stoty commented on PR #7517:
URL: https://github.com/apache/hbase/pull/7517#issuecomment-3615517117

   > The spotless issues will be fixed by #7507.
   > 
   > I think we'd better try to understand the code added in HBASE_27558, and 
see what is the correct way to fix this, instead of just adding a side logic to 
bypass the check...
   
   Reading the text of HBASE-27558 , it's not reset because the block limit was 
meant for quota enforcement, not for limiting the RPC response time / size, 
like the rest of the limits.
   
   I agree that this is not the cleanest design, but ScannerContext is the 
mechanism for we have for tracking resource usage  in StoreScanner and 
returning early, and adding a separate mechanism for quota enforcement would 
not necessarily be an improvement.
   
   We could perhaps generalize the behaviour to be per limit to make it more 
explicit.
   
   Today we are already treating the time limit differently, as it can not be 
reset by implementation,
   the batch limit is always checked between cells, and blockSize is never 
reset.
   so there are many implicit behavioural differences between how the limits 
are handled.
   
   Do you think making these explicit per-limit options help ? I'm not sure.
   Maybe we should just add more comments to clarify behaviour ?
   
   @bbeaudreault what's your take ?


-- 
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