ci: remove redundant tests, and parallelize - #1323
Merged
Merged
Conversation
The Go Test (ubuntu) job was a 33-minute serial critical path that paid for the same work repeatedly. Per-job timing from run 30671070765 showed three avoidable costs: - The race step first re-ran internal/duckdb without -race (421s) — a verbatim duplicate of the invocation the ./... step had already run minutes earlier. Excluding duckdb from -race never required re-running it plain. - make test-evalingest ran ./internal/db (260s) even though the evalingest build tag only gates files in internal/server, so db was byte-identical to the untagged run that had just finished. - The standalone coverage job re-ran the entire suite a third time (629s of runner time). Measured instrumentation overhead of -coverprofile is ~0 (artifact: 577.9s with vs 579.8s without), so the main test job can emit the profile for free and upload to Codecov itself. Moving the race run into its own job lets it start alongside the plain run instead of after it. Combined, the push-CI critical path drops from ~33 min to roughly the max of the plain and race lanes (~12 min), and one full-suite job's worth of self-hosted runner occupancy disappears per push. Follow-up for repo admins: if branch protection lists the old coverage job as a required check, it must be repointed at the test job. Generated with Claude Code (claude-fable-5) Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
roborev: Combined Review (
|
mjacobs
changed the title
wesm
pushed a commit
that referenced
this pull request
Aug 2, 2026
Two background timing tests depended on receiver or scheduler ordering and could exercise the wrong state under differently scheduled CI runners. The unwatched poll cooldown test now retains its one-shot completion notification so a completed reconcile pass cannot become a timeout. The daemon replacement startup-wait test now waits until ensureBackgroundServe enters the startup wait before publishing the incompatible runtime, ensuring it exercises replacement rather than a fresh launch. Both changes are test-only; production polling and daemon lifecycle behavior are unchanged. The race/non-race job split already exists on main via #1323. Co-authored-by: Marius van Niekerk <mariusvniekerk@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
CI takes about 33 minutes, and nearly all of it is the serial
Go Test (ubuntu)job. Timing a recent run showed most of that is the same tests running twice.internal/duckdbwithout-racecosting 421s (a fifth of the job), repeating what the./...step had already finished minutes earlier.make test-evalingestalso ran./internal/db, another 260s for an identical result.coveragejob was a third full pass over the suite, contributing 629s of "runner time".After this,
Go Test (windows-latest)(~23 min) becomes the critical path.