DanielLeens commented on issue #13324:
URL: https://github.com/apache/seatunnel/issues/13324#issuecomment-6983655698

   Confirmed the swallowed-write/stale-decision mismatch with a focused Java 11 
probe on `89f91a89af15f33d97a942d8b65f4fc1b0c80bf5`.
   
   Steps:
   1. Give the real coordinator a final, non-restored `CompletedCheckpoint`; 
assert the `isCompleted()` control is true.
   2. Let the IMap return stale coordinator status `RUNNING`, and make 
`set(FINISHED)` throw an injected `IllegalStateException`.
   3. Invoke actual `updateStatus(FINISHED)`. It logs the failure and returns 
normally.
   4. Invoke actual `isNoErrorCompleted()`: it returns `false` while the stale 
entry remains `RUNNING`.
   5. Change only the IMap read result to `FINISHED`: `isNoErrorCompleted()` 
returns `true`.
   
   [Executable Java probe and exact build/run 
commands](https://github.com/apache/seatunnel/issues/13313#issuecomment-6983599115):
 see `issue13324()`.
   
   The exception here is deliberately **non-retryable**. This confirms the 
production catch/decision mismatch, not retry exhaustion during a real 
Hazelcast outage. I have not reproduced a multi-node failover or duplicate 
external sink side effects.
   


-- 
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