Skip sync safekeepers if possible

## Problem
[#4325](https://github.com/neondatabase/neon/issues/4325)
## Summary of changes
Use delete_objects() method

.github/workflows/build_and_test.yml

@@ -738,7 +738,7 @@ jobs:
       run:
         shell: sh -eu {0}
     env:
-      VM_BUILDER_VERSION: v0.8.0
+      VM_BUILDER_VERSION: v0.11.0
 
     steps:
       - name: Checkout

Dockerfile.compute-node

@@ -10,10 +10,10 @@ ARG BUILD_TAG
 #
 #########################################################################################
 FROM debian:bullseye-slim AS build-deps
-RUN apt update && \
+RUN 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 \
-    libicu-dev libxslt1-dev liblz4-dev libzstd-dev python3 python3-dev python3-pip libpython3-dev
+    libicu-dev libxslt1-dev liblz4-dev libzstd-dev
 
 #########################################################################################
 #
@@ -26,7 +26,7 @@ ARG PG_VERSION
 COPY vendor/postgres-${PG_VERSION} postgres
 RUN cd postgres && \
     export CONFIGURE_CMD="./configure CFLAGS='-O2 -g3' --enable-debug --with-openssl --with-uuid=ossp \
-    --with-icu --with-libxml --with-libxslt --with-lz4 --with-python" && \
+    --with-icu --with-libxml --with-libxslt --with-lz4" && \
     if [ "${PG_VERSION}" != "v14" ]; then \
         # zstd is available only from PG15
         export CONFIGURE_CMD="${CONFIGURE_CMD} --with-zstd"; \
@@ -41,13 +41,9 @@ RUN cd postgres && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/autoinc.control && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/bloom.control && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/earthdistance.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/hstore_plpython3u.control && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/insert_username.control && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/intagg.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/jsonb_plpython3u.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/ltree_plpythonu.control && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/moddatetime.control && \
-    echo 'trusted = true' >> /usr/local/pgsql/share/extension/plpython3u.control && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/pg_stat_statements.control && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgrowlocks.control && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/pgstattuple.control && \
@@ -502,6 +498,23 @@ RUN wget https://github.com/fboulnois/pg_uuidv7/archive/refs/tags/v1.0.1.tar.gz
     make -j $(getconf _NPROCESSORS_ONLN) install && \
     echo 'trusted = true' >> /usr/local/pgsql/share/extension/pg_uuidv7.control
 
+#########################################################################################
+#
+# Layer "pg-roaringbitmap-pg-build"
+# compile pg_roaringbitmap extension
+#
+#########################################################################################
+FROM build-deps AS pg-roaringbitmap-pg-build
+COPY --from=pg-build /usr/local/pgsql/ /usr/local/pgsql/
+
+ENV PATH "/usr/local/pgsql/bin/:$PATH"
+RUN wget https://github.com/ChenHuajun/pg_roaringbitmap/archive/refs/tags/v0.5.4.tar.gz -O pg_roaringbitmap.tar.gz && \
+    echo "b75201efcb1c2d1b014ec4ae6a22769cc7a224e6e406a587f5784a37b6b5a2aa pg_roaringbitmap.tar.gz" | sha256sum --check && \
+    mkdir pg_roaringbitmap-src && cd pg_roaringbitmap-src && tar xvzf ../pg_roaringbitmap.tar.gz --strip-components=1 -C . && \
+    make -j $(getconf _NPROCESSORS_ONLN) && \
+    make -j $(getconf _NPROCESSORS_ONLN) install && \
+    echo 'trusted = true' >> /usr/local/pgsql/share/extension/roaringbitmap.control
+
 #########################################################################################
 #
 # Layer "rust extensions"
@@ -636,6 +649,7 @@ COPY --from=pg-cron-pg-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=pg-pgx-ulid-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=rdkit-pg-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY --from=pg-uuidv7-pg-build /usr/local/pgsql/ /usr/local/pgsql/
+COPY --from=pg-roaringbitmap-pg-build /usr/local/pgsql/ /usr/local/pgsql/
 COPY pgxn/ pgxn/
 
 RUN make -j $(getconf _NPROCESSORS_ONLN) \
@@ -713,7 +727,6 @@ COPY --from=compute-tools --chown=postgres /home/nonroot/target/release-line-deb
 # libxml2, libxslt1.1 for xml2
 # libzstd1 for zstd
 # libboost*, libfreetype6, and zlib1g for rdkit
-# python3* for plpython3u
 RUN apt update &&  \
     apt install --no-install-recommends -y \
         gdb \
@@ -737,8 +750,6 @@ RUN apt update &&  \
         libcurl4-openssl-dev \
         locales \
         procps \
-        python3 \
-        python3-pip \
         zlib1g && \
     rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* && \
     localedef -i en_US -c -f UTF-8 -A /usr/share/locale/locale.alias en_US.UTF-8

compute_tools/src/compute.rs

@@ -338,9 +338,13 @@ impl ComputeNode {
         let lsn = match spec.mode {
             ComputeMode::Primary => {
                 info!("starting safekeepers syncing");
-                let lsn = self
-                    .sync_safekeepers(pspec.storage_auth_token.clone())
-                    .with_context(|| "failed to sync safekeepers")?;
+                let lsn = if let Some(synced_lsn) = spec.skip_sync_safekeepers {
+                    info!("no need to sync");
+                    synced_lsn
+                } else {
+                    self.sync_safekeepers(pspec.storage_auth_token.clone())
+                        .with_context(|| "failed to sync safekeepers")?
+                };
                 info!("safekeepers synced at LSN {}", lsn);
                 lsn
             }

control_plane/src/endpoint.rs

@@ -68,6 +68,7 @@ pub struct EndpointConf {
     http_port: u16,
     pg_version: u32,
     skip_pg_catalog_updates: bool,
+    skip_sync_safekeepers: Option<utils::lsn::Lsn>,
 }
 
 //
@@ -137,6 +138,7 @@ impl ComputeControlPlane {
             tenant_id,
             pg_version,
             skip_pg_catalog_updates: false,
+            skip_sync_safekeepers: None,
         });
 
         ep.create_endpoint_dir()?;
@@ -151,6 +153,7 @@ impl ComputeControlPlane {
                 pg_port,
                 pg_version,
                 skip_pg_catalog_updates: false,
+                skip_sync_safekeepers: None,
             })?,
         )?;
         std::fs::write(
@@ -189,6 +192,7 @@ pub struct Endpoint {
 
     // Optimizations
     skip_pg_catalog_updates: bool,
+    skip_sync_safekeepers: Option<utils::lsn::Lsn>,
 }
 
 impl Endpoint {
@@ -223,6 +227,7 @@ impl Endpoint {
             tenant_id: conf.tenant_id,
             pg_version: conf.pg_version,
             skip_pg_catalog_updates: conf.skip_pg_catalog_updates,
+            skip_sync_safekeepers: conf.skip_sync_safekeepers,
         })
     }
 
@@ -457,6 +462,7 @@ impl Endpoint {
 
         // Create spec file
         let spec = ComputeSpec {
+            skip_sync_safekeepers: self.skip_sync_safekeepers,
             skip_pg_catalog_updates: self.skip_pg_catalog_updates,
             format_version: 1.0,
             operation_uuid: None,

libs/compute_api/src/spec.rs

@@ -33,6 +33,15 @@ pub struct ComputeSpec {
     #[serde(default)] // Default false
     pub skip_pg_catalog_updates: bool,
 
+    /// An optinal hint that can be passed to speed up startup time if we know
+    /// that safekeepers have already been synced at the given LSN.
+    ///
+    /// NOTE: If there's any possibility that the safekeepers could have advanced
+    ///       (e.g. if we started compute, and it crashed) we should stay on the
+    ///       safe side and provide None.
+    #[serde(default)]
+    pub skip_sync_safekeepers: Option<Lsn>,
+
     // Information needed to connect to the storage layer.
     //
     // `tenant_id`, `timeline_id` and `pageserver_connstring` are always needed.

