:
+        timeline_id = timeline_id or TimelineId.generate()
+        tenant_id = tenant_id or self.initial_tenant
+
+        self.neon_cli.timeline_create(new_branch_name, tenant_id, timeline_id, self.pg_version)
+
+        return timeline_id
+
 
 @pytest.fixture(scope="function")
 def neon_simple_env(
@@ -1425,597 +1510,6 @@ class PageserverPort:
     http: int
 
 
-class AbstractNeonCli(abc.ABC):
-    """
-    A typed wrapper around an arbitrary Neon CLI tool.
-    Supports a way to run arbitrary command directly via CLI.
-    Do not use directly, use specific subclasses instead.
-    """
-
-    def __init__(self, env: NeonEnv):
-        self.env = env
-
-    COMMAND: str = cast(str, None)  # To be overwritten by the derived class.
-
-    def raw_cli(
-        self,
-        arguments: List[str],
-        extra_env_vars: Optional[Dict[str, str]] = None,
-        check_return_code=True,
-        timeout=None,
-        local_binpath=False,
-    ) -> "subprocess.CompletedProcess[str]":
-        """
-        Run the command with the specified arguments.
-
-        Arguments must be in list form, e.g. ['pg', 'create']
-
-        Return both stdout and stderr, which can be accessed as
-
-        >>> result = env.neon_cli.raw_cli(...)
-        >>> assert result.stderr == ""
-        >>> log.info(result.stdout)
-
-        If `check_return_code`, on non-zero exit code logs failure and raises.
-
-        If `local_binpath` is true, then we are invoking a test utility
-        """
-
-        assert isinstance(arguments, list)
-        assert isinstance(self.COMMAND, str)
-
-        if local_binpath:
-            # Test utility
-            bin_neon = str(self.env.neon_local_binpath / self.COMMAND)
-        else:
-            # Normal binary
-            bin_neon = str(self.env.neon_binpath / self.COMMAND)
-
-        args = [bin_neon] + arguments
-        log.info('Running command "{}"'.format(" ".join(args)))
-
-        env_vars = os.environ.copy()
-        env_vars["NEON_REPO_DIR"] = str(self.env.repo_dir)
-        env_vars["POSTGRES_DISTRIB_DIR"] = str(self.env.pg_distrib_dir)
-        if self.env.rust_log_override is not None:
-            env_vars["RUST_LOG"] = self.env.rust_log_override
-        for extra_env_key, extra_env_value in (extra_env_vars or {}).items():
-            env_vars[extra_env_key] = extra_env_value
-
-        # Pass coverage settings
-        var = "LLVM_PROFILE_FILE"
-        val = os.environ.get(var)
-        if val:
-            env_vars[var] = val
-
-        # Intercept CalledProcessError and print more info
-        try:
-            res = subprocess.run(
-                args,
-                env=env_vars,
-                check=False,
-                universal_newlines=True,
-                stdout=subprocess.PIPE,
-                stderr=subprocess.PIPE,
-                timeout=timeout,
-            )
-        except subprocess.TimeoutExpired as e:
-            if e.stderr:
-                stderr = e.stderr.decode(errors="replace")
-            else:
-                stderr = ""
-
-            if e.stdout:
-                stdout = e.stdout.decode(errors="replace")
-            else:
-                stdout = ""
-
-            log.warn(f"CLI timeout: stderr={stderr}, stdout={stdout}")
-            raise
-
-        indent = "  "
-        if not res.returncode:
-            stripped = res.stdout.strip()
-            lines = stripped.splitlines()
-            if len(lines) < 2:
-                log.debug(f"Run {res.args} success: {stripped}")
-            else:
-                log.debug("Run %s success:\n%s" % (res.args, textwrap.indent(stripped, indent)))
-        elif check_return_code:
-            # this way command output will be in recorded and shown in CI in failure message
-            indent = indent * 2
-            msg = textwrap.dedent(
-                """\
-            Run %s failed:
-              stdout:
-            %s
-              stderr:
-            %s
-            """
-            )
-            msg = msg % (
-                res.args,
-                textwrap.indent(res.stdout.strip(), indent),
-                textwrap.indent(res.stderr.strip(), indent),
-            )
-            log.info(msg)
-            raise RuntimeError(msg) from subprocess.CalledProcessError(
-                res.returncode, res.args, res.stdout, res.stderr
-            )
-        return res
-
-
-class NeonCli(AbstractNeonCli):
-    """
-    A typed wrapper around the `neon` CLI tool.
-    Supports main commands via typed methods and a way to run arbitrary command directly via CLI.
-    """
-
-    COMMAND = "neon_local"
-
-    def raw_cli(self, *args, **kwargs) -> subprocess.CompletedProcess[str]:
-        kwargs["local_binpath"] = True
-        return super().raw_cli(*args, **kwargs)
-
-    def create_tenant(
-        self,
-        tenant_id: Optional[TenantId] = None,
-        timeline_id: Optional[TimelineId] = None,
-        conf: Optional[Dict[str, Any]] = None,
-        shard_count: Optional[int] = None,
-        shard_stripe_size: Optional[int] = None,
-        placement_policy: Optional[str] = None,
-        set_default: bool = False,
-        aux_file_policy: Optional[AuxFileStore] = None,
-    ) -> Tuple[TenantId, TimelineId]:
-        """
-        Creates a new tenant, returns its id and its initial timeline's id.
-        """
-        tenant_id = tenant_id or TenantId.generate()
-        timeline_id = timeline_id or TimelineId.generate()
-
-        args = [
-            "tenant",
-            "create",
-            "--tenant-id",
-            str(tenant_id),
-            "--timeline-id",
-            str(timeline_id),
-            "--pg-version",
-            self.env.pg_version,
-        ]
-        if conf is not None:
-            args.extend(
-                chain.from_iterable(
-                    product(["-c"], (f"{key}:{value}" for key, value in conf.items()))
-                )
-            )
-
-        if aux_file_policy is AuxFileStore.V2:
-            args.extend(["-c", "switch_aux_file_policy:v2"])
-        elif aux_file_policy is AuxFileStore.V1:
-            args.extend(["-c", "switch_aux_file_policy:v1"])
-        elif aux_file_policy is AuxFileStore.CrossValidation:
-            args.extend(["-c", "switch_aux_file_policy:cross-validation"])
-
-        if set_default:
-            args.append("--set-default")
-
-        if shard_count is not None:
-            args.extend(["--shard-count", str(shard_count)])
-
-        if shard_stripe_size is not None:
-            args.extend(["--shard-stripe-size", str(shard_stripe_size)])
-
-        if placement_policy is not None:
-            args.extend(["--placement-policy", str(placement_policy)])
-
-        res = self.raw_cli(args)
-        res.check_returncode()
-        return tenant_id, timeline_id
-
-    def import_tenant(self, tenant_id: TenantId):
-        args = ["tenant", "import", "--tenant-id", str(tenant_id)]
-        res = self.raw_cli(args)
-        res.check_returncode()
-
