Problem
The CI workflow is failing during the ast10x0_qemu_tests build step due to Bazel output base lock contention.
Error Messages
Another command (pid=1964907 (for shutdown)) is running. Waiting for it to complete on the server (server_pid=1964965)...
Another command holds the output base lock:
[2/3] Finished ===> Recipe: build ast10x0_qemu_tests (FAIL)
Root Cause
Multiple Bazel commands are competing for the same output base, either running concurrently or sequentially without proper cleanup. This causes lock contention and build failures.
Job Reference
Proposed Solutions
Option 1: Add explicit output base directories (Recommended)
Isolate Bazel output bases for each job by setting TEST_TMPDIR and explicitly cleaning Bazel state:
jobs:
test:
name: Test
needs: presubmit
runs-on: openprot-ci-runner
env:
TEST_TMPDIR: /tmp/bazel-test-${{ github.run_id }}
steps:
- uses: actions/checkout@v7
- name: Setup Bazel
uses: bazel-contrib/setup-bazel@0.19.0
with:
bazelisk-cache: false
disk-cache: false
repository-cache: false
- name: Clean Bazel
run: ./pw bazel clean --expunge
- name: Run tests
run: ./pw ciOption 2: Add explicit shutdown between steps
Ensure Bazel server is properly shut down to release locks:
- name: Run tests
run: |
bazel shutdown
./pw ci
bazel shutdownOption 3: Add retry logic
If the issue is transient, implement a retry mechanism:
- name: Run tests
run: |
for i in {1..3}; do
./pw ci && break || sleep 30
doneRecommendation
Implement Option 1 first, as it provides isolation between jobs and prevents lock contention at the source.
Problem
The CI workflow is failing during the
ast10x0_qemu_testsbuild step due to Bazel output base lock contention.Error Messages
Root Cause
Multiple Bazel commands are competing for the same output base, either running concurrently or sequentially without proper cleanup. This causes lock contention and build failures.
Job Reference
Proposed Solutions
Option 1: Add explicit output base directories (Recommended)
Isolate Bazel output bases for each job by setting
TEST_TMPDIRand explicitly cleaning Bazel state:Option 2: Add explicit shutdown between steps
Ensure Bazel server is properly shut down to release locks:
Option 3: Add retry logic
If the issue is transient, implement a retry mechanism:
Recommendation
Implement Option 1 first, as it provides isolation between jobs and prevents lock contention at the source.