CI: the process-calibration step retries a run a busy GPU left unmetered #476
Loading…
Reference in a new issue
No description provided.
Delete branch "ci/energy-step-retry"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
The macOS energy job's process-calibration test meters only once a GPU liveness load has run, and that load gives up after 500 ms on a GPU too busy to grant it. The test then passes on its refusal branch but never prints the
process calibration:line the step greps for. So host load alone fails the job, and because the forge cannot rerun one job, the whole pull request has to run again: about 55 jobs for #465.The step now runs the calibration test up to three times, 20 s apart, and stops at the first run on live counters. The line it requires must still come from such a run, and a failing assertion still fails the step at once.
Checks: the workflow parses as YAML, and
wfsteps.pyreports 0 problems. The public-name check is clean.🤖 Generated with Claude Code