mirror of
https://github.com/elastic/kibana.git
synced 2025-06-27 10:40:07 -04:00
We just had an issue where two PRs were merged and it caused the limit of the `triggerActionsUi` bundle to be exceeded, breaking PR builds. The issue is that we didn't see any indication of this in the on-merge jobs because we don't produce the PR report for on-merge jobs or ask ci-stats if we should fail the job. Instead, we just ship the metrics for baseline purposes. This fixes that problem by adding a `--validate` flag to `node scripts/ship_ci_stats`, which takes care of sending at least some ci-stats and will verify that the bundle limits are not exceeded. Since we didn't catch this issue in the on-merge job the limits were incorrect for over an hour and merged into many PRs, wasting engineering and CI time. Co-authored-by: kibanamachine <42973632+kibanamachine@users.noreply.github.com> |
||
---|---|---|
.. | ||
jest.config.js | ||
kibana.jsonc | ||
package.json | ||
README.md | ||
ship_ci_stats_cli.ts | ||
tsconfig.json |
@kbn/ci-stats-shipper-cli
Simple CLI that runs in CI to ship the metrics produced by the build process. We used to ship these metrics as part of the build, but in order to enable distributed caching of optimizer bundles (which we still don't do) we broke the metrics out of the build and write them to disk instead, to be shipped at a later time.
Run node scripts/ship_ci_stats --help
for usage information.