Merge branch 'main' into trd/disable-eviction

Author: trungda

.github/workflows/build_and_test.yml

@@ -986,6 +986,7 @@ jobs:
       - name: Verify docker-compose example and test extensions
         timeout-minutes: 60
         env:
+          PARALLEL_COMPUTES: 3
           TAG: >-
             ${{
               needs.meta.outputs.run-kind == 'compute-rc-pr'

.gitignore

@@ -15,6 +15,7 @@ neon.iml
 /.neon
 /integration_tests/.neon
 compaction-suite-results.*
+docker-compose/docker-compose-parallel.yml
 
 # Coverage
 *.profraw

clippy.toml

@@ -1,9 +1,12 @@
 disallowed-methods = [
     "tokio::task::block_in_place",
+
     # Allow this for now, to deny it later once we stop using Handle::block_on completely
     # "tokio::runtime::Handle::block_on",
-    # use tokio_epoll_uring_ext instead
-    "tokio_epoll_uring::thread_local_system",
+
+    # tokio-epoll-uring:
+    # - allow-invalid because the method doesn't exist on macOS
+    { path = "tokio_epoll_uring::thread_local_system", replacement = "tokio_epoll_uring_ext module inside pageserver crate", allow-invalid = true }
 ]
 
 disallowed-macros = [

compute/compute-node.Dockerfile

@@ -1915,10 +1915,10 @@ RUN cd /ext-src/pg_repack-src && patch -p1 </ext-src/pg_repack.patch && rm -f /e
 
 COPY --chmod=755 docker-compose/run-tests.sh /run-tests.sh
 RUN echo /usr/local/pgsql/lib > /etc/ld.so.conf.d/00-neon.conf && /sbin/ldconfig
-RUN apt-get update && apt-get install -y libtap-parser-sourcehandler-pgtap-perl jq \
+RUN apt-get update && apt-get install -y libtap-parser-sourcehandler-pgtap-perl jq parallel \
    && apt clean && rm -rf /ext-src/*.tar.gz /ext-src/*.patch /var/lib/apt/lists/*
 ENV PATH=/usr/local/pgsql/bin:$PATH
-ENV PGHOST=compute
+ENV PGHOST=compute1
 ENV PGPORT=55433
 ENV PGUSER=cloud_admin
 ENV PGDATABASE=postgres

compute_tools/src/compute_prewarm.rs

@@ -105,7 +105,8 @@ impl ComputeNode {
                 cloned.state.lock().unwrap().lfc_prewarm_state = LfcPrewarmState::Completed;
                 return;
             };
-            error!(%err);
+            crate::metrics::LFC_PREWARM_ERRORS.inc();
+            error!(%err, "prewarming lfc");
             cloned.state.lock().unwrap().lfc_prewarm_state = LfcPrewarmState::Failed {
                 error: err.to_string(),
             };
@@ -180,7 +181,8 @@ impl ComputeNode {
             self.state.lock().unwrap().lfc_offload_state = LfcOffloadState::Completed;
             return;
         };
-        error!(%err);
+        crate::metrics::LFC_OFFLOAD_ERRORS.inc();
+        error!(%err, "offloading lfc");
         self.state.lock().unwrap().lfc_offload_state = LfcOffloadState::Failed {
             error: err.to_string(),
         };

compute_tools/src/metrics.rs

@@ -105,6 +105,14 @@ pub(crate) static LFC_PREWARMS: Lazy<IntCounter> = Lazy::new(|| {
     .expect("failed to define a metric")
 });
 
+pub(crate) static LFC_PREWARM_ERRORS: Lazy<IntCounter> = Lazy::new(|| {
+    register_int_counter!(
+        "compute_ctl_lfc_prewarm_errors_total",
+        "Total number of LFC prewarms errors requested by compute_ctl or autoprewarm option",
+    )
+    .expect("failed to define a metric")
+});
+
 pub(crate) static LFC_OFFLOADS: Lazy<IntCounter> = Lazy::new(|| {
     register_int_counter!(
         "compute_ctl_lfc_offloads_total",
@@ -113,6 +121,14 @@ pub(crate) static LFC_OFFLOADS: Lazy<IntCounter> = Lazy::new(|| {
     .expect("failed to define a metric")
 });
 
