SEPURI-SAI-KRISHNA commented on PR #12595:
URL: https://github.com/apache/seatunnel/pull/12595#issuecomment-6117615259
Thanks @DanielLeens. Nothing outstanding on my side either, and I appreciate
the note that the javadoc does not overclaim; after rewriting it twice that was
the one thing I was unsure of.
**One prediction from my last comment came good, so closing it out.** I said
the `TestFilterRowKindIT` failure was #12116, fixed by #12216, and that the
sync would clear it rather than reroll it. At the merged head all four
`transform-v2-it` jobs are green, so that one is resolved rather than lucky.
**On the three failures, I narrowed the Cassandra one rather than leave it
to you.** Both land on the *first* assertion of their test:
```text
CassandraIT.java:134 Assertions.assertEquals(0, execResult.getExitCode());
testCassandra
CassandraIT.java:144 Assertions.assertEquals(0, execResult.getExitCode());
testCassandraMultiTableSource
```
So `expected: <0> but was: <1>` is the **job exit code**, not a row count:
the job never ran successfully on that container, and the test failed before
touching Cassandra.
The invocation pattern is the useful part. From the failsafe report, of the
nine invocations only `[2]` fails:
```text
testCassandraMultiTableSource{TestContainer}[1] pass [2] FAIL [3] pass
[4] pass
testCassandra{TestContainer}[1] pass [2] FAIL [3] pass
[4] pass
```
and `[2]` fails identically on **both** the JDK 11 and the JDK 17 leg,
taking 31.7 to 35.7 seconds across the four failing invocations rather than
hitting a timeout. Two independent runners failing the same engine invocation
the same way is not a race; it reads as that one engine not being able to run
this job on current `dev`. I did not identify which engine `[2]` maps to, so I
am not guessing at it.
Correcting myself on tracking before you spend time on it: no issue I can
find names `CassandraIT`, so that one is not a known flake. `DatabendCDCSinkIT`
**is** named, in #12482, among the signatures it lists as known flaky or
shared, though #12482 is a CI matrix proposal rather than a bug report for it.
I had drafted "neither is tracked" before reading #12482's body properly.
Not attributable to this PR either way: the diff is `BasicAuthFilter` plus
its two tests, `enable-basic-auth` defaults to `false`, and the Cassandra e2e
module submits jobs through the container CLI rather than the REST filter.
Also worth noting the matrix moved while this branch was behind: the
connector E2E legs now run JDK **11 and 17**, where earlier runs here were 8
and 11.
--
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]