-    def set_default(self, tenant_id: TenantId):
-        """
-        Update default tenant for future operations that require tenant_id.
-        """
-        res = self.raw_cli(["tenant", "set-default", "--tenant-id", str(tenant_id)])
-        res.check_returncode()
-
-    def config_tenant(self, tenant_id: TenantId, conf: Dict[str, str]):
-        """
-        Update tenant config.
-        """
-
-        args = ["tenant", "config", "--tenant-id", str(tenant_id)]
-        if conf is not None:
-            args.extend(
-                chain.from_iterable(
-                    product(["-c"], (f"{key}:{value}" for key, value in conf.items()))
-                )
-            )
-
-        res = self.raw_cli(args)
-        res.check_returncode()
-
-    def list_tenants(self) -> "subprocess.CompletedProcess[str]":
-        res = self.raw_cli(["tenant", "list"])
-        res.check_returncode()
-        return res
-
-    def create_timeline(
-        self,
-        new_branch_name: str,
-        tenant_id: Optional[TenantId] = None,
-        timeline_id: Optional[TimelineId] = None,
-    ) -> TimelineId:
-        if timeline_id is None:
-            timeline_id = TimelineId.generate()
-
-        cmd = [
-            "timeline",
-            "create",
-            "--branch-name",
-            new_branch_name,
-            "--tenant-id",
-            str(tenant_id or self.env.initial_tenant),
-            "--timeline-id",
-            str(timeline_id),
-            "--pg-version",
-            self.env.pg_version,
-        ]
-
-        res = self.raw_cli(cmd)
-        res.check_returncode()
-
-        return timeline_id
-
-    def create_branch(
-        self,
-        new_branch_name: str = DEFAULT_BRANCH_NAME,
-        ancestor_branch_name: Optional[str] = None,
-        tenant_id: Optional[TenantId] = None,
-        ancestor_start_lsn: Optional[Lsn] = None,
-        new_timeline_id: Optional[TimelineId] = None,
-    ) -> TimelineId:
-        if new_timeline_id is None:
-            new_timeline_id = TimelineId.generate()
-        cmd = [
-            "timeline",
-            "branch",
-            "--branch-name",
-            new_branch_name,
-            "--timeline-id",
-            str(new_timeline_id),
-            "--tenant-id",
-            str(tenant_id or self.env.initial_tenant),
-        ]
-        if ancestor_branch_name is not None:
-            cmd.extend(["--ancestor-branch-name", ancestor_branch_name])
-        if ancestor_start_lsn is not None:
-            cmd.extend(["--ancestor-start-lsn", str(ancestor_start_lsn)])
-
-        res = self.raw_cli(cmd)
-        res.check_returncode()
-
-        return TimelineId(str(new_timeline_id))
-
-    def list_timelines(self, tenant_id: Optional[TenantId] = None) -> List[Tuple[str, TimelineId]]:
-        """
-        Returns a list of (branch_name, timeline_id) tuples out of parsed `neon timeline list` CLI output.
-        """
-
-        # main [b49f7954224a0ad25cc0013ea107b54b]
-        # ┣━ @0/16B5A50: test_cli_branch_list_main [20f98c79111b9015d84452258b7d5540]
-        TIMELINE_DATA_EXTRACTOR: re.Pattern = re.compile(  # type: ignore[type-arg]
-            r"\s?(?P<branch_name>[^\s]+)\s\[(?P<timeline_id>[^\]]+)\]", re.MULTILINE
-        )
-        res = self.raw_cli(
-            ["timeline", "list", "--tenant-id", str(tenant_id or self.env.initial_tenant)]
-        )
-        timelines_cli = sorted(
-            map(
-                lambda branch_and_id: (branch_and_id[0], TimelineId(branch_and_id[1])),
-                TIMELINE_DATA_EXTRACTOR.findall(res.stdout),
-            )
-        )
-        return timelines_cli
-
-    def init(
-        self,
-        init_config: Dict[str, Any],
-        force: Optional[str] = None,
-    ) -> "subprocess.CompletedProcess[str]":
-        with tempfile.NamedTemporaryFile(mode="w+") as init_config_tmpfile:
-            init_config_tmpfile.write(toml.dumps(init_config))
-            init_config_tmpfile.flush()
-
-            cmd = [
-                "init",
-                f"--config={init_config_tmpfile.name}",
-            ]
-
-            if force is not None:
-                cmd.extend(["--force", force])
-
-            res = self.raw_cli(cmd)
-            res.check_returncode()
-        return res
-
-    def storage_controller_start(
-        self,
-        timeout_in_seconds: Optional[int] = None,
-        instance_id: Optional[int] = None,
-        base_port: Optional[int] = None,
-    ):
-        cmd = ["storage_controller", "start"]
-        if timeout_in_seconds is not None:
-            cmd.append(f"--start-timeout={timeout_in_seconds}s")
-        if instance_id is not None:
-            cmd.append(f"--instance-id={instance_id}")
-        if base_port is not None:
-            cmd.append(f"--base-port={base_port}")
-        return self.raw_cli(cmd)
-
-    def storage_controller_stop(self, immediate: bool, instance_id: Optional[int] = None):
-        cmd = ["storage_controller", "stop"]
-        if immediate:
-            cmd.extend(["-m", "immediate"])
-        if instance_id is not None:
-            cmd.append(f"--instance-id={instance_id}")
-        return self.raw_cli(cmd)
-
-    def pageserver_start(
-        self,
-        id: int,
-        extra_env_vars: Optional[Dict[str, str]] = None,
-        timeout_in_seconds: Optional[int] = None,
-    ) -> "subprocess.CompletedProcess[str]":
-        start_args = ["pageserver", "start", f"--id={id}"]
-        if timeout_in_seconds is not None:
-            start_args.append(f"--start-timeout={timeout_in_seconds}s")
-        storage = self.env.pageserver_remote_storage
-
-        if isinstance(storage, S3Storage):
-            s3_env_vars = storage.access_env_vars()
-            extra_env_vars = (extra_env_vars or {}) | s3_env_vars
-
-        return self.raw_cli(start_args, extra_env_vars=extra_env_vars)
-
-    def pageserver_stop(self, id: int, immediate=False) -> "subprocess.CompletedProcess[str]":
-        cmd = ["pageserver", "stop", f"--id={id}"]
-        if immediate:
-            cmd.extend(["-m", "immediate"])
-
-        log.info(f"Stopping pageserver with {cmd}")
-        return self.raw_cli(cmd)
-
-    def safekeeper_start(
-        self,
-        id: int,
-        extra_opts: Optional[List[str]] = None,
-        timeout_in_seconds: Optional[int] = None,
-    ) -> "subprocess.CompletedProcess[str]":
-        s3_env_vars = None
-        if isinstance(self.env.safekeepers_remote_storage, S3Storage):
-            s3_env_vars = self.env.safekeepers_remote_storage.access_env_vars()
-
-        if extra_opts is not None:
-            extra_opts = [f"-e={opt}" for opt in extra_opts]
-        else:
-            extra_opts = []
-        if timeout_in_seconds is not None:
-            extra_opts.append(f"--start-timeout={timeout_in_seconds}s")
-        return self.raw_cli(
-            ["safekeeper", "start", str(id), *extra_opts], extra_env_vars=s3_env_vars
-        )
-
-    def safekeeper_stop(
-        self, id: Optional[int] = None, immediate=False
-    ) -> "subprocess.CompletedProcess[str]":
-        args = ["safekeeper", "stop"]
-        if id is not None:
-            args.append(str(id))
-        if immediate:
-            args.extend(["-m", "immediate"])
-        return self.raw_cli(args)
-
-    def broker_start(
-        self, timeout_in_seconds: Optional[int] = None
-    ) -> "subprocess.CompletedProcess[str]":
-        cmd = ["storage_broker", "start"]
-        if timeout_in_seconds is not None:
-            cmd.append(f"--start-timeout={timeout_in_seconds}s")
-        return self.raw_cli(cmd)
-
-    def broker_stop(self) -> "subprocess.CompletedProcess[str]":
-        cmd = ["storage_broker", "stop"]
-        return self.raw_cli(cmd)
-
-    def endpoint_create(
-        self,
-        branch_name: str,
-        pg_port: int,
-        http_port: int,
-        endpoint_id: Optional[str] = None,
-        tenant_id: Optional[TenantId] = None,
-        hot_standby: bool = False,
-        lsn: Optional[Lsn] = None,
-        pageserver_id: Optional[int] = None,
-        allow_multiple=False,
-    ) -> "subprocess.CompletedProcess[str]":
-        args = [
-            "endpoint",
-            "create",
-            "--tenant-id",
-            str(tenant_id or self.env.initial_tenant),
-            "--branch-name",
-            branch_name,
-            "--pg-version",
-            self.env.pg_version,
-        ]
-        if lsn is not None:
-            args.extend(["--lsn", str(lsn)])
-        if pg_port is not None:
-            args.extend(["--pg-port", str(pg_port)])
-        if http_port is not None:
-            args.extend(["--http-port", str(http_port)])
-        if endpoint_id is not None:
-            args.append(endpoint_id)
-        if hot_standby:
-            args.extend(["--hot-standby", "true"])
-        if pageserver_id is not None:
-            args.extend(["--pageserver-id", str(pageserver_id)])
-        if allow_multiple:
-            args.extend(["--allow-multiple"])
-
-        res = self.raw_cli(args)
-        res.check_returncode()
-        return res
-
-    def endpoint_start(
-        self,
-        endpoint_id: str,
-        safekeepers: Optional[List[int]] = None,
-        remote_ext_config: Optional[str] = None,
-        pageserver_id: Optional[int] = None,
-        allow_multiple=False,
-        basebackup_request_tries: Optional[int] = None,
-    ) -> "subprocess.CompletedProcess[str]":
-        args = [
-            "endpoint",
-            "start",
-        ]
-        extra_env_vars = {}
-        if basebackup_request_tries is not None:
-            extra_env_vars["NEON_COMPUTE_TESTING_BASEBACKUP_TRIES"] = str(basebackup_request_tries)
-        if remote_ext_config is not None:
-            args.extend(["--remote-ext-config", remote_ext_config])
-
-        if safekeepers is not None:
-            args.extend(["--safekeepers", (",".join(map(str, safekeepers)))])
-        if endpoint_id is not None:
-            args.append(endpoint_id)
-        if pageserver_id is not None:
-            args.extend(["--pageserver-id", str(pageserver_id)])
-        if allow_multiple:
-            args.extend(["--allow-multiple"])
-
-        res = self.raw_cli(args, extra_env_vars)
-        res.check_returncode()
-        return res
-
-    def endpoint_reconfigure(
-        self,
-        endpoint_id: str,
-        tenant_id: Optional[TenantId] = None,
-        pageserver_id: Optional[int] = None,
-        safekeepers: Optional[List[int]] = None,
-        check_return_code=True,
-    ) -> "subprocess.CompletedProcess[str]":
-        args = ["endpoint", "reconfigure", endpoint_id]
-        if tenant_id is not None:
-            args.extend(["--tenant-id", str(tenant_id)])
-        if pageserver_id is not None:
-            args.extend(["--pageserver-id", str(pageserver_id)])
-        if safekeepers is not None:
-            args.extend(["--safekeepers", (",".join(map(str, safekeepers)))])
-        return self.raw_cli(args, check_return_code=check_return_code)
-
-    def endpoint_stop(
-        self,
-        endpoint_id: str,
-        destroy=False,
-        check_return_code=True,
-        mode: Optional[str] = None,
-    ) -> "subprocess.CompletedProcess[str]":
-        args = [
-            "endpoint",
-            "stop",
-        ]
-        if destroy:
-            args.append("--destroy")
-        if mode is not None:
-            args.append(f"--mode={mode}")
-        if endpoint_id is not None:
-            args.append(endpoint_id)
-
-        return self.raw_cli(args, check_return_code=check_return_code)
-
-    def map_branch(
-        self, name: str, tenant_id: TenantId, timeline_id: TimelineId
-    ) -> "subprocess.CompletedProcess[str]":
-        """
-        Map tenant id and timeline id to a neon_local branch name. They do not have to exist.
-        Usually needed when creating branches via PageserverHttpClient and not neon_local.
-
-        After creating a name mapping, you can use EndpointFactory.create_start
-        with this registered branch name.
-        """
-        args = [
-            "mappings",
-            "map",
-            "--branch-name",
-            name,
-            "--tenant-id",
-            str(tenant_id),
-            "--timeline-id",
-            str(timeline_id),
-        ]
-
-        return self.raw_cli(args, check_return_code=True)
-
-    def start(self, check_return_code=True) -> "subprocess.CompletedProcess[str]":
-        return self.raw_cli(["start"], check_return_code=check_return_code)
-
-    def stop(self, check_return_code=True) -> "subprocess.CompletedProcess[str]":
-        return self.raw_cli(["stop"], check_return_code=check_return_code)
-
-
-class WalCraft(AbstractNeonCli):
-    """
-    A typed wrapper around the `wal_craft` CLI tool.
-    Supports main commands via typed methods and a way to run arbitrary command directly via CLI.
-    """
-
-    COMMAND = "wal_craft"
-
-    def postgres_config(self) -> List[str]:
-        res = self.raw_cli(["print-postgres-config"])
-        res.check_returncode()
-        return res.stdout.split("\n")
-
-    def in_existing(self, type: str, connection: str) -> None:
-        res = self.raw_cli(["in-existing", type, connection])
-        res.check_returncode()
-
-
-class ComputeCtl(AbstractNeonCli):
-    """
-    A typed wrapper around the `compute_ctl` CLI tool.
-    """
-
-    COMMAND = "compute_ctl"
-
-
-class Pagectl(AbstractNeonCli):
-    """
-    A typed wrapper around the `pagectl` utility CLI tool.
-    """
-
-    COMMAND = "pagectl"
-
-    def dump_index_part(self, path: Path) -> IndexPartDump:
-        res = self.raw_cli(["index-part", "dump", str(path)])
-        res.check_returncode()
-        parsed = json.loads(res.stdout)
-        return IndexPartDump.from_json(parsed)
-
-
 class LogUtils:
     """
     A mixin class which provides utilities for inspecting the logs of a service.
@@ -2933,6 +2427,10 @@ class NeonPageserver(PgProtocol, LogUtils):
         """
         assert self.running is False
 
