Repository navigation
test: fix flaky CI tests and test against Weaviate 1.39 - #486
Open
dudanogueira wants to merge 7 commits into
Open
dudanogueira wants to merge 7 commits into
dudanogueira wants to merge 7 commits into
Conversation
There was a problem hiding this comment.
Orca Security Scan Summary
| Status | Check | Issues by priority | |
|---|---|---|---|
| Infrastructure as Code | View in Orca | ||
| SAST | View in Orca | ||
| Secrets | View in Orca | ||
| Vulnerabilities | View in Orca |
…st deterministic The journey test runs against a shared WCD cluster whose default for replication.asyncEnabled flipped to true, breaking every tests-without-auth job. That value is owned by the cluster config, so accept any boolean. The timeDecay boost test inserted all objects at Date.now() while decaying from 2024-01-01 over 35d, so every object's boost was ~0 and identical and the assertion only passed when float noise reordered ties. Spread createdAt around the origin so the boost reorders results deterministically. Fixes weaviate#485
The delete and create calls were fired without awaiting, so beforeAll could resolve before the 'test' role existed, or the delete could land after the create. assignRoles/revokeRoles then failed with 404 'role doesn't exist'.
1.38.18 includes the server fix for the backup status race where a poll could read FAILED without an error (weaviate/weaviate 895cd4d409, first in 1.38.10), which flaked the 'fail restoring backup for existing class' test.
Before 1.38.10 a poll could read a failed restore as FAILED with no error (fixed upstream in weaviate/weaviate 895cd4d409), so the error-message assertion flaked. 1.33-1.37 have no backport.
… 1.38.16 The concept and extension endpoints were removed in 1.38.16 and 1.39.6 (weaviate/weaviate#13083), so these tests 404 on the latest patches. Adds a requireLowerThan helper next to requireAtLeast.
…eaviate >= 1.39.3 1.39.3 added both fields to the RQ config (weaviate/weaviate#12502), so the schema journey's deep-equal failed on 1.39.9.
dudanogueira
force-pushed
the
fix/flaky-ci-tests
branch
from
October 5, 2026 14:49
7c89904 to
4e5eed7
Compare
This branch has not been deployed
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.
Fixes #485. Unblocks CI for #459 and any other open PR.
Journey test (
test/collections/journey.test.ts)Runs against a shared WCD cluster, not the version-matrix container, so it fails identically on every
tests-without-authjob. The cluster now defaultsreplication.asyncEnabledtotrue. That's server configuration, not client behaviour, so the test now accepts any boolean instead of pinningfalse.timeDecay boost test (
test/collections/query/integration.test.ts)All objects were inserted with
createdAt: Date.now()while the decay origin is2024-01-01with a35dscale, so every object's decay was ~0 and identical — the boost did nothing, andnot.toEqual(baseline)only passed when float noise reordered ties.createdAtis now spread around the origin (origin, +30d, +1y), so the object at the origin —red, which ranks last on vector distance alone — gets a real boost.Testing
Lint, prettier, and
tscpass. The boost suite is gated to ≥1.38 and my local server is 1.36, so it is verified by this PR's CI rather than locally.