wenzhenghu commented on PR #66006:
URL: https://github.com/apache/doris/pull/66006#issuecomment-5075688854
### P0 Regression build 1005119 — analysis of the 3 remaining failures
This build ran on commit `9a956e8` (which includes the
`visitCreateScheduledJob`/`visitAlterJob` cast fix). Notably,
**`job_p0.test_base_insert_job` no longer fails**, confirming the cast guard
works. The remaining 3 failures are all unrelated to this PR:
| Test | Root cause | Related to this PR? |
|---|---|---|
| `nereids_rules_p0.pkfk.eliminate_inner` | Shape-plan baseline drift on
master: expected `cast(f as ...)` vs actual `cast(fkt_not_null.f as ...)` —
only a column-alias rendering difference. The `.out` baseline was last touched
by #56802 ("add table alias for catalog relation", already in this PR's base),
and this case has a history of stabilization fixes (#64957, #64457, #30507). It
fails on the master baseline as well. | No — this PR does not touch the
optimizer or plan-shape generation. |
| `export_p0.test_export_basic` | Assertion on the cancelled export job's
error message (`res[0][10].contains(exception_msg)`). Fails in the exact same
way as the previous build 1004852 (before the cast fix). | No — this PR does
not change any export logic. |
| `export_p0.outfile.s3.test_outfile_with_different_s3` | `Nereids cost too
much time (56s > 30s)` — all plan stages show `-1`; the query stalls on
external IO when listing/reading the parquet file from Tencent COS
(`cos.ap-beijing.myqcloud.com`). Environment/network flakiness in CI. | No. |
**Conclusion.** This PR's changes are limited to FE log printing and
audit-SQL masking (`ai.api_key`), plus the cast guard for job DML. None of the
3 failures above is caused by these changes. Suggest re-running P0 Regression;
if `eliminate_inner` keeps failing, it indicates a broken master baseline (it
would fail for every PR) that should be fixed separately on master (refreshing
the `.out` baseline or the alias rendering) rather than in this PR.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]