ci(e2e): stop gunicorn worker recycling mid-run; slow-test budget for mixed-chart filter spec - #42867
Open
sadpandajoe wants to merge 2 commits into
Open
ci(e2e): stop gunicorn worker recycling mid-run; slow-test budget for mixed-chart filter spec#42867sadpandajoe wants to merge 2 commits into
sadpandajoe wants to merge 2 commits into
Conversation
The E2E backend runs with `--workers 1` (deliberately — multi-worker exposes races in the SQL Lab to Explore flow) alongside `--max-requests 500 --max-requests-jitter 50`. With one worker there is nothing to roll over to, so every recycle takes the whole backend offline: gunicorn drains for the full 30s graceful timeout because browser keep-alive connections stay open, then the replacement worker spends ~5s booting Superset. A Playwright run issues roughly 3800 requests over ~8 minutes, so the cap fired seven times, producing seven ~35s total outages. Specs that happened to navigate during one timed out waiting for the page to render and were rescued by retries — visible as "flaky" rather than as failures. Cypress runs hit the same windows, absorbed by `cypress_run.py --retries 5`. Lowering `--graceful-timeout` is not sufficient: even at 5s the gap exceeds the 6-10s a dashboard load plus chart render needs. Removing the recycle removes the outage, and there is no leak evidence to justify keeping it for an 8-minute process. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Every other dashboard spec that builds its fixtures over the API — the clear-all-filters, gauge-interval, and dashboard-controls specs — raises its budget to TIMEOUT.SLOW_TEST. This one did not, so it ran on the 30s config default despite doing four API round-trips before a full dashboard load with a preselected native filter. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
sadpandajoe
requested review from
betodealmeida,
dpgaspar,
eschutho,
geido,
hainenber,
kgabryje,
mistercrunch,
nytai,
rusackas,
sha174n and
villebro
as code owners
August 6, 2026 22:36
dosubot
Bot
added
the
infra:webserver
Contributor
Code Review Agent Run #10ba52Actionable Suggestions - 0Review Details
Bito Usage GuideCommands Type the following command in the pull request comment and save the comment.
Refer to the documentation for additional commands. Configuration This repository uses Documentation & Help AI Code Review powered by |
github-actions
Bot
added
the
github_actions
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #42867 +/- ##
==========================================
+ Coverage 65.73% 66.29% +0.55%
==========================================
Files 2843 2857 +14
Lines 162666 162825 +159
Branches 37239 37356 +117
==========================================
+ Hits 106926 107940 +1014
+ Misses 53647 52792 -855
Partials 2093 2093
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
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
1 participant
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.
SUMMARY
Two small E2E follow-ups discovered while working on the edit-mode Playwright migration (#41438), which had already merged before these were pushed:
--workers 1alongside--max-requests 500 --max-requests-jitter 50. With one worker there is nothing to roll over to, so every recycle takes the whole backend offline: gunicorn drains for the full 30s graceful timeout (browser keep-alive connections hold it open), then the replacement worker spends ~5s booting. A Playwright run issues roughly 3800 requests over ~8 minutes, so the cap fired seven times, producing seven ~35s outages — visible as flaky tests (retried past a hung navigation) rather than as an infra failure. Removing the recycle removes the outage; there's no leak evidence over an 8-minute process to justify keeping it.mixed-chart-dashboard-filters.spec.tsthe slow-test budget. Every sibling dashboard spec that builds its fixtures over the API (clear-all-filters, gauge-interval, dashboard-controls) raises its timeout toTIMEOUT.SLOW_TEST. This one didn't, despite doing four API round-trips before a full dashboard load with a preselected native filter.TESTING INSTRUCTIONS
playwright-tests/playwright-tests-requiredshould show fewer/no flaky retries tied to worker-recycle windows.mixed-chart-dashboard-filters.spec.tsruns under its own budget instead of the 30s default.ADDITIONAL INFORMATION