+        storage = self.env.pageserver_remote_storage
+        if isinstance(storage, S3Storage):
+            s3_env_vars = storage.access_env_vars()
+            extra_env_vars = (extra_env_vars or {}) | s3_env_vars
         self.env.neon_cli.pageserver_start(
             self.id, extra_env_vars=extra_env_vars, timeout_in_seconds=timeout_in_seconds
         )
@@ -3953,6 +3451,7 @@ class Endpoint(PgProtocol, LogUtils):
             hot_standby=hot_standby,
             pg_port=self.pg_port,
             http_port=self.http_port,
+            pg_version=self.env.pg_version,
             pageserver_id=pageserver_id,
             allow_multiple=allow_multiple,
         )
@@ -4395,8 +3894,16 @@ class Safekeeper(LogUtils):
             extra_opts = self.extra_opts
 
         assert self.running is False
+
+        s3_env_vars = None
+        if isinstance(self.env.safekeepers_remote_storage, S3Storage):
+            s3_env_vars = self.env.safekeepers_remote_storage.access_env_vars()
+
         self.env.neon_cli.safekeeper_start(
-            self.id, extra_opts=extra_opts, timeout_in_seconds=timeout_in_seconds
+            self.id,
+            extra_opts=extra_opts,
+            timeout_in_seconds=timeout_in_seconds,
+            extra_env_vars=s3_env_vars,
         )
         self.running = True
         # wait for wal acceptor start by checking its status
@@ -4542,7 +4049,7 @@ class Safekeeper(LogUtils):
         1) wait for remote_consistent_lsn and wal_backup_lsn on safekeeper to reach it.
         2) checkpoint timeline on safekeeper, which should remove WAL before this LSN; optionally wait for that.
         """
-        cli = self.http_client()
+        client = self.http_client()
 
         target_segment_file = lsn.segment_name()
 
@@ -4554,7 +4061,7 @@ class Safekeeper(LogUtils):
             assert all(target_segment_file <= s for s in segments)
 
         def are_lsns_advanced():
-            stat = cli.timeline_status(tenant_id, timeline_id)
+            stat = client.timeline_status(tenant_id, timeline_id)
             log.info(
                 f"waiting for remote_consistent_lsn and backup_lsn on sk {self.id} to reach {lsn}, currently remote_consistent_lsn={stat.remote_consistent_lsn}, backup_lsn={stat.backup_lsn}"
             )
@@ -4563,7 +4070,7 @@ class Safekeeper(LogUtils):
         # xxx: max wait is long because we might be waiting for reconnection from
         # pageserver to this safekeeper
         wait_until(30, 1, are_lsns_advanced)
-        cli.checkpoint(tenant_id, timeline_id)
+        client.checkpoint(tenant_id, timeline_id)
         if wait_wal_removal:
             wait_until(30, 1, are_segments_removed)
 
@@ -4591,13 +4098,13 @@ class NeonBroker(LogUtils):
         timeout_in_seconds: Optional[int] = None,
     ):
         assert not self.running
-        self.env.neon_cli.broker_start(timeout_in_seconds)
+        self.env.neon_cli.storage_broker_start(timeout_in_seconds)
         self.running = True
         return self
 
     def stop(self):
         if self.running:
-            self.env.neon_cli.broker_stop()
+            self.env.neon_cli.storage_broker_stop()
             self.running = False
         return self
 
@@ -5226,10 +4733,10 @@ def flush_ep_to_pageserver(
     commit_lsn: Lsn = Lsn(0)
     # In principle in the absense of failures polling single sk would be enough.
     for sk in env.safekeepers:
-        cli = sk.http_client()
+        client = sk.http_client()
         # wait until compute connections are gone
-        wait_walreceivers_absent(cli, tenant, timeline)
-        commit_lsn = max(cli.get_commit_lsn(tenant, timeline), commit_lsn)
+        wait_walreceivers_absent(client, tenant, timeline)
+        commit_lsn = max(client.get_commit_lsn(tenant, timeline), commit_lsn)
 
