Repository navigation
ci(e2e): rightsize namespace runners and fix trivial flakes and ci waste - #25667
Merged
Merged
Conversation
189 of the 231 required E2E jobs finish in about 42 s (p50) on the 4 vCPU runner, so they pay for cores they don't use. They now run on the small profile. The 20 apps whose jobs take longer than about 90 s set `sentryTest.runner` to keep the 4 vCPU profile. Co-Authored-By: Joshua Li <14209781+joshuarli@users.noreply.github.com>
joshuarli
marked this pull request as ready for review
October 22, 2026 18:01
joshuarli
approved these changes
Oct 22, 2026
Contributor
Author
|
Superseded by #25668, which combines these CI changes so the suite runs once. |
joshuarli
enabled auto-merge (squash)
October 22, 2026 18:21
Contributor
Author
|
Correction to my earlier comment: this PR is not superseded and stays open. #25668 is now stacked on top of it and holds the other CI changes. |
joshuarli
requested review from
Lms24,
andreiborza,
isaacs and
logaretm
and removed request for
a team
October 22, 2026 18:28
Contributor
size-limit report 📦Compared: 93dbb22 (PR test merge using develop at 21f2cf1) Against: 21f2cf1 (develop) · base run #63206 ✅ Baseline matches: this diff contains only this PR's changes.
|
joshuarli
changed the title
Contributor
Optional E2E testsSome optional checks were not successful. They do not block the merge. 4 failing, 60 successful. See the run.
Successful checks
|
pnpm dedupe --check fetches package metadata from npm, and PR runs of this job regularly take 8+ minutes, so the 10 minute limit cancels it. Co-Authored-By: Joshua Li <14209781+joshuarli@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
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.
Required E2E jobs are about 60% of the Namespace compute in a full
build.ymlrun (about 1,450 of 2,500 unit-minutes, from 18 first attempts today). All of them run on the 4x8 profile, but most barely use it. 189 of the 231 matrix entries take 42 s at p50 and 66 s at p90, and most of that is install, a small build and a short Playwright run. Namespace bills the larger of vCPUs or GB/2 per minute, so a 4x8 job costs 4 units/min and a 2x4 job costs 2.These jobs now default to
namespace-profile-sentry-javascript-small(2x4). The 20 apps with a variant slower than 90 s set the newsentryTest.runnerfield to stay on 4x8, and every variant of those apps inherits it. Those apps are mostly Next.js, Nuxt and React Router framework, plusaws-serverless-layer,browser-bfcacheandreact-router-6. A light job that stays under about 75 s on 2 vCPUs still bills 1 minute, so this saves about 380 unit-minutes per full run (about 15%). Fewer vCPUs per E2E burst should also shorten the 2–8 min runner queue I saw on E2E jobs this afternoon, when several PRs ran at once.The matrix is the same as before apart from the
runnerfield. Cache volumes are keyed bycache-tag, not by profile, so the small runners use the same E2E volume. The sizing comes from job durations, not CPU or memory peaks. Before marking this ready, check that the light apps don't get slower or flakier on 2 vCPUs, and check their peaks withnsc instance report. Optional E2E jobs stay on 4x8 for now.via @joshuarli.
--
View Junior Session [Sentry]