Random disconnects /clients dropping (openfire 2.5.2)

Hi all,

I have an openfire server running, 83 total users, normally around 60 online during the main portion of the day. It could be 12 hours uptime or 10 min uptime. The server will just start randomly booting people off the server. Sometimes 5+ at a time. Their clients will auto-reconnect… so you will see the roster going up and down until the service is rebooted. The clients are all using the latest Pidgin 2.4.3.

I tried a couple flavors of linux starting with ubuntu, then to SUSE 10.2 then currently its running on SUSE 11. (All 32-bit) They were running on vmware setups (they were the guest OS) execpt this current one on SUSE 11. This is a physical machine running no vmware, just straight connection. I have been trying to troubleshoot this problem. I wanted to rule out vmware as being the issue (I found a couple of posts questioning if it could cause such problems) so native SUSE 11 32-bit. Hardware is a 3.4ghz Pentium 4. 1GB ram.

I have also tried and bumped up the java memory from default 64 to 128megs. I have had previous encounters with programs which use java as the back end loving a little more memory. Unfortunately this didn’t help the problem either.

Here is what I find in the ERROR logfile:

The only other thing worth mentioning is currently the server sits out on the internet, and the company accesses it through a squid proxy. (http) AIM, ICQ, MSN, Googletalk, Yahoo all that stuff. (all these others don’t randomly disconnect…, no network loss of any kind)

I am going to try and catch debug text on what happens on the client when it deicdes to boot people… See if there is any info there as well. I am thinking I might try and get the server internal and get the proxy out of the mix and see if it continues to boot people.

If anyone has any thoughts, ideas or have seen this before with some info that would be great!

Edit: Just had the problem happen again. Client said “Read Error” and disconnected… im leaving a debug window open hoping to catch more.

–Had it happen again I got the debug output from pidgin when it disconnected me…

Thank you for your time,

-Brian

Just some more info.

I took the proxy out of the loop, I moved the server onto our internal network and all clients are connected directly. – Problem still exists.

Next I have no installed the Spark client in place of Pidgin on a few clients – These clients no longer have the problem.

I was hoping we could use Pidgin since it works so well with all of the other protocols (aim,icq,msn etc) but I guess we might have to wait until a proper fix has been made in the pidgin.

Unless anyone has any ideas…

We had this same issue as well. It turned out a user was using an unapproved client to connect which hosed the server, Openfire did not handle the connection properly and caused a constant connect and disconnect of all users. After many hours of troubleshooting we then added the client restriction addon to only allow Pidgin and Spark and that fixed the issue. Needless to say that user then contacted us asking why he could not connect and sure enough he was using a bad client (Octrotalk). I would try that and see if it fixes it.