[8.0.6] Scheduler stops dispatching scheduled searches intermittently.

[sylim_splunk]

Our Splunk  SH cluster scheduler stopping, users complaining that alerts/scheduled reporting not running or processing. We disabled and enabled the scheduler on the captain but that didnt work. We decided to switch captaincy to another and that worked - scheduling/processing resumed.
Today we had reoccurrence but on a different Search head cluster - we switched captains and that remediated issue again. 

Version 8.0.6 and recent change - cascading bundle replication enabled around a month ago.

[sylim_splunk · Accepted solution]

Once scheduler stopped scheduling jobs and stuck with something it will not put logs in scheduler.log at all on the captain - which can be used for us to confirm it must be stuck with something.

Pstack collection would be helpful to address the issue.

After inspection on the pstack collected when the issue was noticed, it turned out to be caused by a deadlock situation by cascading bundle push and the fix is available in 8.0.7+ and 8.1.3+.

If you would like to confirm it is the case please open a support case with a diag and pstack outputs collected when it is outstanding/before restarting of splunk. SPL-200260.