After upgrading our gateway from 8.0.15 to 8.1.2, we’ve seen an increase in CPU usage of about 30%. Is this expected behavior? I’m hoping we’ve overlooked some config change that is causing the issue.
The server is running Ubuntu 18.04.2 LTS and is only used for ignition. A quick look at top confirms that java is the only process using a significant amount of CPU.
No, not expected.
Can you look at threads page in the gateway and see which threads are reporting higher percentages of CPU usage? and maybe get a few thread dumps?
We might be able to diagnose it here but usually it’s easier for support to take a look.
When I look at usage by system, it’s almost all “unclassified.” Most of the higher CPU threads are “milo-shared-thread-pool,” but no one thread seems to be using a huge amount of CPU. I’ve attached a few thread dumps.thread_dump20210209-155230.txt (369.6 KB) thread_dump20210209-155422.txt (441.7 KB) thread_dump20210209-155624.txt (375.6 KB)
Hmm, not really sure, might try support. I only see one thing happening on one of the threads you mentioned and it’s just monitored items being created (tags being subscribed to on the server).
That shouldn’t be too CPU intensive or long-lived, though if you have tags that don’t exist configured it’ll keep happening until they do. But nothing about that has changed in between versions.
Support ticket submitted. Thanks for your help!