libs/remote_storage/tests/test_real_s3.rs

@@ -173,10 +173,15 @@ async fn s3_delete_objects_works(ctx: &mut MaybeEnabledS3) -> anyhow::Result<()>
     let path2 = RemotePath::new(&PathBuf::from(format!("{}/path2", ctx.base_prefix,)))
         .with_context(|| "RemotePath conversion")?;
 
+    let path3 = RemotePath::new(&PathBuf::from(format!("{}/path3", ctx.base_prefix,)))
+        .with_context(|| "RemotePath conversion")?;
+
     let data1 = "remote blob data1".as_bytes();
     let data1_len = data1.len();
     let data2 = "remote blob data2".as_bytes();
     let data2_len = data2.len();
+    let data3 = "remote blob data3".as_bytes();
+    let data3_len = data3.len();
     ctx.client
         .upload(std::io::Cursor::new(data1), data1_len, &path1, None)
         .await?;
@@ -185,8 +190,18 @@ async fn s3_delete_objects_works(ctx: &mut MaybeEnabledS3) -> anyhow::Result<()>
         .upload(std::io::Cursor::new(data2), data2_len, &path2, None)
         .await?;
 
+    ctx.client
+        .upload(std::io::Cursor::new(data3), data3_len, &path3, None)
+        .await?;
+
     ctx.client.delete_objects(&[path1, path2]).await?;
 
+    let prefixes = ctx.client.list_prefixes(None).await?;
+
+    assert_eq!(prefixes.len(), 1);
+
+    ctx.client.delete_objects(&[path3]).await?;
+
     Ok(())
 }
 

pageserver/src/http/openapi_spec.yml

@@ -186,10 +186,8 @@ paths:
               schema:
                 $ref: "#/components/schemas/Error"
     delete:
-      description: "Attempts to delete specified timeline. On 500 errors should be retried"
+      description: "Attempts to delete specified timeline. 500 and 409 errors should be retried"
       responses:
-        "200":
-          description: Ok
         "400":
           description: Error when no tenant id found in path or no timeline id
           content:
@@ -214,6 +212,12 @@ paths:
             application/json:
               schema:
                 $ref: "#/components/schemas/NotFoundError"
+        "409":
+          description: Deletion is already in progress, continue polling
+          content:
+            application/json:
+              schema:
+                $ref: "#/components/schemas/ConflictError"
         "412":
           description: Tenant is missing, or timeline has children
           content:

pageserver/src/http/routes.rs

@@ -187,6 +187,7 @@ impl From<crate::tenant::DeleteTimelineError> for ApiError {
                 format!("Cannot delete timeline which has child timelines: {children:?}")
                     .into_boxed_str(),
             ),
+            a @ AlreadyInProgress => ApiError::Conflict(a.to_string()),
             Other(e) => ApiError::InternalServerError(e),
         }
     }

pageserver/src/tenant.rs

@@ -440,8 +440,13 @@ pub enum GetTimelineError {
 pub enum DeleteTimelineError {
     #[error("NotFound")]
     NotFound,
+
     #[error("HasChildren")]
     HasChildren(Vec<TimelineId>),
+
+    #[error("Timeline deletion is already in progress")]
+    AlreadyInProgress,
+
     #[error(transparent)]
     Other(#[from] anyhow::Error),
 }
@@ -1755,14 +1760,11 @@ impl Tenant {
             timeline = Arc::clone(timeline_entry.get());
 
             // Prevent two tasks from trying to delete the timeline at the same time.
-            delete_lock_guard =
-                DeletionGuard(Arc::clone(&timeline.delete_lock).try_lock_owned().map_err(
-                    |_| {
-                        DeleteTimelineError::Other(anyhow::anyhow!(
-                            "timeline deletion is already in progress"
-                        ))
-                    },
-                )?);
+            delete_lock_guard = DeletionGuard(
+                Arc::clone(&timeline.delete_lock)
+                    .try_lock_owned()
+                    .map_err(|_| DeleteTimelineError::AlreadyInProgress)?,
+            );
 
             // If another task finished the deletion just before we acquired the lock,
             // return success.