+pub(crate) static LFC_OFFLOAD_ERRORS: Lazy<IntCounter> = Lazy::new(|| {
+    register_int_counter!(
+        "compute_ctl_lfc_offload_errors_total",
+        "Total number of LFC offload errors requested by compute_ctl or lfc_offload_period_seconds option",
+    )
+    .expect("failed to define a metric")
+});
+
 pub fn collect() -> Vec<MetricFamily> {
     let mut metrics = COMPUTE_CTL_UP.collect();
     metrics.extend(INSTALLED_EXTENSIONS.collect());
@@ -123,6 +139,8 @@ pub fn collect() -> Vec<MetricFamily> {
     metrics.extend(PG_CURR_DOWNTIME_MS.collect());
     metrics.extend(PG_TOTAL_DOWNTIME_MS.collect());
     metrics.extend(LFC_PREWARMS.collect());
+    metrics.extend(LFC_PREWARM_ERRORS.collect());
     metrics.extend(LFC_OFFLOADS.collect());
+    metrics.extend(LFC_OFFLOAD_ERRORS.collect());
     metrics
 }

docker-compose/compute_wrapper/shell/compute.sh

@@ -54,14 +54,16 @@ else
     printf '%s\n' "${result}" | jq .
   fi
 
-  echo "Check if a timeline present"
-  PARAMS=(
-       -X GET
-       -H "Content-Type: application/json"
-       "http://pageserver:9898/v1/tenant/${tenant_id}/timeline"
-  )
-  timeline_id=$(curl "${PARAMS[@]}" | jq -r .[0].timeline_id)
-  if [[ -z "${timeline_id}" || "${timeline_id}" = null ]]; then
+  if [[ "${RUN_PARALLEL:-false}" != "true" ]]; then
+    echo "Check if a timeline present"
+    PARAMS=(
+         -X GET
+         -H "Content-Type: application/json"
+        "http://pageserver:9898/v1/tenant/${tenant_id}/timeline"
+    )
+    timeline_id=$(curl "${PARAMS[@]}" | jq -r .[0].timeline_id)
+  fi
+  if [[ -z "${timeline_id:-}" || "${timeline_id:-}" = null ]]; then
     generate_id timeline_id
     PARAMS=(
         -sbf

docker-compose/docker-compose.yml

@@ -142,7 +142,7 @@ services:
       - "storage_broker"
       - "--listen-addr=0.0.0.0:50051"
 
-  compute:
+  compute1:
     restart: always
     build:
       context: ./compute_wrapper/
@@ -152,6 +152,7 @@ services:
         - TAG=${COMPUTE_TAG:-${TAG:-latest}}
         - http_proxy=${http_proxy:-}
         - https_proxy=${https_proxy:-}
+    image: built-compute
     environment:
       - PG_VERSION=${PG_VERSION:-16}
       - TENANT_ID=${TENANT_ID:-}
@@ -166,6 +167,11 @@ services:
       - 3080:3080 # http endpoints
     entrypoint:
       - "/shell/compute.sh"
+    # Ad an alias for compute1 for compatibility
+    networks:
+      default:
+        aliases:
+            - compute
     depends_on:
       - safekeeper1
       - safekeeper2
@@ -174,15 +180,20 @@ services:
 
   compute_is_ready:
     image: postgres:latest
+    environment:
+      - PARALLEL_COMPUTES=1
     entrypoint:
-      - "/bin/bash"
+      - "/bin/sh"
       - "-c"
     command:
-      - "until pg_isready -h compute -p 55433 -U cloud_admin ; do
-            echo 'Waiting to start compute...' && sleep 1;
-         done"
+      - "for i in $(seq 1 $${PARALLEL_COMPUTES}); do
+           until pg_isready -h compute$$i -p 55433 -U cloud_admin ; do
+             sleep 1;
+           done;
+         done;
+         echo All computes are started"
     depends_on:
-      - compute
+      - compute1
 
   neon-test-extensions:
     profiles: ["test-extensions"]
@@ -196,4 +207,4 @@ services:
     command:
       - sleep 3600
     depends_on:
-      - compute
+      - compute1

docker-compose/docker_compose_test.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 
 # A basic test to ensure Docker images are built correctly.
 # Build a wrapper around the compute, start all services and runs a simple SQL query.
@@ -13,9 +13,36 @@
 #
 set -eux -o pipefail
 
