You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
test(collections): target the actual WAL format in split recovery - #5391
The split-insert recovery fixture assumes its mutated WAL is fixed v3. Seeding a closed fixed predecessor does not ensure that: Open advances to a new writer, whose container follows the existing medium policy. This patch locates the actual closed acknowledged segment and injects corruption or a torn tail in its real fixed/stream layout.
Reuse the strict V2 reader, offsets and terminal-tail inspector; retain fixed extent/witness geometry and truncate genuine stream bytes after complete frame identity. Remove ineffective pre-open seed plumbing. No product format, medium policy, durability, journal or recovery implementation changes.
Validation:
Exact current head eacbfd1/tree6b0fc6bd9f6a22621c85526342727509fc3bcafe received independent complete source review plus review of its four-space/inventory correction. Inventory refresh/check passes at17workflows/63members, with only discovery binding changed. Exact-head Codex review completed with no findings.
Current TreeDB41773223457 Linux normal and race raw artifacts each prove3tops/19namedoutcomes/16leaves run/PASS and collections packagePASS, bound to actual synthetic840559dbdaa1a0862e2499eee6f42d67fef031e2 whose tree equals the candidate. Existing accepted-prefix, no false row/LSN advance, retry identity and reopen assertions remain unchanged. Focused selector: ^Test(SplitVectorInsertTerminalTailRecoveryV1|SplitCollectionRecoveryActualContainerFaultsV1|SplitCollectionRecoveryFaultRefusalsV1)$.
macOS and Windows each prove16 direct control outcomes PASS; the existing Linux-only namespace guard skips the original recovery parent, so its two children do not enter on those platforms. macOS real fixed/stream controls enter; Windows stream controls are positive, while fixed cases assert unsupported-factory refusal, not positive fixed-format recovery.
Corrected-head hosted Go1.26.10 Format41773218504 passed. All8 current-head Actions workflows passed, including all38 TreeDB jobs and required app15368 gate126396180941. All54 check contexts passed, including the completed exact-head CodeRabbit review with no actionable comments. Its generic docstring-coverage warning is advisory; no repository configuration applied and the check concluded SUCCESS. The existing branch-policy and issue gates remain unchanged. All local Go checks remain UNRUN; no local host runtime is admitted. Previous head10861 Format failure is retained.
Fixture cost adds a bounded image (max64MiB), strict scan and control oracle copies; fixed writes retain reservation. Sparse capacity refusal is counted in logical output. Existing resource/deadline controls stay unchanged. No public performance claim.
Semantic source review SHA256 ef4bf0ff13f1b52c53e53f53a5bea9cf34f1fd00b45035b8ca52e07b846381f6; exact corrected-head review efa531fcb2af486e755aa4f29ec3735d7554d7492ed1c1b75c92e8f3edf7ab44; named/platform entry audit b3a5afd4da52a056cba24c59c0d46035730631645f6e9bb280c974aaa8adc5a4. Current required CI, independent source/entry review and exact-head Codex and CodeRabbit reviews are complete. Final live policy/base/head/threads/merge-tree readiness checks govern integration. #5386's distinct Windows startup timeout and broader performance acceptance stay open.
No actionable comments were generated in the recent review.
Walkthrough
The recovery tests now use production-selected WAL containers and mutate frames identified in the actual WAL image. They cover stream and fixed-container layouts, verify recovery-reader outcomes, and reject unsupported or invalid inputs without modifying the image.
Changes: Split-insert recovery tests
Layer / File(s)
Summary
Use standard production fixtures TreeDB/collections/column_vector_graph_typed_column_test.go, TreeDB/collections/vector_index_partition_live_v1_test.go, TreeDB/collections/vector_partition_split_insert_recovery_v1_test.go
The fixture helpers no longer pass pre-open setup callbacks. The terminal-tail recovery test uses the standard production fixture.
Locate and damage terminal frames TreeDB/collections/vector_partition_split_insert_recovery_v1_test.go
Frame selection checks the durable command identity and terminal position. Shared helpers locate frames and apply corruption or relaxed-tail damage to stream and fixed-container images.
Exercise container formats and refusal cases TreeDB/collections/vector_partition_split_insert_recovery_v1_test.go, .github/ci/ci_impact.json
Tests create WALs with maintained writers and check mutation outcomes and refusal cases. The CI impact inventory hash also changes.
snissn/gomap#5224: Adds fixed-WAL reader and frame-inspection support used by this PR’s split-insert recovery tests.