pageserver/src/tenant/remote_timeline_client.rs

@@ -862,10 +862,8 @@ impl RemoteTimelineClient {
                 "Found {} files not bound to index_file.json, proceeding with their deletion",
                 remaining.len()
             );
-            for file in remaining {
-                warn!("Removing {}", file.object_name().unwrap_or_default());
-                self.storage_impl.delete(&file).await?;
-            }
+            warn!("About to remove {} files", remaining.len());
+            self.storage_impl.delete_objects(&remaining).await?;
         }
 
         let index_file_path = timeline_storage_path.join(Path::new(IndexPart::FILE_NAME));

pageserver/src/tenant/storage_layer.rs

@@ -12,7 +12,7 @@ use crate::context::RequestContext;
 use crate::repository::{Key, Value};
 use crate::task_mgr::TaskKind;
 use crate::walrecord::NeonWalRecord;
-use anyhow::{Context, Result};
+use anyhow::Result;
 use bytes::Bytes;
 use enum_map::EnumMap;
 use enumset::EnumSet;
@@ -343,8 +343,7 @@ impl LayerAccessStats {
 /// All layers should implement a minimal `std::fmt::Debug` without tenant or
 /// timeline names, because those are known in the context of which the layers
 /// are used in (timeline).
-#[async_trait::async_trait]
-pub trait Layer: std::fmt::Debug + Send + Sync + 'static {
+pub trait Layer: std::fmt::Debug + Send + Sync {
     /// Range of keys that this layer covers
     fn get_key_range(&self) -> Range<Key>;
 
@@ -374,42 +373,13 @@ pub trait Layer: std::fmt::Debug + Send + Sync + 'static {
     /// is available. If this returns ValueReconstructResult::Continue, look up
     /// the predecessor layer and call again with the same 'reconstruct_data' to
     /// collect more data.
-    fn get_value_reconstruct_data_blocking(
+    fn get_value_reconstruct_data(
         &self,
         key: Key,
         lsn_range: Range<Lsn>,
-        reconstruct_data: ValueReconstructState,
-        ctx: RequestContext,
-    ) -> Result<(ValueReconstructState, ValueReconstructResult)>;
-
-    /// CANCEL SAFETY: if the returned future is dropped,
-    /// the wrapped closure still run to completion and the return value discarded.
-    /// For the case of get_value_reconstruct_data, we expect the closure to not
-    /// have any side effects, as it only attempts to read a layer (and stuff like
-    /// page cache isn't considered a real side effect).
-    /// But, ...
-    /// TRACING:
-    /// If the returned future is cancelled, the spawn_blocking span can outlive
-    /// the caller's span.
-    /// So, technically, we should be using `parent: None` and `follows_from: current`
-    /// instead. However, in practice, the advantage of maintaining the span stack
-    /// in logs outweighs the disadvantage of having a dangling span in a case that
-    /// is not expected to happen because in pageserver we generally don't drop pending futures.
-    async fn get_value_reconstruct_data(
-        self: Arc<Self>,
-        key: Key,
-        lsn_range: Range<Lsn>,
-        reconstruct_data: ValueReconstructState,
-        ctx: RequestContext,
-    ) -> Result<(ValueReconstructState, ValueReconstructResult)> {
-        let span = tracing::info_span!("get_value_reconstruct_data_spawn_blocking");
-        tokio::task::spawn_blocking(move || {
-            let _enter = span.enter();
-            self.get_value_reconstruct_data_blocking(key, lsn_range, reconstruct_data, ctx)
-        })
-        .await
-        .context("spawn_blocking")?
-    }
+        reconstruct_data: &mut ValueReconstructState,
+        ctx: &RequestContext,
+    ) -> Result<ValueReconstructResult>;
 
     /// A short ID string that uniquely identifies the given layer within a [`LayerMap`].
     fn short_id(&self) -> String;
@@ -529,7 +499,6 @@ impl LayerDescriptor {
     }
 }
 
-#[async_trait::async_trait]
 impl Layer for LayerDescriptor {
     fn get_key_range(&self) -> Range<Key> {
         self.key.clone()
@@ -543,13 +512,13 @@ impl Layer for LayerDescriptor {
         self.is_incremental
     }
 
