Openfire 3.10.0 eventual crash

We have been running Openfire on CentOS 5 for about 7 years. Since upgrading to 3.10.0 our server will stop responding to new connections after about 3 days.

We have a very small install with about 15 clients at peak time. For the first couple days after a reboot all clients can connect without issue. After about 3 days existing connections work but new connections fail. So as long as everyone stays connected they are fine. Restarting openfire clears the problem.

Also, when the server stops responding to new clients, the “Client Sessions” tab will not load in the web interface. I think most of the other pages work fine.

Other details:

  • Centos 5 has all the latest updates

  • Most users use Psi as a client

  • We use LDAP to authenticate to our Zimbra mail system

That’s probably same issue reported below on the “100% CPU usage” thread. So far the only working solution is to downgrade to 3.9.3.

Maybe but I don’t see the high CPU usage. Actually my CPU floats between 100 and 99.7% Idle. I think top is using more CPU time.

I’m waiting for the issue to pop up again. Should happen some time Friday.

It could be the webchat plugin is missing a class. We had the same problem on a windows 2008 server and had to restart. We noticed in the logs the webchat plugin was missing a class (See below). I’m not sure if it is a permanent fix yet but I’ll update if it is. We lasted almost 5 days before it crashed.

  1. 2015.05.15
    16:23:08 org.jivesoftware.openfire.container.PluginManager
  • Error loading plugin: C:\Program Files (x86)\Openfire\plugins\webchat
    java.lang.NoClassDefFoundError: org/mortbay/jetty/servlet/Context

It looks like this is the correct answer. Everything was fine until Friday afternoon and then bam… 100% CPU.

Any idea what triggers it, do you use the webchat?

My setup is very simple. Just a base Openfire install with the search plugin. Most users connect with Psi clients and I use KDE Telepathy. No webchat.

I have 7 Openfire 3.10 Server installed and 6 of them are on Qnap NAS using tar.gz version and the 7th one is on Ubuntu 14.04. They all crashed eventually but none of them using the CPU to 100% but they are just stop accepting client new sessions and in the openfire web admin, session-summary.jsp would display error or never loads and we have to restart it in order to solve the problem. Right now I almost have to write a cron job to restart openfire server every two hours.

I have remove most plugins and also most of servers only have 20 clients connected, however, all 7 servers have s2s connected. Please fix this issue and thanks in advance for your all your development effort, openfire xmpp is so important to our organization inner communication