Hey guys, I’ve been running native batch ingestion for a while now and have had subtasks tasks randomly run indefinitely. The task logs indicate SUCCESS but when i go to the /druid/indexer/v1/task endpoint for the task, all the statuses there are RUNNING. Here are some extra details: • My druid cluster is on EKS with the help of the Druid operator • There is only one coordinator running and I noticed there’s no restarts from it • I can ping the middlemanagers from the coordinator instance and get results back fine • Restarting only the coordinator when this occurs does nothing to solve the issue • This issue occurs even if both main task and subtask are on the same middlemanager • The master task will continually log “Found [1] expired supervisor tasks” if it has a subtask running indefinitely • The logs of an indefinite subtask mention “Starting graceful shutdown of task”. The logs between a normal subtask and indefinite running one differ a bit because an indefinitely running one does NOT log this:

2023-02-09T06:15:58,602 INFO [main] org.apache.druid.java.util.common.lifecycle.Lifecycle$CloseableHandler - Closing object[org.asynchttpclient.DefaultAsyncHttpClient@2c846d55]
2023-02-09T06:15:58,604 INFO [main] org.apache.druid.java.util.common.lifecycle.Lifecycle - Stopping lifecycle [module] stage [INIT]
Finished peon task

The only resolution is to kill a task and when it restarts it finishes immediately. Is there any druid configuration I can set to automatically handle something like this?