+cd "$(dirname "${0}")"
 export COMPOSE_FILE='docker-compose.yml'
 export COMPOSE_PROFILES=test-extensions
-cd "$(dirname "${0}")"
+export PARALLEL_COMPUTES=${PARALLEL_COMPUTES:-1}
+READY_MESSAGE="All computes are started"
+COMPUTES=()
+for i in $(seq 1 "${PARALLEL_COMPUTES}"); do
+  COMPUTES+=("compute${i}")
+done
+CURRENT_TMPDIR=$(mktemp -d)
+trap 'rm -rf ${CURRENT_TMPDIR} docker-compose-parallel.yml' EXIT
+if [[ ${PARALLEL_COMPUTES} -gt 1 ]]; then
+  export COMPOSE_FILE=docker-compose-parallel.yml
+  cp docker-compose.yml docker-compose-parallel.yml
+  # Replace the environment variable PARALLEL_COMPUTES with the actual value
+  yq eval -i ".services.compute_is_ready.environment |=  map(select(. | test(\"^PARALLEL_COMPUTES=\") | not)) + [\"PARALLEL_COMPUTES=${PARALLEL_COMPUTES}\"]" ${COMPOSE_FILE}
+  for i in $(seq 2 "${PARALLEL_COMPUTES}"); do
+    # Duplicate compute1 as compute${i} for parallel execution
+    yq eval -i ".services.compute${i} = .services.compute1" ${COMPOSE_FILE}
+    # We don't need these sections, so delete them
+    yq eval -i "(del .services.compute${i}.build) | (del .services.compute${i}.ports) | (del .services.compute${i}.networks)" ${COMPOSE_FILE}
+    # Let the compute 1 be the only dependence
+    yq eval -i ".services.compute${i}.depends_on = [\"compute1\"]" ${COMPOSE_FILE}
+    # Set RUN_PARALLEL=true for compute2. They will generate tenant_id and timeline_id to avoid using the same as other computes
+    yq eval -i ".services.compute${i}.environment += [\"RUN_PARALLEL=true\"]" ${COMPOSE_FILE}
+    # Remove TENANT_ID and TIMELINE_ID from the environment variables of the generated computes
+    # They will create new TENANT_ID and TIMELINE_ID anyway.
+    yq eval -i ".services.compute${i}.environment |= map(select(. | (test(\"^TENANT_ID=\") or test(\"^TIMELINE_ID=\")) | not))" ${COMPOSE_FILE}
+  done
+fi
 PSQL_OPTION="-h localhost -U cloud_admin -p 55433 -d postgres"
 
 function cleanup() {
@@ -27,11 +54,11 @@ function cleanup() {
 
 for pg_version in ${TEST_VERSION_ONLY-14 15 16 17}; do
     pg_version=${pg_version/v/}
-    echo "clean up containers if exists"
+    echo "clean up containers if exist"
     cleanup
     PG_TEST_VERSION=$((pg_version < 16 ? 16 : pg_version))
-    PG_VERSION=${pg_version} PG_TEST_VERSION=${PG_TEST_VERSION} docker compose up --quiet-pull --build -d
-
+    PG_VERSION=${pg_version} PG_TEST_VERSION=${PG_TEST_VERSION} docker compose build compute1
+    PG_VERSION=${pg_version} PG_TEST_VERSION=${PG_TEST_VERSION} docker compose up --quiet-pull -d
     echo "wait until the compute is ready. timeout after 60s. "
     cnt=0
     while sleep 3; do
@@ -41,45 +68,50 @@ for pg_version in ${TEST_VERSION_ONLY-14 15 16 17}; do
             echo "timeout before the compute is ready."
             exit 1
         fi
-        if docker compose logs "compute_is_ready" | grep -q "accepting connections"; then
+        if docker compose logs compute_is_ready | grep -q "${READY_MESSAGE}"; then
             echo "OK. The compute is ready to connect."
             echo "execute simple queries."
-            docker compose exec compute /bin/bash -c "psql ${PSQL_OPTION} -c 'SELECT 1'"
+            for compute in "${COMPUTES[@]}"; do
+              docker compose exec "${compute}" /bin/bash -c "psql ${PSQL_OPTION} -c 'SELECT 1'"
+            done
             break
         fi
     done
 
