You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This very much feels tied to the OOM framework that we are working on #7253.
The thought here is to make sure that the thread with highest priority has less of a chance to have its buffers spilled. We have some exec nodes that will make spillable and then do some compute, in a chain. If the chain is broken because we run out of memory, the buffers added by this high priority thread should probably be the last ones to spill.
This issue is to visit this as we work on the retry framework.
The text was updated successfully, but these errors were encountered:
This very much feels tied to the OOM framework that we are working on #7253.
The thought here is to make sure that the thread with highest priority has less of a chance to have its buffers spilled. We have some exec nodes that will make spillable and then do some compute, in a chain. If the chain is broken because we run out of memory, the buffers added by this high priority thread should probably be the last ones to spill.
This issue is to visit this as we work on the retry framework.
The text was updated successfully, but these errors were encountered: