Workflow [PR], commit [574a244]

Summary: ❌

AI Review

Summary

This PR updates the contrib/silk submodule pointer from 997641b85cec5792183cf1970ddfd79ff1a74f72 to 79a46061e9ee662053f4ad4090562e342628d03b (the current tip of ClickHouse/silk clickhouse-public). I reviewed the PR diff, current PR discussion, and upstream submodule delta between these commits; no unresolved correctness, safety, or integration problems were identified in the ClickHouse tree.

Final Verdict

• Status: ✅ Approve

@groeneai there are some failed checks - they cannot be affected by this change

@vadimskipin Confirmed — both failing tests are chronic flakes tracked in existing issues, fully unrelated to the contrib/silk bump.

PR scope: Single-line submodule pointer change (997641b → 79a4606), no other files touched.

Failing check: Only Unit tests (asan_ubsan, function_prop_fuzzer) — 1 of 167 checks. Two tests inside it:

Both started failing days before this PR was opened and hit hundreds of unrelated PRs — no causal link possible. Safe to ignore for merge purposes.

This was fixed by #105146. Let's update the branch.

@groeneai there is a new failure - what is this?

@vadimskipin The new failure is chronic — STID 3593-560a "Table has no columns" on Stress test (arm_msan), fully unrelated to the contrib/silk bump.

CIDB cross-PR evidence (30 days): 14 hits across 13 distinct PRs, only 1 master hit (2026-04-20). Affects all stress sanitizer variants (arm_msan / arm_tsan / arm_release / arm_ubsan / arm_asan_ubsan / amd_tsan). Recent unrelated PRs: #104965, #105114, #104079, #104440, #104496, #103862, #100141, #104299, #96844, #103246, #100752 — none touch contrib/silk.

PR scope reminder: contrib/silk submodule pointer change (997641b → 79a4606), no source files touched — no plausible path from a contrib submodule pin to a StorageMerge abort under server-side stress fuzzing.

The previous FunctionsStress.stress / AllTests failures from your last question were unblocked by @alexey-milovidov's #105146 merge (per his 2026-05-17 rebase directive), and they are no longer in the rollup. Once @azat's #104919 merges, the arm_msan stress flake will clear on rebase. The PR is safe to keep moving — no action needed on your side for this failure.