     # Note: depending on WAL filtering implementation, probably most shards
     # won't be able to reach commit_lsn (unless gaps are also ack'ed), so this
@@ -5282,7 +4789,12 @@ def fork_at_current_lsn(
     the WAL up to that LSN to arrive in the pageserver before creating the branch.
     """
     current_lsn = endpoint.safe_psql("SELECT pg_current_wal_lsn()")[0][0]
-    return env.neon_cli.create_branch(new_branch_name, ancestor_branch_name, tenant_id, current_lsn)
+    return env.create_branch(
+        new_branch_name=new_branch_name,
+        tenant_id=tenant_id,
+        ancestor_branch_name=ancestor_branch_name,
+        ancestor_start_lsn=current_lsn,
+    )
 
 
 def import_timeline_from_vanilla_postgres(
@@ -5301,9 +4813,9 @@ def import_timeline_from_vanilla_postgres(
     """
 
     # Take backup of the existing PostgreSQL server with pg_basebackup
-    basebackup_dir = os.path.join(test_output_dir, "basebackup")
-    base_tar = os.path.join(basebackup_dir, "base.tar")
-    wal_tar = os.path.join(basebackup_dir, "pg_wal.tar")
+    basebackup_dir = test_output_dir / "basebackup"
+    base_tar = basebackup_dir / "base.tar"
+    wal_tar = basebackup_dir / "pg_wal.tar"
     os.mkdir(basebackup_dir)
     pg_bin.run(
         [
@@ -5313,40 +4825,28 @@ def import_timeline_from_vanilla_postgres(
             "-d",
             vanilla_pg_connstr,
             "-D",
-            basebackup_dir,
+            str(basebackup_dir),
         ]
     )
 
     # Extract start_lsn and end_lsn form the backup manifest file
     with open(os.path.join(basebackup_dir, "backup_manifest")) as f:
         manifest = json.load(f)
-        start_lsn = manifest["WAL-Ranges"][0]["Start-LSN"]
-        end_lsn = manifest["WAL-Ranges"][0]["End-LSN"]
+        start_lsn = Lsn(manifest["WAL-Ranges"][0]["Start-LSN"])
+        end_lsn = Lsn(manifest["WAL-Ranges"][0]["End-LSN"])
 
     # Import the backup tarballs into the pageserver
-    env.neon_cli.raw_cli(
-        [
-            "timeline",
-            "import",
-            "--tenant-id",
-            str(tenant_id),
-            "--timeline-id",
-            str(timeline_id),
-            "--branch-name",
-            branch_name,
-            "--base-lsn",
-            start_lsn,
-            "--base-tarfile",
-            base_tar,
-            "--end-lsn",
-            end_lsn,
-            "--wal-tarfile",
-            wal_tar,
-            "--pg-version",
-            env.pg_version,
-        ]
+    env.neon_cli.timeline_import(
+        tenant_id=tenant_id,
+        timeline_id=timeline_id,
+        new_branch_name=branch_name,
+        base_lsn=start_lsn,
+        base_tarfile=base_tar,
+        end_lsn=end_lsn,
+        wal_tarfile=wal_tar,
+        pg_version=env.pg_version,
     )
-    wait_for_last_record_lsn(env.pageserver.http_client(), tenant_id, timeline_id, Lsn(end_lsn))
+    wait_for_last_record_lsn(env.pageserver.http_client(), tenant_id, timeline_id, end_lsn)
 
 
 def last_flush_lsn_upload(

test_runner/fixtures/pageserver/http.py

@@ -586,6 +586,7 @@ class PageserverHttpClient(requests.Session, MetricsGetter):
         timeline_id: TimelineId,
         force_repartition=False,
         force_image_layer_creation=False,
+        force_l0_compaction=False,
         wait_until_uploaded=False,
         enhanced_gc_bottom_most_compaction=False,
     ):
@@ -595,6 +596,8 @@ class PageserverHttpClient(requests.Session, MetricsGetter):
             query["force_repartition"] = "true"
         if force_image_layer_creation:
             query["force_image_layer_creation"] = "true"
+        if force_l0_compaction:
+            query["force_l0_compaction"] = "true"
         if wait_until_uploaded:
             query["wait_until_uploaded"] = "true"
         if enhanced_gc_bottom_most_compaction:
@@ -701,6 +704,7 @@ class PageserverHttpClient(requests.Session, MetricsGetter):
         timeline_id: TimelineId,
         force_repartition=False,
         force_image_layer_creation=False,
+        force_l0_compaction=False,
         wait_until_uploaded=False,
         compact: Optional[bool] = None,
         **kwargs,
@@ -711,6 +715,8 @@ class PageserverHttpClient(requests.Session, MetricsGetter):
             query["force_repartition"] = "true"
         if force_image_layer_creation:
             query["force_image_layer_creation"] = "true"
+        if force_l0_compaction:
+            query["force_l0_compaction"] = "true"
         if wait_until_uploaded:
             query["wait_until_uploaded"] = "true"
 