Merge Risk: Minimal · up to eacbf
The test-only changes have no established merge-blocking defect. Complete the required exact-head CI before merging.
Pre-merge checks: 4 · 1
Failed checks: 1 warning
Check name
Status
Explanation
Resolution
Docstring Coverage
Docstring coverage is 44.44% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff with cyclomatic complexity of at least 2. Documenta…
Write docstrings for the functions missing them to satisfy the coverage threshold.
Passed checks: 4 passed
Check name
Status
Explanation
Description Check
Check skipped - CodeRabbit’s high-level summary is enabled.
Title check
The title "test(collections): target the actual WAL format in split recovery" matches the main change. The changeset rewrites split-insert recovery tests to locate and mutate the real WAL container (f…
Linked Issues check
Issue #5389 requires test-fixture-only fault placement in the actual closed, acknowledged WAL segment. The recovery diff removes pre-open fixed-WAL seeding and locates the acknowledged frame with `New…
Out of Scope Changes check
The changes remain within issue #5389. The main change is in vector_partition_split_insert_recovery_v1_test.go. The two helper-file changes remove the now-unused pre-open setup callback that enabled…
Full details: Title check
Explanation
The title "test(collections): target the actual WAL format in split recovery" matches the main change. The changeset rewrites split-insert recovery tests to locate and mutate the real WAL container (fixed or stream layout) instead of pre-creating fixed WAL segments, and removes the ineffective pre-open setup plumbing. The title is concise, specific, and follows a conventional commit style. It does not list every detail, which is expected.
Full details: Linked Issues check
Explanation
Issue #5389 requires test-fixture-only fault placement in the actual closed, acknowledged WAL segment. The recovery diff removes pre-open fixed-WAL seeding and locates the acknowledged frame with NewReaderWithOptions, ReadCommandFrameV2, offsets, and terminal-tail inspection. It validates frame identity, bounds, fixed extent geometry, stream layout, durability, and refusal cases. It preserves the existing recovery, retry, lifecycle, and no-false-advance assertions. The added controls exercise fixed and stream writers, corruption, incomplete frames, and unsupported fixed creation. The reported focused normal/race tests and hosted validation passed. No product implementation changed.
Full details: Out of Scope Changes check
Explanation
The changes remain within issue #5389. The main change is in vector_partition_split_insert_recovery_v1_test.go. The two helper-file changes remove the now-unused pre-open setup callback that enabled the ineffective seed path. The CI inventory change updates discovery metadata. The diff contains no product format, medium-policy, durability, journal, recovery, workload, or unrelated #5386/#5373 changes.
Full details: Docstring Coverage
Explanation
Docstring coverage is 44.44% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff with cyclomatic complexity of at least 2. Documentation-only edits and deletions within surviving functions are included; unchanged diff context and fully deleted functions are excluded. Analyzed 9 functions across 2 files. (1 skipped: 1 unsupported.)
Resolution
Write docstrings for the functions missing them to satisfy the coverage threshold.
Finishing Touches
Generate docstrings
Commit to this branch
Create a new PR
Generate unit tests (beta)
Commit to this branch
Create a new PR
Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts
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.
Included review availability: This review used your included allowance. Your plan provides up to 1 included review per hour; 0 remain after this review.
Please review exact current head eacbfd1 (tree6b0fc6bd9f6a22621c85526342727509fc3bcafe). Focus on actual closed acknowledged WAL target identity, fixed/stream physical offsets and witness geometry, fail-closed refusal controls, preserved recovery/retry assertions and bounded fixture cost. This is test-only; no product medium-policy change. The preceding Format failure is retained; corrected-head hosted Go1.26.10 Format passed. Full current-head normal/race/affected-scope CI remains running and independently blocking.
The reason will be displayed to describe this comment to others. Learn more.
Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota. Try requesting a review at a lower effort level.
Coordinator merge verified on GitHub: 8e1133085b4052e344a73063f28c2ab142485c73, tree 6b0fc6bd9f6a22621c85526342727509fc3bcafe, parents actual main b2f9 + qualified eac. The actual merge tree equals the qualified candidate/synthetic tree. All54 exact-head checks passed (all38 TreeDB jobs), Codex and CodeRabbit completed with no actionable findings, threads empty; Copilot quota exhaustion was not counted as review. Live rules/head/base were rechecked before expected-head merge.
