CI: bazel build failure on recent PRs - no such file clang

[asraa]

I'm noticiing PRs with:

[8,465 / 12,740] [Prepa] TdGenerate lib/Dialect/BGV/IR/BGVOps.cpp.inc [for tool] ... (2 actions, 0 running)
ERROR: /home/runner/work/heir/heir/lib/Dialect/BGV/IR/BUILD:11:11: Compiling lib/Dialect/BGV/IR/BGVDialect.cpp [for tool] failed: (Exit 1): clang failed: error executing command (from target //lib/Dialect/BGV/IR:Dialect) /usr/lib/llvm-14/bin/clang -U_FORTIFY_SOURCE -fstack-protector -Wall -Wthread-safety -Wself-assign -Wunused-but-set-parameter -Wno-free-nonheap-object -fcolor-diagnostics -fno-omit-frame-pointer -g0 ... (remaining 95 arguments skipped)

Use --sandbox_debug to see verbose messages from the sandbox and retain the sandbox build root for debugging
src/main/tools/process-wrapper-legacy.cc:80: "execvp(/usr/lib/llvm-14/bin/clang, ...)": No such file or directory
INFO: Elapsed time: 18.094s, Critical Path: 0.26s
INFO: 80 processes: 18 internal, 62 processwrapper-sandbox.
FAILED: Build did NOT complete successfully

I'll investigate and see if I can fix.

[asraa]

The runner's don't seem to have changed, e.g. from
https://github.com/google/heir/actions/runs/12747567920/job/35525904809
to
https://github.com/google/heir/actions/runs/12822845640/job/35756413432

since my first thought is maybe the runners don't have llvm-14. But that doesn't seem to be the case.

[asraa]

Even though the runner name is ubuntu-22.04 in the failing link above toggling the runner image seems like its ubuntu 24

https://github.com/actions/runner-images/blob/ubuntu24/20250105.1/images/ubuntu/Ubuntu2404-Readme.mdhttps://github.com/actions/runner-images/blob/ubuntu24/20250105.1/images/ubuntu/Ubuntu2404-Readme.md

Operating System
Ubuntu
24.04.1
LTS
Runner Image
Image: ubuntu-24.04
Version: 20250105.1.0
Included Software: https://github.com/actions/runner-images/blob/ubuntu24/20250105.1/images/ubuntu/Ubuntu2404-Readme.md
Image Release: https://github.com/actions/runner-images/releases/tag/ubuntu24%2F20250105.1

and looking at the release the software inclued is clang 16+
https://github.com/actions/runner-images/blob/ubuntu24/20250105.1/images/ubuntu/Ubuntu2404-Readme.md

[asraa]

clang 14 was the default in the previous runner

[asraa]

Our merges to main use

set -euo pipefail
  
  if [[ "${CACHE_HIT}" == "true" ]]; then
    echo "runner=ubuntu-22.04-8core" >> "$GITHUB_OUTPUT"
  else
    echo "runner=ubuntu-22.04-[3](https://github.com/google/heir/actions/runs/12815041446/job/35732851436#step:4:3)2core" >> "$GITHUB_OUTPUT"
  fi

and when i toggle the runner image they are all explicitly 22.04 so maybe that's why main is fine so far

[asraa]

In this failing run we are pulling the cache from main: https://github.com/google/heir/actions/runs/12822845640/job/35781192813

So I thinkkkk the issue is because of these runner mismatches

[asraa]

I'm going to try a test PR with no cache on build and test and confirm that we can bazel build w/o cache on the new runner assuming it pulls clang 16+

[j2kun]

I also see this warning: Your workflow is using a version of actions/cache that is scheduled for deprecation, actions/cache@88522ab. Please update your workflow to use either v3 or v4 of actions/cache to avoid interruptions

Might be worth updating the cache action version.

[asraa]

Nice find! OK I'll update my test PR in #1276 since it appears to be making progress in building and finding a suitable clang version:

• Updating actions/cache
• Updating actions/upload-artifact (c/f https://github.com/google/heir/actions/runs/12815041436/job/35732850966)
• Updating our manual runner selection to 24.x runners
• Clearing the stale bazel caches on main to start populating the cache with 24.x runners

[asraa]

For posterity this was really the issue: actions/runner-images#11327

Despite labeling the runner as 22.04 we were still seeing 24.04 runners