No new auto-compaction tasks have been created by the coordinator in the last ~40 hours. This leads to segment count increase substantially.
After debugging found that there is a long running compaction task in running state. Needed to kill it manually for auto-compaction to resume. The coordinator process is waiting for long-running compaction tasks to complete before submitting new compaction tasks.
Even when taskSlots are empty this issue was seem. We are running compaction task in separate tier.
Druid is on 25.0.0 running on kubernetes.
No new auto-compaction tasks have been created by the coordinator in the last ~40 hours. This leads to segment count increase substantially.
After debugging found that there is a long running compaction task in running state. Needed to kill it manually for auto-compaction to resume. The coordinator process is waiting for long-running compaction tasks to complete before submitting new compaction tasks.
Even when taskSlots are empty this issue was seem. We are running compaction task in separate tier.