Proposed PR branch: qe/issue-15-s390x-perf-regression-automation
Category: Feature Request (test/infra investment)
Problem
- Correctness-only testing does not detect throughput/latency regressions.
- Need repeatable s390x workflow with artifacts, thresholds, and a CI hook.
Justification / why this is needed
- Many performance regressions are “correct” but slower; they will pass unit
tests and still break real deployments.
- A reproducible battery with saved artifacts accelerates diagnosis.
- Having a CI workflow makes performance hygiene continuous instead of manual.
Proposed change
- Add sync/run scripts for s390x:
scripts/s390x_sync_and_run.sh
scripts/s390x_run_battery.sh
- Add perf regression runner + thresholds:
scripts/s390x_perf_regression.sh
scripts/s390x_perf_thresholds.env
- Add CI workflow for self-hosted s390x:
.github/workflows/s390x-perf-regression.yml
- (Optional follow-up) Update
README.md validation/perf guidance once workflow/scripts stabilize.
Acceptance criteria
- Scripts run non-interactively and produce stable logs/artifacts.
- Threshold failures are actionable (clear output and captured artifacts).
- CI workflow runs scheduled and manually on s390x runners.
Test plan
- Execute scripts on s390x and confirm:
make build + make test success
- perf battery runs and produces artifacts
References
- Key files:
.github/workflows/s390x-perf-regression.yml, scripts/*.
Proposed PR branch:
qe/issue-15-s390x-perf-regression-automationCategory:
Feature Request(test/infra investment)Problem
Justification / why this is needed
tests and still break real deployments.
Proposed change
scripts/s390x_sync_and_run.shscripts/s390x_run_battery.shscripts/s390x_perf_regression.shscripts/s390x_perf_thresholds.env.github/workflows/s390x-perf-regression.ymlREADME.mdvalidation/perf guidance once workflow/scripts stabilize.Acceptance criteria
Test plan
make build+make testsuccessReferences
.github/workflows/s390x-perf-regression.yml,scripts/*.