DanielLeens commented on PR #11545:
URL: https://github.com/apache/seatunnel/pull/11545#issuecomment-5997474037
Correction to my earlier comment on this PR (the one that pointed to run
37317968844), @SEZ9, plus the fix.
I said the unit-test lane no longer repeats the six module flags and that CI
therefore exercises the `surefire.module.args` pom property. That was wrong for
modules that define their own surefire `argLine`. Fork run 37317968844 failed
`unit-test (17)` on ubuntu and windows in `connector-lance` (`LanceSinkTest`,
`LanceCatalogTest`, `FragmentConverterTest`) with `InaccessibleObjectException:
Unable to make field long java.nio.Buffer.address accessible: module java.base
does not "opens java.nio"` (Arrow `MemoryUtil`). Cause:
`seatunnel-connectors-v2/connector-lance/pom.xml` sets
`<argLine>${surefire.jvm.args}</argLine>` itself, so it never sees
`surefire.module.args`, and the lane's `-Dsurefire.jvm.args` CLI override also
replaces that module's own JDK-activated profile value. Before `4ce5edfba` the
repeated flags in `SUREFIRE_JVM_ARGS` were covering for this; I removed them as
redundant and uncovered it.
Fix: `c284df271` restores the flags in both unit-test steps and rewrites the
comments to give the real reason. Head is `d1794e8fb` (also picks up #12529
from `dev`, no conflict, 0 behind). New fork run:
https://github.com/DanielLeens/seatunnel/actions/runs/37331927562. It has only
just started, so I am not claiming any result.
What this means for F6/F8, stated plainly: the unit lane repeats the flags
again, so it does not by itself prove the pom property. The property is still
the only source on the lanes that pass no `-Dsurefire.jvm.args` (the
integration-test lanes). I did not touch `connector-lance/pom.xml`; making that
module inherit `surefire.module.args` would be a pom change, which I would
raise separately.
For the record, the other red jobs in that run were not caused by this PR's
changes: `all-connectors-it-2 (11)` died on an HTTP 429 downloading the Maven
Wrapper, and `edge-agent-it (11)` hit `IllegalStateException: Node failed to
start!` after 300s in `JobMasterMasterFailoverResourceTest`. I read those from
the logs but have not yet seen them pass on a rerun.
--
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]
