fix: resolve WAL archiving performance and memory issues (#746)

The barman-cloud plugin experienced significant performance degradation
and memory growth compared to the embedded solution. WAL archiving was
noticeably slower and memory consumption grew over time.

Root cause: The sidecar uses a read-only filesystem which prevents
Python from creating bytecode at runtime. When Python finds missing or
stale bytecode (.pyc files), it attempts to recompile on every
invocation, causing high CPU usage and memory consumption. The previous
approach pre-compiled bytecode in a separate base image, but the
bytecode was marked as stale when copied between Docker stages,
triggering runtime recompilation attempts.

This change eliminates bytecode staleness by ensuring all Python
bytecode is properly compiled in the final image before the sidecar
starts. The image is now fully distroless and based on trixie
(previously it was distroless-based but copied unnecessary files from
the build stage), reducing size from 463MB to 270MB and package count
from 188 to 35, while maintaining zero HIGH/CRITICAL vulnerabilities.

Closes #656 
Closes #711 
Closes #735

Signed-off-by: Marco Nenciarini <marco.nenciarini@enterprisedb.com>

Author: mnencia

.github/workflows/barman-base-image.yml

@@ -381,34 +381,6 @@ tasks:
         build --dir . --file containers/Dockerfile.sidecar --platform linux/amd64 --platform linux/arm64
         publish --ref {{.SIDECAR_IMAGE_NAME}} --tags {{.IMAGE_VERSION}}
 
-  publish-barman-base:
-    desc: Build and publish a barman-cloud base container image
-    vars:
-      BARMAN_BASE_IMAGE_NAME: ghcr.io/{{.GITHUB_REPOSITORY}}-base{{if not (hasPrefix "refs/heads/main" .GITHUB_REF)}}-testing{{end}}
-      BARMAN_VERSION:
-        sh: grep "^barman" containers/sidecar-requirements.in | sed -E 's/.*==([^ ]+)/\1/'
-      BUILD_DATE:
-        sh: date +"%Y%m%d%H%M"
-    requires:
-      # We expect this to run in a GitHub workflow, so we put a few GitHub-specific vars here
-      # to prevent running this task locally by accident.
-      vars:
-        - CI
-        - GITHUB_REPOSITORY
-        - GITHUB_REF
-        - GITHUB_REF_NAME
-        - REGISTRY_USER
-        - REGISTRY_PASSWORD
-    env:
-      # renovate: datasource=git-refs depName=docker lookupName=https://github.com/purpleclay/daggerverse currentValue=main
-      DAGGER_DOCKER_SHA: ee12c1a4a2630e194ec20c5a9959183e3a78c192
-    cmds:
-      - >
-        dagger call -m github.com/purpleclay/daggerverse/docker@${DAGGER_DOCKER_SHA}
-        --registry ghcr.io --username $REGISTRY_USER --password env:REGISTRY_PASSWORD
-        build --dir . --file containers/Dockerfile.barmanbase --platform linux/amd64 --platform linux/arm64
-        publish --ref {{.BARMAN_BASE_IMAGE_NAME}} --tags "{{.BARMAN_VERSION}}-{{.BUILD_DATE}}"
-
   controller-gen:
     desc: Run controller-gen
     run: once

Taskfile.yml

@@ -10,7 +10,7 @@ ARG TARGETOS
 ARG TARGETARCH
 
 WORKDIR /workspace
-# Copy the Go Modules manifests
+
 COPY ../go.mod go.mod
 COPY ../go.sum go.sum
 # cache deps before building and copying source so that we don't need to re-download as much
@@ -20,35 +20,73 @@ RUN go mod download
 ENV GOCACHE=/root/.cache/go-build
 ENV GOMODCACHE=/go/pkg/mod
 
-# Copy the go source
 COPY ../cmd/manager/main.go cmd/manager/main.go
 COPY ../api/ api/
 COPY ../internal/ internal/
 
-# Build
-# the GOARCH has not a default value to allow the binary be built according to the host where the command
-# was called. For example, if we call make docker-build in a local env which has the Apple Silicon M1 SO
-# the docker BUILDPLATFORM arg will be linux/arm64 when for Apple x86 it will be linux/amd64. Therefore,
-# by leaving it empty we can ensure that the container and binary shipped on it will have the same platform.
+# Build Go binary for target platform (TARGETOS/TARGETARCH)
+# Docker BuildKit sets these based on --platform flag or defaults to the build host platform
 RUN --mount=type=cache,target=/go/pkg/mod --mount=type=cache,target=/root/.cache/go-build \
     CGO_ENABLED=0 GOOS=${TARGETOS:-linux} GOARCH=${TARGETARCH} go build -a -o manager cmd/manager/main.go
 
