andygrove commented on issue #3949:
URL:
https://github.com/apache/datafusion-comet/issues/3949#issuecomment-4270358440
I think I understand the issue now, and I now have additional logging which
seems to confirm.
Comet planning happens for the initial plan and again during AQE
re-planning. If Comet makes different decisions during the different phases of
planning then it can lead to mismatches in the plan.
In the follow log there are two exchanges with fallback reasons, but only
one correctly fell back.
```
: +- CometSinkPlaceHolder
: +- CometColumnarExchange [COMET: Stage contains a scan with
Dynamic Partition Pruning]
: +- HashAggregate
: +- AQEShuffleRead
: +- Exchange [COMET: Stage contains a scan with Dynamic
Partition Pruning]
: +- HashAggregate
```
--
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]