-    fn get_value_reconstruct_data_blocking(
+    fn get_value_reconstruct_data(
         &self,
         _key: Key,
         _lsn_range: Range<Lsn>,
-        _reconstruct_data: ValueReconstructState,
-        _ctx: RequestContext,
-    ) -> Result<(ValueReconstructState, ValueReconstructResult)> {
+        _reconstruct_data: &mut ValueReconstructState,
+        _ctx: &RequestContext,
+    ) -> Result<ValueReconstructResult> {
         todo!("This method shouldn't be part of the Layer trait")
     }
 

pageserver/src/tenant/storage_layer/delta_layer.rs

@@ -218,7 +218,6 @@ impl std::fmt::Debug for DeltaLayerInner {
     }
 }
 
-#[async_trait::async_trait]
 impl Layer for DeltaLayer {
     /// debugging function to print out the contents of the layer
     fn dump(&self, verbose: bool, ctx: &RequestContext) -> Result<()> {
@@ -295,13 +294,13 @@ impl Layer for DeltaLayer {
         Ok(())
     }
 
-    fn get_value_reconstruct_data_blocking(
+    fn get_value_reconstruct_data(
         &self,
         key: Key,
         lsn_range: Range<Lsn>,
-        mut reconstruct_state: ValueReconstructState,
-        ctx: RequestContext,
-    ) -> anyhow::Result<(ValueReconstructState, ValueReconstructResult)> {
+        reconstruct_state: &mut ValueReconstructState,
+        ctx: &RequestContext,
+    ) -> anyhow::Result<ValueReconstructResult> {
         ensure!(lsn_range.start >= self.desc.lsn_range.start);
         let mut need_image = true;
 
@@ -309,7 +308,7 @@ impl Layer for DeltaLayer {
 
         {
             // Open the file and lock the metadata in memory
-            let inner = self.load(LayerAccessKind::GetValueReconstructData, &ctx)?;
+            let inner = self.load(LayerAccessKind::GetValueReconstructData, ctx)?;
 
             // Scan the page versions backwards, starting from `lsn`.
             let file = &inner.file;
@@ -375,9 +374,9 @@ impl Layer for DeltaLayer {
         // If an older page image is needed to reconstruct the page, let the
         // caller know.
         if need_image {
-            Ok((reconstruct_state, ValueReconstructResult::Continue))
+            Ok(ValueReconstructResult::Continue)
         } else {
-            Ok((reconstruct_state, ValueReconstructResult::Complete))
+            Ok(ValueReconstructResult::Complete)
         }
     }
 

pageserver/src/tenant/storage_layer/image_layer.rs

@@ -149,7 +149,6 @@ impl std::fmt::Debug for ImageLayerInner {
     }
 }
 
-#[async_trait::async_trait]
 impl Layer for ImageLayer {
     /// debugging function to print out the contents of the layer
     fn dump(&self, verbose: bool, ctx: &RequestContext) -> Result<()> {
@@ -182,18 +181,18 @@ impl Layer for ImageLayer {
     }
 
     /// Look up given page in the file
-    fn get_value_reconstruct_data_blocking(
+    fn get_value_reconstruct_data(
         &self,
         key: Key,
         lsn_range: Range<Lsn>,
-        mut reconstruct_state: ValueReconstructState,
-        ctx: RequestContext,
-    ) -> anyhow::Result<(ValueReconstructState, ValueReconstructResult)> {
+        reconstruct_state: &mut ValueReconstructState,
+        ctx: &RequestContext,
+    ) -> anyhow::Result<ValueReconstructResult> {
         assert!(self.desc.key_range.contains(&key));
         assert!(lsn_range.start >= self.lsn);
         assert!(lsn_range.end >= self.lsn);
 