test_runner/fixtures/pageserver/remote_storage.py

@@ -7,7 +7,7 @@ from pathlib import Path
 from typing import Any, List, Tuple
 
 from fixtures.common_types import TenantId, TimelineId
-from fixtures.neon_fixtures import NeonEnv, Pagectl
+from fixtures.neon_fixtures import NeonEnv
 from fixtures.pageserver.common_types import (
     InvalidFileName,
     parse_layer_file_name,
@@ -35,7 +35,7 @@ def duplicate_one_tenant(env: NeonEnv, template_tenant: TenantId, new_tenant: Te
         for file in tl.iterdir():
             shutil.copy2(file, dst_tl_dir)
             if "__" in file.name:
-                Pagectl(env).raw_cli(
+                env.pagectl.raw_cli(
                     [
                         "layer",
                         "rewrite-summary",

test_runner/fixtures/workload.py

@@ -175,7 +175,9 @@ class Workload:
                 if upload:
                     # Wait for written data to be uploaded to S3 (force a checkpoint to trigger upload)
                     ps_http.timeline_checkpoint(
-                        tenant_shard_id, self.timeline_id, wait_until_uploaded=True
+                        tenant_shard_id,
+                        self.timeline_id,
+                        wait_until_uploaded=True,
                     )
                     log.info(f"Churn: waiting for remote LSN {last_flush_lsn}")
                 else:

test_runner/performance/pageserver/interactive/test_many_small_tenants.py

@@ -53,7 +53,7 @@ def setup_env(
             "checkpoint_distance": 268435456,
             "image_creation_threshold": 3,
         }
-        template_tenant, template_timeline = env.neon_cli.create_tenant(set_default=True)
+        template_tenant, template_timeline = env.create_tenant(set_default=True)
         env.pageserver.tenant_detach(template_tenant)
         env.pageserver.tenant_attach(template_tenant, config)
         ep = env.endpoints.create_start("main", tenant_id=template_tenant)

test_runner/performance/pageserver/pagebench/test_large_slru_basebackup.py

@@ -81,7 +81,7 @@ def setup_tenant_template(env: NeonEnv, n_txns: int):
         "image_creation_threshold": 3,
     }
 
-    template_tenant, template_timeline = env.neon_cli.create_tenant(set_default=True)
+    template_tenant, template_timeline = env.create_tenant(set_default=True)
     env.pageserver.tenant_detach(template_tenant)
     env.pageserver.tenant_attach(template_tenant, config)
 

test_runner/performance/pageserver/pagebench/test_pageserver_max_throughput_getpage_at_latest_lsn.py

@@ -162,7 +162,7 @@ def setup_tenant_template(env: NeonEnv, pg_bin: PgBin, scale: int):
         "checkpoint_distance": 268435456,
         "image_creation_threshold": 3,
     }
-    template_tenant, template_timeline = env.neon_cli.create_tenant(set_default=True)
+    template_tenant, template_timeline = env.create_tenant(set_default=True)
     env.pageserver.tenant_detach(template_tenant)
     env.pageserver.tenant_attach(template_tenant, config)
     ps_http = env.pageserver.http_client()

test_runner/performance/test_branch_creation.py

@@ -41,7 +41,7 @@ def test_branch_creation_heavy_write(neon_compare: NeonCompare, n_branches: int)
     pg_bin = neon_compare.pg_bin
 
     # Use aggressive GC and checkpoint settings, so GC and compaction happen more often during the test
-    tenant, _ = env.neon_cli.create_tenant(
+    tenant, _ = env.create_tenant(
         conf={
             "gc_period": "5 s",
             "gc_horizon": f"{4 * 1024 ** 2}",
@@ -64,7 +64,7 @@ def test_branch_creation_heavy_write(neon_compare: NeonCompare, n_branches: int)
 
         endpoint.stop()
 
-    env.neon_cli.create_branch("b0", tenant_id=tenant)
+    env.create_branch("b0", tenant_id=tenant)
 
     threads: List[threading.Thread] = []
     threads.append(threading.Thread(target=run_pgbench, args=("b0",), daemon=True))
@@ -78,7 +78,7 @@ def test_branch_creation_heavy_write(neon_compare: NeonCompare, n_branches: int)
         p = random.randint(0, i)
 
         timer = timeit.default_timer()
-        env.neon_cli.create_branch(f"b{i + 1}", f"b{p}", tenant_id=tenant)
+        env.create_branch(f"b{i + 1}", ancestor_branch_name=f"b{p}", tenant_id=tenant)
         dur = timeit.default_timer() - timer
 
         log.info(f"Creating branch b{i+1} took {dur}s")
@@ -104,7 +104,7 @@ def test_branch_creation_many(neon_compare: NeonCompare, n_branches: int, shape:
     # seed the prng so we will measure the same structure every time
     rng = random.Random("2024-02-29")
 
