Repository navigation
test(coroot): bind consumer evidence to the verified parser - #4935
Conversation
Validation at bbec9ef: reproduced #4926’s actual v1.4.2 module selection, including its real checksums. The CRI severity and deprecation consumers now fail before reporting compatibility. Restoring v1.4.0 passes the native Helm 4.2.0 chart-and-consumer suite under race detection; affected vet and lint pass. Strict Renovate validation passes with cached validator 44.155.3. The full root suite is still running with bounded package parallelism; no full-suite success is claimed yet. The profile was independently checked against coroot/coroot-node-agent commit ee62018e73b90b1549f5b81ba0d3d6e28851a1df. It is a recorded source profile, not a live deployment attestation. A replacement or an unknown/unreadable selection cannot provide evidence. Proposals stay enabled; only unattended adoption is disabled for this test-only dependency. Production logging and thresholds are unchanged. Correctness/security and simplification self-review found no additional actionable issue. Current-head independent review and CI remain required. |
@coderabbitai full review |
|
|
Warning Review limit reachedYou've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository. Next included review available in 29 minutes. View limit details
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
CodeRabbit declined this head because its included review limit is reached (comment 6199142347). No paid capacity is requested. @codex review |
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
Codex Review: Didn't find any major issues. You're on a roll. Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
Full local-suite result: FAILED, not green. The run completed in roughly nine minutes with bounded package parallelism, but TestCurrentTalosNodesSkipTalosAPI failed its owned-process-group cleanup verification because the process-group read was killed. Other packages completed, including the new consumer gate. Root vet and introduced-findings lint pass. This local failure is not dismissed as harmless; exact-head hosted full-suite proof remains required before promotion, and the fixture cleanup failure needs its own diagnosis rather than a relaxed check. |
Ready at bbec9ef. The complete exact-head hosted suite and required checks passed in 38784219160, including the full Go helpers and the native consumer controls. Codex independently reviewed this head without findings (comment 6199181656); the fresh complete thread/comment scan has no unresolved or unaddressed finding, and GitHub reports CLEAN. As a user of the evidence collector, I exercised the real CRI severity and deprecation consumers: selecting the independently proposed v1.4.2 refuses the witness before parsing; restoring the independently verified v1.4.0 profile passes the native chart-and-consumer race suite. Replacements and unreadable selections cannot provide evidence. Strict Renovate validation passes; proposals remain enabled and production logging is unchanged. This is an offline source profile, not a live attestation. The earlier local full-suite failure remains FAILED and is tracked in #4936; its isolated case later passed but does not clear that failure. The independent full hosted run supplies the full-suite execution proof for this head. No timeout, fixture safety check or required gate was weakened. Merge-group validation and actual merge remain outstanding. |
Merged at 309e97d. Merge-queue CI 38816764494 passed: the complete 47-context read has no failing or unfinished check. The production deploy was skipped, as expected for this test-evidence-only change. The verified consumer witness stays on the recorded producer's logparser v1.4.0. The regression refuses the unsupported standalone v1.4.2 selection and replacements; restoring the verified selection passes the native consumer tests. Updates remain visible for paired review. Exact-head source CI and independent review were recorded before queue admission. No production parser version, severity classification, log filter, alert threshold or suppression rule changed. This closes #4934's offline evidence-correctness defect; it does not claim that live Coroot is clear. The separate local cleanup-probe failure remains tracked in #4936. |
Why
Our Coroot compatibility tests can silently switch to a parser that the verified agent does not use. Passing those tests would then overstate what we know about production.
What
Reject an independently updated or replaced parser before collecting consumer evidence. Keep dependency proposals visible and require a verified source refresh when deliberately updating this test witness; production logging is unchanged.
Fixes #4934