-        let inner = self.load(LayerAccessKind::GetValueReconstructData, &ctx)?;
+        let inner = self.load(LayerAccessKind::GetValueReconstructData, ctx)?;
 
         let file = inner.file.as_ref().unwrap();
         let tree_reader = DiskBtreeReader::new(inner.index_start_blk, inner.index_root_blk, file);
@@ -211,9 +210,9 @@ impl Layer for ImageLayer {
             let value = Bytes::from(blob);
 
             reconstruct_state.img = Some((self.lsn, value));
-            Ok((reconstruct_state, ValueReconstructResult::Complete))
+            Ok(ValueReconstructResult::Complete)
         } else {
-            Ok((reconstruct_state, ValueReconstructResult::Missing))
+            Ok(ValueReconstructResult::Missing)
         }
     }
 

pageserver/src/tenant/storage_layer/inmemory_layer.rs

@@ -110,7 +110,6 @@ impl InMemoryLayer {
     }
 }
 
-#[async_trait::async_trait]
 impl Layer for InMemoryLayer {
     fn get_key_range(&self) -> Range<Key> {
         Key::MIN..Key::MAX
@@ -191,13 +190,13 @@ impl Layer for InMemoryLayer {
     }
 
     /// Look up given value in the layer.
-    fn get_value_reconstruct_data_blocking(
+    fn get_value_reconstruct_data(
         &self,
         key: Key,
         lsn_range: Range<Lsn>,
-        mut reconstruct_state: ValueReconstructState,
-        _ctx: RequestContext,
-    ) -> anyhow::Result<(ValueReconstructState, ValueReconstructResult)> {
+        reconstruct_state: &mut ValueReconstructState,
+        _ctx: &RequestContext,
+    ) -> anyhow::Result<ValueReconstructResult> {
         ensure!(lsn_range.start >= self.start_lsn);
         let mut need_image = true;
 
@@ -214,7 +213,7 @@ impl Layer for InMemoryLayer {
                 match value {
                     Value::Image(img) => {
                         reconstruct_state.img = Some((*entry_lsn, img));
-                        return Ok((reconstruct_state, ValueReconstructResult::Complete));
+                        return Ok(ValueReconstructResult::Complete);
                     }
                     Value::WalRecord(rec) => {
                         let will_init = rec.will_init();
@@ -234,9 +233,9 @@ impl Layer for InMemoryLayer {
         // If an older page image is needed to reconstruct the page, let the
         // caller know.
         if need_image {
-            Ok((reconstruct_state, ValueReconstructResult::Continue))
+            Ok(ValueReconstructResult::Continue)
         } else {
-            Ok((reconstruct_state, ValueReconstructResult::Complete))
+            Ok(ValueReconstructResult::Complete)
         }
     }
 }

pageserver/src/tenant/storage_layer/remote_layer.rs

@@ -6,7 +6,7 @@ use crate::context::RequestContext;
 use crate::repository::Key;
 use crate::tenant::layer_map::BatchedUpdates;
 use crate::tenant::remote_timeline_client::index::LayerFileMetadata;
-use crate::tenant::storage_layer::{Layer, ValueReconstructState};
+use crate::tenant::storage_layer::{Layer, ValueReconstructResult, ValueReconstructState};
 use anyhow::{bail, Result};
 use pageserver_api::models::HistoricLayerInfo;
 use std::ops::Range;
@@ -21,7 +21,7 @@ use utils::{
 use super::filename::{DeltaFileName, ImageFileName};
 use super::{
     DeltaLayer, ImageLayer, LayerAccessStats, LayerAccessStatsReset, LayerIter, LayerKeyIter,
-    LayerResidenceStatus, PersistentLayer, PersistentLayerDesc, ValueReconstructResult,
+    LayerResidenceStatus, PersistentLayer, PersistentLayerDesc,
 };
 
 /// RemoteLayer is a not yet downloaded [`ImageLayer`] or
@@ -63,15 +63,14 @@ impl std::fmt::Debug for RemoteLayer {
     }
 }
 
-#[async_trait::async_trait]
 impl Layer for RemoteLayer {
-    fn get_value_reconstruct_data_blocking(
+    fn get_value_reconstruct_data(
         &self,
         _key: Key,
         _lsn_range: Range<Lsn>,
-        _reconstruct_state: ValueReconstructState,
-        _ctx: RequestContext,
-    ) -> Result<(ValueReconstructState, ValueReconstructResult)> {
+        _reconstruct_state: &mut ValueReconstructState,
+        _ctx: &RequestContext,
+    ) -> Result<ValueReconstructResult> {
         bail!(
             "layer {} needs to be downloaded",
             self.filename().file_name()

pageserver/src/tenant/timeline.rs

@@ -555,14 +555,13 @@ impl Timeline {
             None => None,
         };
 
