> Hmm, good point. I thought it makes sense to kill them because the supervisor 
> is currently killing running tasks if they are not allocated to 
> middleManagers yet.

I think killing unassigned running tasks does make sense, since if a task 
hasn't even started running yet, it has no hope of catching up so we should 
just cancel it. However, if there is some risk that the task actually is 
running but the supervisor just doesn't know where yet, this killing might be 
over-eager. If that's the case I think it'd be an issue for a separate PR 
though.

[ Full content available at: 
https://github.com/apache/incubator-druid/pull/6206 ]
This message was relayed via gitbox.apache.org for [email protected]

Reply via email to