Troubleshooting high CPU load in tag-provider-subs-default

Looking for general advice on how to isolate the cause of high CPU load on this thread. Looking at the stack trace, I see a lot of methods having to do with on value change and change distribution. So I'm assume the issue is with the number of tag changes, or value change scripts.

We tried shutting down clients, designers, and SFC. We also tried temporarily deleting large sections of tags. None of this reduce CPU load on this thread. Further testing we have thought about doing would be to delete all tags and add them back in a few at a time until the culprit(s) are found. Another is to turn off OPC connections.

The thread is using 24% CPU, and the gateway is using 30-35% overall. Not terrible, but it is significantly higher that what I've seen on other gateways running similar projects. With some load from SQL the server is starting to have some performance issues.

Is there a better way to isolate the cause of the load other than the method I've been using?

Look at all of the tag events. Look for while loops first. Then add duration instrumentation with java.lang.System.nanoTime() at beginning and end, logging anything over a millisecond.

1 Like