Eric, I certainly agree with what Joe said. I would also recommend checking in nifi.properties if you have a value set for:
nifi.monitor.long.running.task.schedule I recommend setting that to “9999 hours” In 1.14.0, we introduced the notion of a Long-Running Task Monitor. It’s generally very fast. Typically runs in 10s of milliseconds on my macbook. But it relies on JVM-specific code, and we’ve seen in some environments that can cause the UI responsiveness to be very adversely affected. We disabled the task monitor by default in 1.15, I believe, because of this. Thanks -Mark On Apr 19, 2022, at 3:44 PM, Joe Witt <joe.w...@gmail.com<mailto:joe.w...@gmail.com>> wrote: Eric When the UI isn't responsive it would be great to have a snapshot of: - CPU usage at that time - GC behavior/logging at/around that time. - IO Utilization around that time - NiFi Thread dump precisely during it and ideally also one after it responds again NiFi Restarting itself is very interesting of course. There should be more in the app log and bootstrap that will help illuminate the issue then. Thanks On Tue, Apr 19, 2022 at 12:42 PM Eric Secules <esecu...@gmail.com<mailto:esecu...@gmail.com>> wrote: By the way, I am running NiFi 1.14.0 and it looks like it keeps restarting itself. I am seeing this in the logs about once an hour. {"level":"INFO","thread":"main","message":"org.apache.nifi.NiFi Controller initialization took 4737354582168 nanoseconds (4737 seconds)."} On Tue, Apr 19, 2022 at 12:34 PM Eric Secules <esecu...@gmail.com<mailto:esecu...@gmail.com>> wrote: Hello, When my nifi system goes under high load the web UI becomes unresponsive until load comes down. Is there a way I can see what's going on (processor status summary, queued count, active thread count) when the UI is unresponsive? The logs are not showing any errors and the various repositories are all mounted to separate volumes with elastic capacity so I am sure that none of them ran out of space. Our monitoring shows bursts of CPU usage and memory use lower than normal. The logs show that the StandardProcessScheduler stops processors followed by starting them, but I never see logs related to the UI being ready to serve. It does this about once an hour. I see that the flow is slowly processing based on log activity and databases. How can I see what's going on when the web UI is not responding? Thanks, Eric