-# Use plugin-barman-cloud-base to get the dependencies.
-# pip will build everything inside /usr, so we copy every file into a new
-# destination that will then be copied into the distroless container
-FROM ghcr.io/cloudnative-pg/plugin-barman-cloud-base:3.17.0-202601131704 AS pythonbuilder
-# Prepare a new /usr/ directory with the files we'll need in the final image
-RUN mkdir /new-usr/ && \
-    cp -r --parents /usr/local/lib/ /usr/lib/*-linux-gnu/ /usr/local/bin/ \
-    /new-usr/
+# Build Python virtualenv with all dependencies
+FROM debian:trixie-slim AS pythonbuilder
+WORKDIR /build
+
+# Install postgresql-common and setup pgdg repository first
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends postgresql-common && \
+    /usr/share/postgresql-common/pgdg/apt.postgresql.org.sh -y
+
+# Install build dependencies
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends \
+        python3 \
+        python3-venv \
+        python3-dev \
+        build-essential \
+        libpq-dev \
+        liblz4-dev \
+        libsnappy-dev
+
+COPY containers/sidecar-requirements.txt .
 
-# Joint process
-# Now we put everything that was build from the origin into our
-# distroless container
-FROM gcr.io/distroless/python3-debian12:nonroot
+# Create virtualenv and install dependencies
+RUN python3 -m venv /venv && \
+    /venv/bin/pip install --upgrade pip setuptools wheel && \
+    /venv/bin/pip install --no-cache-dir -r sidecar-requirements.txt
+
+# Download and extract runtime library packages and their dependencies
+# Using apt-cache to automatically resolve dependencies, filtering out packages
+# already present in the distroless base image.
+# Distroless package list from: https://github.com/GoogleContainerTools/distroless/blob/main/base/config.bzl
+# and https://github.com/GoogleContainerTools/distroless/blob/main/python3/config.bzl
+RUN mkdir -p /dependencies /build/downloads && \
+    cd /build/downloads && \
+    DISTROLESS_PACKAGES="libc6 libssl3t64 libzstd1 zlib1g libgcc-s1 libstdc++6 \
+        libbz2-1.0 libdb5.3t64 libexpat1 liblzma5 libsqlite3-0 libuuid1 \
+        libncursesw6 libtinfo6 libcom-err2 libcrypt1 libgssapi-krb5-2 \
+        libk5crypto3 libkeyutils1 libkrb5-3 libkrb5support0 libnsl2 \
+        libreadline8t64 libtirpc3t64 libffi8 libpython3.13-minimal \
+        libpython3.13-stdlib python3.13-minimal python3.13-venv" && \
+    apt-cache depends --recurse --no-recommends --no-suggests \
+        --no-conflicts --no-breaks --no-replaces --no-enhances \
+        $DISTROLESS_PACKAGES 2>/dev/null | grep "^\w" | sort -u > /tmp/distroless.txt && \
+    apt-cache depends --recurse --no-recommends --no-suggests \
+        --no-conflicts --no-breaks --no-replaces --no-enhances \
+        libpq5 liblz4-1 libsnappy1v5 2>/dev/null | grep "^\w" | sort -u | \
+        grep -v -F -x -f /tmp/distroless.txt > /tmp/packages.txt && \
+    apt-get download $(cat /tmp/packages.txt) && \
+    for deb in *.deb; do \
+        dpkg -x "$deb" /dependencies; \
+    done
+
+# Final sidecar image using distroless base for minimal size and fewer packages
+FROM gcr.io/distroless/python3-debian13:nonroot
 
 ENV SUMMARY="CloudNativePG Barman plugin" \
-    DESCRIPTION="Container image that provides the barman-cloud sidecar"
+    DESCRIPTION="Container image that provides the barman-cloud sidecar" \
+    PATH="/venv/bin:$PATH"
 
 LABEL summary="$SUMMARY" \
       description="$DESCRIPTION" \
@@ -60,7 +98,13 @@ LABEL summary="$SUMMARY" \
       version="" \
       release="1"
 
-COPY --from=pythonbuilder /new-usr/* /usr/
+COPY --from=pythonbuilder /venv /venv
+COPY --from=pythonbuilder /dependencies/usr/lib /usr/lib
 COPY --from=gobuilder /workspace/manager /manager
+
+# Compile all Python bytecode as root to avoid runtime compilation
+USER 0:0
+RUN ["/venv/bin/python3", "-c", "import sysconfig, compileall; compileall.compile_dir(sysconfig.get_path('stdlib'), quiet=1); compileall.compile_dir('/venv', quiet=1)"]
+
 USER 26:26
 ENTRYPOINT ["/manager"]

containers/Dockerfile.barmanbase

@@ -79,12 +79,6 @@
     enabled: false,
   },
   packageRules: [
-    {
-      matchPackageNames: [
-       'ghcr.io/cloudnative-pg/plugin-barman-cloud-base',
-      ],
-      versioning: 'loose',
-    },
     {
       matchDatasources: [
         'go',