Coordinator Node Errors - Uh... I have no servers. Not assigning anything


I am seeing this error in the coordinator log “Uh… I have no servers. Not assigning anything…

Seems like it is happening here,

Has anyone encountered this error?

The historical node is running fine because I can see it memory mapping the segments.

This error is causing the middlemanager (Kafka Indexing Service) not be able to handoff the segments, and hence the ingestion is blocked.

I tried plenty of things but I am unable to figure the root cause for this error.

all druid services are on the same box and in their own docker containers. The disk is at ~63% use.

  • A few changes that I made were to turn off segment caching on the historical.

  • Changed the taskCounts

  • Following these changes, I started seeing this issue. Not sure if it is just coincidental or related.

~ Ankush