jakubmatyszewski commented on PR #14934:
URL: https://github.com/apache/druid/pull/14934#issuecomment-1708132309
> thank you for your contribution. can you elaborate a bit on following
>
> > but since MM instances that I use are created by k8s statefulsets it
wasn't consistent enough.
>
> what is not consistent?
For statefulset to scale down we need to take into consideration that:
> When Pods are being deleted, they are terminated in reverse order, from
{N-1..0}.
But with existing `fillCapacity` overlord strategy it doesn't care on what
host ordinal index it accumulates tasks - it **just** checks which
middlemanager instance has highest `currCapacityUsed` value. So in scenarios
when there are multiple middlemanagers with same amount of `currCapacityUsed`
it seems like this strategy picks middlemanagers at random.
--
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]
