morningman opened a new pull request, #64580:
URL: https://github.com/apache/doris/pull/64580

   INSERT OVERWRITE ... PARTITION(*) auto-detect into an auto-partition table 
whose source rows match no existing partition 
(enable_auto_create_when_overwrite=false) correctly fails every time. But with 
multiple parallel sink instances the surfaced error message races between the 
semantic "Cannot found origin partitions in auto detect overwriting" 
(vrow_distribution.cpp) and the collateral "no partition for this tuple" 
(vtablet_finder.cpp) that gets wrapped into a node-channel "add batch ... 
status isn't ok" error. One instance cancels the shared node channels, the 
other instance's in-flight add_batch then reports the wrapped variant; which 
one FE reports to the client is non-deterministic.
   
   The test pinned the strict "Cannot found origin partitions" message, so it 
failed intermittently when the wrapped variant won the race (observed in 
CloudP0 build 971274: same SQL run 3x, 2 pass 1 fail; single-sender loads are 
deterministic, >=2 sender loads race). Relax the assertion to accept all 
equivalent failure messages, matching the sibling test 
insert_overwrite_auto_detect.groovy, while still requiring the statement to 
fail (guards against silently creating a partition or succeeding).
   


-- 
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