Reclaim runner disk before load: true smoke builds
Validate / docs-check (push) Successful in 22s
Validate / validate-base (push) Successful in 18m10s
Validate / validate-omos (push) Failing after 25m54s
Publish Docker Image / smoke-base (push) Successful in 11m50s
Publish Docker Image / build-base (linux/amd64) (push) Failing after 38s
Publish Docker Image / build-base (linux/arm64) (push) Failing after 21s
Publish Docker Image / merge-base (push) Has been skipped
Publish Docker Image / smoke-omos (push) Failing after 19m18s
Publish Docker Image / build-omos (linux/amd64) (push) Has been skipped
Publish Docker Image / build-omos (linux/arm64) (push) Has been skipped
Publish Docker Image / merge-omos (push) Has been skipped
Publish Docker Image / update-description (push) Has been skipped

v1.14.31 publish and validate both hit 'No space left on device' on
single-arch amd64 smoke/validate builds. The image has crossed ~3 GB
and the runner's ~40 GB overlay starts ~70% full, so 'load: true'
peak disk (tarball + unpacked image + buildx cache) no longer fits.

Add a 'Reclaim runner disk' step to validate-base, validate-omos,
smoke-base, smoke-omos. Strips catthehacker-resident toolchains we
never use (hosted-tool-cache, dotnet, android, powershell, swift,
ghc, jvm, microsoft, chromium, boost), then runs 'docker system
prune -af --volumes' + 'docker builder prune -af' against the
runner's dockerd before setup-buildx-action. Expected reclaim is
6-12 GB depending on what's resident.

Deliberately NOT in the per-arch matrix build jobs — push-by-digest
doesn't need it and pruning in parallel jobs risks one job nuking
another's in-flight buildx cache. Also add workflow-level
concurrency on docker-publish.yml so concurrent tag pushes serialize
cleanly.
This commit is contained in:
Joakim Persson
2026-05-01 09:34:52 +00:00
parent f0918ba915
commit 23894bc19f
3 changed files with 124 additions and 0 deletions
+52
View File
@@ -46,6 +46,34 @@ jobs:
run: |
echo 'precedence ::ffff:0:0/96 100' >> /etc/gai.conf
# The runner's overlay disk starts ~70% full. `load: true` peak disk
# is tarball + unpacked image + buildx cache, which tips it over
# once the image crosses ~3 GB. Strip catthehacker-resident
# toolchains we never use and any stale docker state up front.
- name: Reclaim runner disk
run: |
set -x
df -h / || true
rm -rf \
/opt/hostedtoolcache \
/opt/microsoft \
/opt/az \
/opt/ghc \
/usr/local/.ghcup \
/usr/share/dotnet \
/usr/share/swift \
/usr/local/lib/android \
/usr/local/share/powershell \
/usr/local/share/chromium \
/usr/local/share/boost \
/usr/lib/jvm 2>/dev/null || true
apt-get clean || true
rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* || true
docker system df || true
docker system prune -af --volumes || true
docker builder prune -af || true
df -h / || true
- name: Set up Docker Buildx
uses: docker/setup-buildx-action@v4
with:
@@ -76,6 +104,30 @@ jobs:
run: |
echo 'precedence ::ffff:0:0/96 100' >> /etc/gai.conf
- name: Reclaim runner disk
run: |
set -x
df -h / || true
rm -rf \
/opt/hostedtoolcache \
/opt/microsoft \
/opt/az \
/opt/ghc \
/usr/local/.ghcup \
/usr/share/dotnet \
/usr/share/swift \
/usr/local/lib/android \
/usr/local/share/powershell \
/usr/local/share/chromium \
/usr/local/share/boost \
/usr/lib/jvm 2>/dev/null || true
apt-get clean || true
rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/* || true
docker system df || true
docker system prune -af --volumes || true
docker builder prune -af || true
df -h / || true
- name: Set up Docker Buildx
uses: docker/setup-buildx-action@v4
with: