DDtKey commented on issue #5108:
URL: 
https://github.com/apache/arrow-datafusion/issues/5108#issuecomment-1414475265

   > > And probably related to usage of unbounded channels.
   > 
   > Interestingly, the point of #4867 was in fact to remove the unbounded 
channels. I am looking more into this issue
   
   As I already said, my case it's not about sorting, so sorry for referring to 
it in this issue. It's somehow related to join-repartitions and channels 
(because it started to happen after getting rid of unbounded ones). I like that 
unbounded channels were refactored, but somehow it introduced an issue that I 
faced with.
   
   It returns correct error with `16.0.0` & just don't respect memory limit 
with `17.0.0`. 
   However I can't reproduce it on small file sizes with lower mem-limits (it 
works correctly for both versions - returns error), so I think this might be 
related to buffering.
   
   So [the code is attached above in the my 
message](https://github.com/apache/arrow-datafusion/issues/5108#issuecomment-1412879293),
 and I'm attaching file (about 1.3 GiB) - [GDrive 
link](https://drive.google.com/file/d/1112hJDYXhWc7z8SrmEQ2X_9JXLipFLaV/view?usp=share_link).
 It's actually artificial case, but it shows the some kind of regression is 
exists. 
   


-- 
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