We have a process that publishes volatile messages to a UM connection alias. Triggers are configured to be concurrent. My question is how do I determine the optimal number of max execution threads for the trigger? Can these threads be identified in a JVM thread dump? I see NirvanaReadThread threads in thread dumps but not sure if these are the trigger threads.
Yes, we’re expecting fairly heavy volumes on this trigger, probably up to around 2000 TPS max. Although the issue now is that when volume is increased the Nirvana nodes disconnect from the cluster and occasionally crash.
I would say please open a SR with SAG tech support and follow up since getting the nodes disconnect is not normal and so they can troubleshoot more and advise some resolution.