GitHub user yrenat added a comment to the discussion: Proposal: Automatically 
terminate idle Kubernetes Computing Units

Thanks for the comments. I just checked the code. Completed execution results 
are stored safely in Iceberg/VFS and are not deleted by CU garbage collection. 
However, you are right that the current result retrieval path does link to the 
original/live CU, so returning later after that CU is Garbage-Collected may 
still fail from the UI/API perspective. 

Some fixes that come into mind: 

1. Add an execution-result retrieval path that is keyed by eid, with no need 
from the CU side
2. Serve historical results from an always-on service, or route them to any 
live service that can read shared DB/Iceberg storage, not necessarily the 
original Garbage-Collected CU.
3. Update frontend execution-history/result viewing so selecting an old 
execution sends the eid explicitly.

Also happy to hear your inputs 
 



GitHub link: 
https://github.com/apache/texera/discussions/6264#discussioncomment-17630877

----
This is an automatically sent email for [email protected].
To unsubscribe, please send an email to: [email protected]

Reply via email to