Is anyone else seeing 100% cpu usage with 3.10 beta?
After a few days of us, with a small user base, openfire service goes to 100%, and and stops accepting connections… I have to kill the task in task manager. After restarting, everything seems fine for a few days. This is the first time ive experienced this issue with any openfire build. Logs look normal. Any thoughts?
Running on windows 2008 r2 with bundled jre 1.7.0_72
Looks like there may have been a jar or plugin left over from the upgrade that was causing the problem. I removed openfire, and deleted any leftover files/folders. After a fresh install, the problem seems to have been resolved.
Interesting. Will keep in mind when upgrading. Though i usually do in-place upgrades and everything was fine so far. But i delete built-in jre folder and let it use system’s Java 8.
I saw this as well right before the beta. I rolled back to stable. I suspected a plugin as well. I’m on different hardware and Java version than you - OS X 10.10, Java 1.8_11.
I have this exact same issue. I have tried uninstalling it and cleaning out basically all data but the config files, and reinstalling. The only plugin I use now is ofmeet which we need. Can’t find anything in the logs but this:
"2015.02.20 10:42:20 org.jivesoftware.openfire.net.SocketReader - Closing session due to incorrect hostname in stream header. Host: [our server IP]. Connection: org.jivesoftware.openfire.net.SocketConnection@15f8bdf socket: Socket[addr=/[same server IP],port=60170,localport=5269] session: null "
This repeats once a minute. Unsure if it’s related.
Have you tried recent builds? It should have newer Jetty version and a few other patches, though i don’t know if they are related to this issue. I will file a ticket anyway.
The current builds may have fixed this. Ignite’s instance of Openfire was pegging CPU as well, but the current nightly build from two days ago has yet to run away (knock on wood).