-        let reconstruct_state = ValueReconstructState {
+        let mut reconstruct_state = ValueReconstructState {
             records: Vec::new(),
             img: cached_page_img,
         };
 
         let timer = self.metrics.get_reconstruct_data_time_histo.start_timer();
-        let reconstruct_state = self
-            .get_reconstruct_data(key, lsn, reconstruct_state, ctx)
+        self.get_reconstruct_data(key, lsn, &mut reconstruct_state, ctx)
             .await?;
         timer.stop_and_record();
 
@@ -2353,9 +2352,9 @@ impl Timeline {
         &self,
         key: Key,
         request_lsn: Lsn,
-        mut reconstruct_state: ValueReconstructState,
+        reconstruct_state: &mut ValueReconstructState,
         ctx: &RequestContext,
-    ) -> Result<ValueReconstructState, PageReconstructError> {
+    ) -> Result<(), PageReconstructError> {
         // Start from the current timeline.
         let mut timeline_owned;
         let mut timeline = self;
@@ -2385,12 +2384,12 @@ impl Timeline {
             // The function should have updated 'state'
             //info!("CALLED for {} at {}: {:?} with {} records, cached {}", key, cont_lsn, result, reconstruct_state.records.len(), cached_lsn);
             match result {
-                ValueReconstructResult::Complete => return Ok(reconstruct_state),
+                ValueReconstructResult::Complete => return Ok(()),
                 ValueReconstructResult::Continue => {
                     // If we reached an earlier cached page image, we're done.
                     if cont_lsn == cached_lsn + 1 {
                         MATERIALIZED_PAGE_CACHE_HIT.inc_by(1);
-                        return Ok(reconstruct_state);
+                        return Ok(());
                     }
                     if prev_lsn <= cont_lsn {
                         // Didn't make any progress in last iteration. Error out to avoid
@@ -2494,19 +2493,13 @@ impl Timeline {
                             // Get all the data needed to reconstruct the page version from this layer.
                             // But if we have an older cached page image, no need to go past that.
                             let lsn_floor = max(cached_lsn + 1, start_lsn);
-                            result = match Arc::clone(open_layer)
-                                .get_value_reconstruct_data(
-                                    key,
-                                    lsn_floor..cont_lsn,
-                                    reconstruct_state,
-                                    ctx.attached_child(),
-                                )
-                                .await
-                            {
-                                Ok((new_reconstruct_state, result)) => {
-                                    reconstruct_state = new_reconstruct_state;
-                                    result
-                                }
+                            result = match open_layer.get_value_reconstruct_data(
+                                key,
+                                lsn_floor..cont_lsn,
+                                reconstruct_state,
+                                ctx,
+                            ) {
+                                Ok(result) => result,
                                 Err(e) => return Err(PageReconstructError::from(e)),
                             };
                             cont_lsn = lsn_floor;
@@ -2527,19 +2520,13 @@ impl Timeline {
                         if cont_lsn > start_lsn {
                             //info!("CHECKING for {} at {} on frozen layer {}", key, cont_lsn, frozen_layer.filename().display());
                             let lsn_floor = max(cached_lsn + 1, start_lsn);
-                            result = match Arc::clone(frozen_layer)
-                                .get_value_reconstruct_data(
-                                    key,
-                                    lsn_floor..cont_lsn,
-                                    reconstruct_state,
-                                    ctx.attached_child(),
-                                )
-                                .await
-                            {
-                                Ok((new_reconstruct_state, result)) => {
-                                    reconstruct_state = new_reconstruct_state;
-                                    result
-                                }
+                            result = match frozen_layer.get_value_reconstruct_data(
+                                key,
+                                lsn_floor..cont_lsn,
+                                reconstruct_state,
+                                ctx,
+                            ) {
+                                Ok(result) => result,
                                 Err(e) => return Err(PageReconstructError::from(e)),
                             };
                             cont_lsn = lsn_floor;
@@ -2568,19 +2555,13 @@ impl Timeline {
                             // Get all the data needed to reconstruct the page version from this layer.
                             // But if we have an older cached page image, no need to go past that.
                             let lsn_floor = max(cached_lsn + 1, lsn_floor);
-                            result = match Arc::clone(&layer)
-                                .get_value_reconstruct_data(
-                                    key,
-                                    lsn_floor..cont_lsn,
-                                    reconstruct_state,
-                                    ctx.attached_child(),
-                                )
-                                .await
-                            {
-                                Ok((new_reconstruct_state, result)) => {
-                                    reconstruct_state = new_reconstruct_state;
-                                    result
-                                }
+                            result = match layer.get_value_reconstruct_data(
+                                key,
+                                lsn_floor..cont_lsn,
+                                reconstruct_state,
+                                ctx,
+                            ) {
+                                Ok(result) => result,
                                 Err(e) => return Err(PageReconstructError::from(e)),
                             };
                             cont_lsn = lsn_floor;

