Spark Disconnects Randomly

I am hoping re-open the can of worms that is this issue since a solution has never been clearly documented here.

Spark is randomly disconnecting for my staff after 6 months of it working perfectly.

Here’s a little background:

  • Latest version of Openfire installed on Windows Server 2003 virtual server.
  • 200 clients using latest version of Spark on 200 XP/Win7 machines.
  • Users authenticate via Active Directory using LDAP setup.
  • Everything worked perfectly up until a few weeks ago when all users experience connectivity issues. Spark randomly disconnects “due to an error.”
  • Openfire lists errors in the log but I have no idea how to read them.
  • I have reinstalled Openfire with no resolution.
  • I have used other clients such as Spark beta and Pidgin with no resolution.
  • I have noticed that when spark disconnects the Openfire admin panel also does not work.
  • I have nothing else running on the windows server.
  • Continual server monitoring has shown me nothing
  • been happening for a couple weeks now

Attached are the current openfire log files. If anyone is willing to look through them and tell me what they think. I would greatly appreciate it.
error.log.zip (10867 Bytes)
warn.log.zip (3570 Bytes)
info.log.zip (2196 Bytes)

Hi,

you may need to increaase the Xmx (memory) value for Openfire. The install documentation describes how to do this and JVM Settings and Debugging may also be interesting.

LG

Thanks for the reply. I increased the memory greatly on the latest reinstall but it did not fix the issue.

Hi,

I don’t see database errors in the log and only a few LDAP errors (group XYZ not found). Anyhow I wonder whether they are related to your problem. Could you enable the GC log as described in the document so one can make sure that you don’t have a memory problem.

It may be possible that your AD does not respond in time, anyhow this should not drop all users.

LG

I enabled the GC log, i think, but I dont see anything different- it appears it wasnt added. i guess I’m not sure WHERE to add the line “-XX:+PrintGCDetails -Xloggc:/tmp/gc.log” and where to view the log afte adding it.

I have reinstalled openfire a couple times and even moved it to a different server, but every 30 minutes the entire server locks up and spark and the openfire admin console disconnect and then reconnect.

We noticed this issue started when we upgraded our 2003 domain to 2008. Anyone have issues like this?

Hi, do you have a folder /tmp/ and there a gc.log file? Otherwise you may want to change “-Xloggc:/tmp/gc.log” to “-Xloggc:c:/temp/gc.log” or simply to “-Xloggc:gc.log” and then search for the “gc.log” file.

LG

After months of failed attempts to fix this disconnect issue, I decided to move openfire directly onto the domain server, cutting out any interference or dns resolution issues between the server Openfire was on originally and the domain server. I did this because it seemed like the disconnect was caused by an authentication timeout that was happening every 30 minutes. So for we’re into a week of Spark/openfire working perfectly. During the brand new setup of Openfire on the domain server, I did NOT chagce any settings outside of the LDAP integration and group sharing. So I don’t know, maybe it wasnt the server afterall but some setting I was changing in the previous installations.

Every environmen/setupt is different so our issues are our issues it seems…since no one had a solution for me.