potiuk commented on issue #35938:
URL: https://github.com/apache/airflow/issues/35938#issuecomment-1956742007

   BTW. I am not an expert in reading those graphs and I have no idea what 
kinds of memory it is and what it means/ what does it mean but it does seem you 
have an interference of some kind. There are at least few level downs you can 
go. Your pods are running on vms that are running on physical machines and your 
k8s is running on all of this and decides to allocate resources etc. so maybe 
that is some kind of interference - betwen the PODS that are running on  the 
same machine for example.
   
   Recently for example someone reported memory issues with airflow and turned 
out that their vms were set to never overcommit memory: 
https://github.com/apache/airflow/discussions/36891#discussion-6094121 - and 
that basically meant that when Pod was allocated on a machine, a memory needed 
were allocated even if it was not used and that did not show in the reports.
   
   Maybe you have similar issues but in order to do that you need to deep dive 
intoy understanding what happens in your systme, I am afraid - no 
airflow-specific knowledge is helpful here.
   
   
   


-- 
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