SEZ9 commented on PR #12298:
URL: https://github.com/apache/seatunnel/pull/12298#issuecomment-5697776151
Correction to my previous comment on `NebulaGraphIT`: I described it in a
way that implied a consistent, cross-the-board regression, and the per-leg
numbers do not support that framing. The accurate picture, from the job
conclusions of every attempt:
| | JDK 8 | JDK 11 |
|---|---|---|
| Before the rebase (base `0d9f9e230`, pre-#11201) — this PR and #12299 |
4/4 pass | 4/4 pass |
| After the rebase (base `75b60fa14`, post-#11201) — this PR | 4/4 pass |
**0/4 pass** |
| After the rebase — #12299 | 4/4 pass | 2/4 pass (failed attempts 1–2, then
passed 3–4) |
So it is **JDK 11 only** — JDK 8 has never failed once — and it is
intermittent rather than deterministic: on #12299 the same commit failed twice
on JDK 11 and then passed twice. My earlier comment omitted both of those
facts, and "well-supported hypothesis" was too strong a label for the
`${testcontainer.version}` explanation on this evidence.
What I think survives: 8 JDK 11 legs after the rebase produced 6 failures,
against 0 failures in 4 JDK 11 legs before it, always at the same line
(`NebulaGraphIT.java:109`, the `adminPool.init(...)` assertion after the three
containers start). That shift is worth a maintainer's attention as a lead, but
the sample is small and confounded — the two bases differ by more than #11201,
and fork runner load differed a lot between the two rounds. It is not
established that #11201 caused it, and I have not bisected.
None of this changes the conclusion for this PR: `NebulaGraphIT` exercises a
graph-database connector and cannot be reached from a change to
`JettyService`'s constructor and one `HttpConfig` field, and it passes on JDK 8
with the identical commit.
For contrast, the other red job really is unambiguous.
`OpengaussCDCIT.testAddFieldWithRestore` has now failed on **24 of 24** legs —
3 runs × 4 attempts × 2 JDKs — across this PR and #12299, before and after the
rebase, without a single pass.
I am filing separate issues for both so they are tracked outside these PRs,
and will link them here.
--
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]
