This message was deleted.
# general
s
This message was deleted.
a
It would be rejected.
l
Ok my problem is that if I have an middle manager with say 8 slots and I schedule msq tasks then sometimes I end up with 8 tasks of type
query_controller
running. Then there is no more room for the actual
query_worker
tasks. Is there any way to prevent this? Thank you for the help.
a
https://github.com/apache/druid/pull/12221 - Introduces a config to restrict the native batch supervisor task slots
I'm checking if there's something similar for MSQ controllers
You would need to divide your middlemanagers into tiers and then you could say that the MSQ controller tasks have an affinity to one of them and that way, the worker tasks would have a certain number of slots available on the other tier
l
Thanks a lot! I think I now have a good understanding of what I need to do!
👍 1
m
Hey @lukas don't you still need to deal with an ingestion task being rejected? I'm not sure if you get an exception, curious how to architect for that...