flysen commented on issue #16016:
URL: https://github.com/apache/iceberg/issues/16016#issuecomment-5599650323

   **Independent confirmation on 1.11.0 / AWS Glue Catalog**
   
   We hit this exact zombie-coordinator signature during the same incident we 
described on #15293 ([our comment 
there](https://github.com/apache/iceberg/issues/15293#issuecomment-5598930908)) 
— triggered by a targeted consumer-group offset reset, not a storage failure or 
cluster rebuild.
   
   Across that ~17-hour incident we found the `"found no partitions assigned 
across all members, cannot determine leader"` WARN logged 88 times across all 6 
tasks — initially dismissed as benign, but it's the exact 
`hasLeaderPartition()`-returns-false-during-rebalance condition described here, 
which skips `stopCoordinator()` in `close()`. Matches @tongwai-wong-appier's 
1.11.0 confirmation: no storage failure needed, an ordinary rebalance is enough.
   
   One caveat: checking the `-coord` group's live membership mid-failure showed 
only one member at that moment, so we can't say this is the sole explanation 
for every failed round we saw — possibly a residual effect once the zombie 
itself is gone. Full evidence (thread dumps, ruled-out causes, timeline) is on 
#15293 if useful.
   
   +1 to @gdmachado's fix direction (remember leader partition at `open()`, 
stop unconditionally on it in `close()` rather than re-querying group state 
mid-rebalance).
   


-- 
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]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to