Automatic actual-main qualification is now running: TreeDB41777285003 plus the six other push workflows. Dependent publications wait for those gates. Original failure/format history and platform-entry limitations remain retained; no public performance claim.
Final readiness SHA256 7b5203ede99563e7d23666efe7d74de42f2d1751a4c065d68cfee0e6384e4615; actual merge receipt 626d4e04d4043c6a7e9aea67b34ec82761ffc32ed3770d29f447bf620ecb99b1.
Actual-main qualification is BLOCKED: windows-nativewire126400474982 failed on actual8e113308/Go1.26.10 Windows. The only functional failing child is TestFixedPeerTCPNativeCollectionCatalogGuardInterpositionV1/bson: fixed_peer_native_collections_v1_test.go:395, submit guard-create returns raftcluster: commit ambiguous joined with HTTP body/context deadline. Child4.46s,parent5.51s; JSON childPASS1.06s; package411.982s. No WARNING: DATA RACE.
This is distinct from #5386's earlier sparse-catalog30s setup timeout and #5373's separately owned Raft failure. Cause remains INCONCLUSIVE. Original artifact13298166791 ZIP SHA256 34710e3fe772926a5aaa5b92f8869481ba0de7d6ec5897d52025e0755c4ac512, rawJSON e336838497e3a76dea13e5698abe63aa604a068c182a1b4d848e451ff3b4667c retained. Existing source owner has one bounded actual-input/call-path diagnosis; no unchanged rerun, timeout/skip/gate change or product repair selected. Other post-main jobs are still running. Dependent publications stay held.
Actual-main post-merge qualification is terminal: run41777285003 has38 jobs,36PASS; windows-nativewire126400474982 and aggregate126409827568 FAIL. All six other automatic push workflows PASS. The sole functional failure is BSON guard-create/body-read deadline; cause/server outcome remain INCONCLUSIVE, distinct from #5386/#5373. No unchanged rerun.
The owning recovery race shard126400474922 did enter and pass all3 parent tests/16leaves/19outcomes on actual8e. Selected-event audit SHA d494cb2035e4c8146141922650928d9d6757ac20c7569ab508eda87f08068a06. This preserves #5391's merged and exact-head/owning evidence in its actual scope; it does not make default-main green. Final post-main receipt SHA624acd7327d315c0220b5db4a9cad2fa7617cdeb8ae4ffd7fa6d659d89067234.
One bounded, independently reviewed client-trace diagnostic is now #5392/#5393 at f696b71/tree91239e, with one fresh exact-head hosted qualification block. It changes no product code, RPC/deadline/assertion or transport/body and is not a claimed repair. A green diagnostic run does not automatically waive the original blocker. #5385/#5390/#5386 remain held.
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
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.
The split-insert recovery fixture assumes its mutated WAL is fixed v3. Seeding a closed fixed predecessor does not ensure that: Open advances to a new writer, whose container follows the existing medium policy. This patch locates the actual closed acknowledged segment and injects corruption or a torn tail in its real fixed/stream layout.
Reuse the strict V2 reader, offsets and terminal-tail inspector; retain fixed extent/witness geometry and truncate genuine stream bytes after complete frame identity. Remove ineffective pre-open seed plumbing. No product format, medium policy, durability, journal or recovery implementation changes.
Validation:
^Test(SplitVectorInsertTerminalTailRecoveryV1|SplitCollectionRecoveryActualContainerFaultsV1|SplitCollectionRecoveryFaultRefusalsV1)$.Semantic source review SHA256 ef4bf0ff13f1b52c53e53f53a5bea9cf34f1fd00b45035b8ca52e07b846381f6; exact corrected-head review efa531fcb2af486e755aa4f29ec3735d7554d7492ed1c1b75c92e8f3edf7ab44; named/platform entry audit b3a5afd4da52a056cba24c59c0d46035730631645f6e9bb280c974aaa8adc5a4. Current required CI, independent source/entry review and exact-head Codex and CodeRabbit reviews are complete. Final live policy/base/head/threads/merge-tree readiness checks govern integration. #5386's distinct Windows startup timeout and broader performance acceptance stay open.
Fixes #5389. Parent #5225.
Summary by CodeRabbit