Repository navigation
CI: fix the checks that fail under parallel workers - #29
Merged
Merged
Conversation
The backend suite runs on 4 workers that share the runner's CPUs. Three alias matcher tests measured wall time and failed on CI (1.4 s, 2.8 s, 7.2 s against limits of 1, 2 and 5 s) while the matcher itself took 0.1 to 0.6 s. They now measure the process's CPU time, which waiting for a core does not inflate; an exponential or quadratic regression still costs far more than the limits. The huge-address sign-in test goes through the API, so it keeps wall time with a 5 s limit instead of 1 s: the regression it guards against takes about a minute.
|
The latest updates on your projects. Learn more about Vercel for GitHub.
1 Skipped Deployment
|
…e does The first main run on 4 workers failed three more checks, none of them a defect in the product: - The cross-tenant list test looked for the other tenant's numeric snapshot id as a substring. On a fresh per-worker database that id is three digits, so "232" was found inside the tenant's own UUIDs, times and decimals. A numeric id now counts only as a whole number (a JSON number, or a digit run not part of a longer number, word or decimal); ids, slugs and addresses keep the substring check. A self-check pins both sides. - The 5,000-keyword import waited 170 s for metrics; the roadmap's limit is 5 minutes, so it waits 290 s. - The web job's unit tests took 11 to 19 minutes today and a 20 minute limit cut one run off; the limit is 30.
jadhavgaurav
changed the title
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Why
PR #28 runs the backend suite on 4 workers. The first runs failed checks that measure the runner, not the code:
time.process_time(); same limitstest_a_list_holds_nothing_of_another_workspacefor/meand/costs("shows 232")serp_snapshot_idwas searched as a substring; on a fresh per-worker DB it is 3 digits and occurs inside the tenant's own UUIDs, times and decimalswebjob cancelled at 20 minProof
test_a_numeric_id_counts_as_shown_only_as_a_whole_value:232inside a UUID, a timestamp,0.232000and2320is not a leak;232as a number, a string, a key, a path or query segment,kw-232,id:232,snap_232,(232)and a sentence is.tests/security/test_cross_tenant.pyon 4 workers: 269 passed. With the import acceptance test: 270 passed.The check that matters is this PR's own
backendandwebruns.