     if [[ ${pg_version} -ge 16 ]]; then
-        # This is required for the pg_hint_plan test, to prevent flaky log message causing the test to fail
-        # It cannot be moved to Dockerfile now because the database directory is created after the start of the container
-        echo Adding dummy config
-        docker compose exec compute touch /var/db/postgres/compute/compute_ctl_temp_override.conf
-        # Prepare for the PostGIS test
-        docker compose exec compute mkdir -p /tmp/pgis_reg/pgis_reg_tmp
-        TMPDIR=$(mktemp -d)
-        docker compose cp neon-test-extensions:/ext-src/postgis-src/raster/test "${TMPDIR}"
-        docker compose cp neon-test-extensions:/ext-src/postgis-src/regress/00-regress-install "${TMPDIR}"
-        docker compose exec compute mkdir -p /ext-src/postgis-src/raster /ext-src/postgis-src/regress /ext-src/postgis-src/regress/00-regress-install
-        docker compose cp "${TMPDIR}/test" compute:/ext-src/postgis-src/raster/test
-        docker compose cp "${TMPDIR}/00-regress-install" compute:/ext-src/postgis-src/regress
-        rm -rf "${TMPDIR}"
-        # The following block copies the files for the pg_hintplan test to the compute node for the extension test in an isolated docker-compose environment
-        TMPDIR=$(mktemp -d)
-        docker compose cp neon-test-extensions:/ext-src/pg_hint_plan-src/data "${TMPDIR}/data"
-        docker compose cp "${TMPDIR}/data" compute:/ext-src/pg_hint_plan-src/
-        rm -rf "${TMPDIR}"
-        # The following block does the same for the contrib/file_fdw test
-        TMPDIR=$(mktemp -d)
-        docker compose cp neon-test-extensions:/postgres/contrib/file_fdw/data "${TMPDIR}/data"
-        docker compose cp "${TMPDIR}/data" compute:/postgres/contrib/file_fdw/data
-        rm -rf "${TMPDIR}"
+        mkdir "${CURRENT_TMPDIR}"/{pg_hint_plan-src,file_fdw,postgis-src}
+        docker compose cp neon-test-extensions:/ext-src/postgis-src/raster/test "${CURRENT_TMPDIR}/postgis-src/test"
+        docker compose cp neon-test-extensions:/ext-src/postgis-src/regress/00-regress-install "${CURRENT_TMPDIR}/postgis-src/00-regress-install"
+        docker compose cp neon-test-extensions:/ext-src/pg_hint_plan-src/data "${CURRENT_TMPDIR}/pg_hint_plan-src/data"
+        docker compose cp neon-test-extensions:/postgres/contrib/file_fdw/data "${CURRENT_TMPDIR}/file_fdw/data"
+
+        for compute in "${COMPUTES[@]}"; do
+          # This is required for the pg_hint_plan test, to prevent flaky log message causing the test to fail
+          # It cannot be moved to Dockerfile now because the database directory is created after the start of the container
+          echo Adding dummy config on "${compute}"
+          docker compose exec "${compute}" touch /var/db/postgres/compute/compute_ctl_temp_override.conf
+          # Prepare for the PostGIS test
+          docker compose exec "${compute}" mkdir -p /tmp/pgis_reg/pgis_reg_tmp /ext-src/postgis-src/raster /ext-src/postgis-src/regress /ext-src/postgis-src/regress/00-regress-install
+          docker compose cp "${CURRENT_TMPDIR}/postgis-src/test" "${compute}":/ext-src/postgis-src/raster/test
+          docker compose cp "${CURRENT_TMPDIR}/postgis-src/00-regress-install" "${compute}":/ext-src/postgis-src/regress
+          # The following block copies the files for the pg_hintplan test to the compute node for the extension test in an isolated docker-compose environment
+          docker compose cp "${CURRENT_TMPDIR}/pg_hint_plan-src/data" "${compute}":/ext-src/pg_hint_plan-src/
+          # The following block does the same for the contrib/file_fdw test
+          docker compose cp "${CURRENT_TMPDIR}/file_fdw/data" "${compute}":/postgres/contrib/file_fdw/data
+        done
         # Apply patches
         docker compose exec -T neon-test-extensions bash -c "(cd /postgres && patch -p1)" <"../compute/patches/contrib_pg${pg_version}.patch"
         # We are running tests now
         rm -f testout.txt testout_contrib.txt
+        # We want to run the longest tests first to better utilize parallelization and reduce overall test time.
+        # Tests listed in the RUN_FIRST variable will be run before others.
+        # If parallelization is not used, this environment variable will be ignored.
+
         docker compose exec -e USE_PGXS=1 -e SKIP=timescaledb-src,rdkit-src,pg_jsonschema-src,kq_imcx-src,wal2json_2_5-src,rag_jina_reranker_v1_tiny_en-src,rag_bge_small_en_v15-src \
+        -e RUN_FIRST=hll-src,postgis-src,pgtap-src -e PARALLEL_COMPUTES="${PARALLEL_COMPUTES}" \
         neon-test-extensions /run-tests.sh /ext-src | tee testout.txt && EXT_SUCCESS=1 || EXT_SUCCESS=0
         docker compose exec -e SKIP=start-scripts,postgres_fdw,ltree_plpython,jsonb_plpython,jsonb_plperl,hstore_plpython,hstore_plperl,dblink,bool_plperl \
+        -e PARALLEL_COMPUTES="${PARALLEL_COMPUTES}" \
         neon-test-extensions /run-tests.sh /postgres/contrib | tee testout_contrib.txt && CONTRIB_SUCCESS=1 || CONTRIB_SUCCESS=0
         if [[ ${EXT_SUCCESS} -eq 0 || ${CONTRIB_SUCCESS} -eq 0 ]]; then
             CONTRIB_FAILED=