andygrove opened a new issue, #25650:
URL: https://github.com/apache/datafusion/issues/25650

   ### Is your feature request related to a problem or challenge?
   
   When running in containers (e.g. in k8s), if a process tries to allocate 
more memory than is available then the process will get OOM killed. This 
happens in downstream projects using DataFusion because the memory pool / 
accouinting is not accurate.
   
   This is a problem in Comet for example. See details at 
https://datafusion.apache.org/comet/contributor-guide/memory_management.html
   
   Let's use this issue to discusss potential solutions and approaches.
   
   One idea is to tie into the actuall allocator (e.g. jemalloc) so we can see 
actual allocated bytes. We could use that information in diferent ways:
   
   - log the actual usage alongside reported memory pool sizes so that we can 
at least see where they drift
   - implement a memory pool that rejects allocations if allocated bytes is 
over some limit
   
   Related: https://github.com/apache/datafusion/issues/22758
   
   ### Describe the solution you'd like
   
   _No response_
   
   ### Describe alternatives you've considered
   
   _No response_
   
   ### Additional context
   
   _No response_


-- 
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