test_runner/performance/test_startup.py

@@ -30,7 +30,18 @@ def test_startup_simple(neon_env_builder: NeonEnvBuilder, zenbenchmark: NeonBenc
     neon_env_builder.num_safekeepers = 3
     env = neon_env_builder.init_start()
 
-    env.neon_cli.create_branch("test_startup")
+    tenant_id = env.initial_tenant
+    timeline_id = env.neon_cli.create_branch("test_startup")
+
+    def get_synced_lsn():
+        """Assert safekeepers are synced and get the LSN."""
+        commit_lsns = [
+            sk.http_client().timeline_status(tenant_id, timeline_id).commit_lsn
+            for sk in env.safekeepers
+        ]
+        assert len(commit_lsns) == 3
+        assert len(set(commit_lsns)) == 1
+        return commit_lsns[0]
 
     endpoint = None
 
@@ -63,7 +74,8 @@ def test_startup_simple(neon_env_builder: NeonEnvBuilder, zenbenchmark: NeonBenc
         endpoint.stop()
 
         # Imitate optimizations that console would do for the second start
-        endpoint.respec(skip_pg_catalog_updates=True)
+        lsn = get_synced_lsn()
+        endpoint.respec(skip_pg_catalog_updates=True, skip_sync_safekeepers=lsn.lsn_int)
 
 
 # This test sometimes runs for longer than the global 5 minute timeout.

test_runner/regress/test_timeline_delete.py

@@ -463,10 +463,10 @@ def test_concurrent_timeline_delete_stuck_on(
 
         # make the second call and assert behavior
         log.info("second call start")
-        error_msg_re = "timeline deletion is already in progress"
+        error_msg_re = "Timeline deletion is already in progress"
         with pytest.raises(PageserverApiException, match=error_msg_re) as second_call_err:
             ps_http.timeline_delete(env.initial_tenant, child_timeline_id)
-        assert second_call_err.value.status_code == 500
+        assert second_call_err.value.status_code == 409
         env.pageserver.allowed_errors.append(f".*{child_timeline_id}.*{error_msg_re}.*")
         # the second call will try to transition the timeline into Stopping state as well
         env.pageserver.allowed_errors.append(
@@ -518,9 +518,9 @@ def test_delete_timeline_client_hangup(neon_env_builder: NeonEnvBuilder):
         ps_http.timeline_delete(env.initial_tenant, child_timeline_id, timeout=2)
 
     env.pageserver.allowed_errors.append(
-        f".*{child_timeline_id}.*timeline deletion is already in progress.*"
+        f".*{child_timeline_id}.*Timeline deletion is already in progress.*"
     )
-    with pytest.raises(PageserverApiException, match="timeline deletion is already in progress"):
+    with pytest.raises(PageserverApiException, match="Timeline deletion is already in progress"):
         ps_http.timeline_delete(env.initial_tenant, child_timeline_id, timeout=2)
 
     # make sure the timeout was due to the failpoint