SEZ9 commented on issue #13319:
URL: https://github.com/apache/seatunnel/issues/13319#issuecomment-6992861970

   Thanks for the detailed report and for the reproduction, which confirms the 
`NullPointerException` from `LinkedBlockingDeque.add(null)` in 
`submitThreadShareTask`. It also shows the latch left at 1, the second task not 
initialized, and the tracker's completion future still incomplete. As noted 
there, the deployment-context publication and redeploy part of the chain has 
not been exercised at runtime yet.
   
   @SEPURI-SAI-KRISHNA thanks for volunteering, a PR is welcome. Your plan 
covers the main point: filtering nulls before `add` is not enough on its own. 
Every task that is not submitted also has to call `taskDone`, so the completion 
latch reaches zero, `finishExecution` runs, and the group reports FAILED with 
the original `init()` error instead of the NPE.
   
   For the regression test, in addition to what you listed (no NPE, second task 
not initialized, group FAILED with the injected error), please also assert, if 
feasible, that:
   
   - the location is removed from `executionContexts` once the group finishes;
   - a second `deployTask` for the same location is not skipped.
   
   That would cover the part of the report that has not been verified at 
runtime.
   
   Leaving `deployLocalTask` unchanged and deferring that to #12218, as you 
propose, sounds reasonable for this change. Please note that scope boundary in 
the PR description.
   
   <!-- streview-comment:2439 -->


-- 
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