-    env.neon_cli.create_branch("b0")
+    env.create_branch("b0")
 
     endpoint = env.endpoints.create_start("b0")
     neon_compare.pg_bin.run_capture(["pgbench", "-i", "-I", "dtGvp", "-s10", endpoint.connstr()])
@@ -121,7 +121,7 @@ def test_branch_creation_many(neon_compare: NeonCompare, n_branches: int, shape:
 
         timer = timeit.default_timer()
         # each of these uploads to remote storage before completion
-        env.neon_cli.create_branch(f"b{i + 1}", parent)
+        env.create_branch(f"b{i + 1}", ancestor_branch_name=parent)
         dur = timeit.default_timer() - timer
         branch_creation_durations.append(dur)
 
@@ -222,7 +222,7 @@ def wait_and_record_startup_metrics(
 def test_branch_creation_many_relations(neon_compare: NeonCompare):
     env = neon_compare.env
 
-    timeline_id = env.neon_cli.create_branch("root")
+    timeline_id = env.create_branch("root")
 
     endpoint = env.endpoints.create_start("root")
     with closing(endpoint.connect()) as conn:
@@ -238,7 +238,7 @@ def test_branch_creation_many_relations(neon_compare: NeonCompare):
     )
 
     with neon_compare.record_duration("create_branch_time_not_busy_root"):
-        env.neon_cli.create_branch("child_not_busy", "root")
+        env.create_branch("child_not_busy", ancestor_branch_name="root")
 
     # run a concurrent insertion to make the ancestor "busy" during the branch creation
     thread = threading.Thread(
@@ -247,6 +247,6 @@ def test_branch_creation_many_relations(neon_compare: NeonCompare):
     thread.start()
 
     with neon_compare.record_duration("create_branch_time_busy_root"):
-        env.neon_cli.create_branch("child_busy", "root")
+        env.create_branch("child_busy", ancestor_branch_name="root")
 
     thread.join()

test_runner/performance/test_branching.py

@@ -41,7 +41,7 @@ def test_compare_child_and_root_pgbench_perf(neon_compare: NeonCompare):
         )
         neon_compare.zenbenchmark.record_pg_bench_result(branch, res)
 
-    env.neon_cli.create_branch("root")
+    env.create_branch("root")
     endpoint_root = env.endpoints.create_start("root")
     pg_bin.run_capture(["pgbench", "-i", "-I", "dtGvp", endpoint_root.connstr(), "-s10"])
 
@@ -55,14 +55,14 @@ def test_compare_child_and_root_pgbench_perf(neon_compare: NeonCompare):
 
 def test_compare_child_and_root_write_perf(neon_compare: NeonCompare):
     env = neon_compare.env
-    env.neon_cli.create_branch("root")
+    env.create_branch("root")
     endpoint_root = env.endpoints.create_start("root")
 
     endpoint_root.safe_psql(
         "CREATE TABLE foo(key serial primary key, t text default 'foooooooooooooooooooooooooooooooooooooooooooooooooooo')",
     )
 
-    env.neon_cli.create_branch("child", "root")
+    env.create_branch("child", ancestor_branch_name="root")
     endpoint_child = env.endpoints.create_start("child")
 
     with neon_compare.record_duration("root_run_duration"):
@@ -73,7 +73,7 @@ def test_compare_child_and_root_write_perf(neon_compare: NeonCompare):
 
 def test_compare_child_and_root_read_perf(neon_compare: NeonCompare):
     env = neon_compare.env
-    env.neon_cli.create_branch("root")
+    env.create_branch("root")
     endpoint_root = env.endpoints.create_start("root")
 
     endpoint_root.safe_psql_many(
@@ -83,7 +83,7 @@ def test_compare_child_and_root_read_perf(neon_compare: NeonCompare):
         ]
     )
 
-    env.neon_cli.create_branch("child", "root")
+    env.create_branch("child", ancestor_branch_name="root")
     endpoint_child = env.endpoints.create_start("child")
 
     with neon_compare.record_duration("root_run_duration"):

test_runner/performance/test_bulk_tenant_create.py

@@ -26,10 +26,8 @@ def test_bulk_tenant_create(
     for i in range(tenants_count):
         start = timeit.default_timer()
 
-        tenant, _ = env.neon_cli.create_tenant()
-        env.neon_cli.create_timeline(
-            f"test_bulk_tenant_create_{tenants_count}_{i}", tenant_id=tenant
-        )
+        tenant, _ = env.create_tenant()
+        env.create_timeline(f"test_bulk_tenant_create_{tenants_count}_{i}", tenant_id=tenant)
 
         # FIXME: We used to start new safekeepers here. Did that make sense? Should we do it now?
         # if use_safekeepers == 'with_sa':

test_runner/performance/test_bulk_update.py

@@ -16,7 +16,7 @@ def test_bulk_update(neon_env_builder: NeonEnvBuilder, zenbenchmark, fillfactor)
     env = neon_env_builder.init_start()
     n_records = 1000000
 
-    timeline_id = env.neon_cli.create_branch("test_bulk_update")
+    timeline_id = env.create_branch("test_bulk_update")
     tenant_id = env.initial_tenant
     endpoint = env.endpoints.create_start("test_bulk_update")
     cur = endpoint.connect().cursor()