DanielLeens commented on PR #11812:
URL: https://github.com/apache/seatunnel/pull/11812#issuecomment-5409207857

   @nzw921rx @goutamadwant Thanks for the quick rerun coordination — appreciate 
the effort keeping this moving.
   
   I want to flag an accuracy correction on the "all checks look clean" note, 
since a fresh sync-from-dev landed right after: the current head (`056033fe`, 
which merged in `dev`'s recent Oracle INTERVAL-type commit) still shows the 
apache-side `Build` check as failing, not green. I dug into the actual fork run 
behind it (`goutamadwant/seatunnel` run `32803825571`) and traced both failures:
   
   1. **`all-connectors-it-2 (8, ubuntu-latest)`** — `connector-couchbase-e2e` 
failed after ~3:25 min. This is the same pre-existing `CouchbaseIT` 
container-bootstrap crash we've been seeing across unrelated PRs since around 
2026-08-15 (the container's `containerIsStarting` REST call isn't covered by 
Testcontainers' startup-timeout retry). There's already an open fix for it 
(adds `startupAttempts=3` + a log consumer); until that lands, this class of 
failure is rerun-proof noise, not something introduced here.
   2. **`jdbc-connectors-it-part-6 (8, ubuntu-latest)`** — `connector-jdbc-e2e` 
failed on `JdbcHanaIT » Completion 
org.testcontainers.containers.ContainerLaunchException`, i.e. the SAP HANA test 
container failed to start. This PR's diff only touches 
`TaskExecutionService.deployTask`/`deployLocalTask`'s classloader 
acquisition/release bookkeeping in `seatunnel-engine-server` — nothing in the 
JDBC/HANA connector path, so this is an independent Testcontainers-level flake.
   
   (`unit-test (8, ubuntu-latest)` also shows `cancelled` in that run, but 
that's the workflow's fail-fast behavior kicking in once the two jobs above 
failed, not a real failure of its own.)
   
   To be clear: I re-checked `TaskExecutionService.java` and 
`TaskExecutionServiceTest.java` against my last approval (2026-08-24T06:45Z) — 
the only commit on top of that is the clean merge from `dev` (Oracle connector 
docs/code only), so the classloader-release logic itself is unchanged and still 
carries my approval. No source-side blocker here.
   
   @goutamadwant — could you do one more job-level rerun, this time targeting 
just `all-connectors-it-2 (8, ubuntu-latest)` and `jdbc-connectors-it-part-6 
(8, ubuntu-latest)` on the current head? Once those two are green (or land 
after the Couchbase startup-retry fix merges), this should be ready to go from 
my side.


-- 
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]

Reply via email to