Spark Client slow, and no updates

We just installed OpenFire 3.8.1 using active directory LDAP. We are also using Contact List (Roster) Sharing.

I have a few issues I am hoping someone can help me with…

Using Spark 2.6.3 we are seeing the client with a large contact list responding very slow. We are still in the testing phase so this particluar clients has mostly offline users in his roster. If I try to scroll through the list or even try to move the Spark window, it will not respond for about 30 seconds.

**- Is this normal for clients with large roster? **If I turn off showing offline users, the client works fine.

Secondly, I have created groups for contact list sharing, and have them shared for each of my offices and departments

  • I was testing a particular user who was in one group called Charlotte. This person was able to see all others in the Charlotte group. Works good!

  • I then moved this particular users to a different group called New York, however his roseter did not update. He does not see the New York group or the Charlotte group, but his offline users are still showing the Charlotte people.

Should these rosters be updating automatically? If so how often should it update? Can the setting be changed?

Thanks for the help!

-Evan

You can try the latest Spark nightly build (#590) http://bamboo.igniterealtime.org/browse/SPARK-INSTALL4J-590/artifact/JOB1/Instal l4j/

I remember some fixes related to big rosters.

Rosters should update automatically, but not immediately. Openfire uses caches, so changes doesn’t propagate very fast. You can try clearing the Roster cache in Server Settings > Cache Summary (Admin Console), though this may not be enough and you probably would need to restart a client too (actually restarting the client is also enough to propagate such changes as group change or name change). You can also try setting smaller caches, though this can impact server performance http://community.igniterealtime.org/docs/DOC-1673

Thanks for the reply…I will look into these links and see if this helps.

You seem fairly knowledgeable regarding OpenFire/Spark. Any idea how many clients can connect to one server without seeing degradation in the service.

My server is a Xeon Processor 5130 @ 2.ghz (dual core), with 2gb of memory. Not the most powerful server.

Thanks again.

wroot wrote:

You can try the latest Spark nightly build (#590) http://bamboo.igniterealtime.org/browse/SPARK-INSTALL4J-590/artifact/JOB1/Instal l4j/

I remember some fixes related to big rosters.

Rosters should update automatically, but not immediately. Openfire uses caches, so changes doesn’t propagate very fast. You can try clearing the Roster cache in Server Settings > Cache Summary (Admin Console), though this may not be enough and you probably would need to restart a client too (actually restarting the client is also enough to propagate such changes as group change or name change). You can also try setting smaller caches, though this can impact server performance http://community.igniterealtime.org/docs/DOC-1673

Never mind…I answered it my self.

Hardware Recommendations

Concurrent Users

Recommendation

1-500

At least 384 MB of available RAM and one 1.5 GHz processor.

501-10,000

At least 768 MB of available RAM and one 3.0 GHz processor.

10,001-25,000

At least 1.5 GB of available RAM, two 3 GHz processors, and one

[connection manager](http://web.petabox.bibalex.org/web/20070808144948/http://www.igniterealtime.org/projects/openfire/connection_manager.jsp)

on the same machine.

25,001-100,000

At least 2.0 GB of available RAM, two 3 GHz processors, and between 1 and 4

[connection managers](http://web.petabox.bibalex.org/web/20070808144948/http://www.igniterealtime.org/projects/openfire/connection_manager.jsp)

based on exact size, each on their own equivalent machine.

It all depends on clients behavior (how many chats, file transfers, etc.) and on users number if it goes beyound a thousand i think. I have only 200 users, so i’m fine with a virtual server (1 virtual core of Xeon and 512 MB RAM, Arch Linux without GUI). I was running similar setup for 150 users on Pentium 3 500 MHz and 64 MB of RAM You should watch your memory and JVM memory status (first page in Admin Console - you can refresh it constantly and see at what point it releases the memory, if it doesn’t release the memory, then you may need to increase memory delegated to JVM; by default it is just 64 or 128 MB i think.). I have increased my JVM memory for it to run for a month without outofmemory errors between my scheduled server maintenance reboots.

Custom Parameters part tells about increasing the JVM memory http://www.igniterealtime.org/builds/openfire/docs/latest/documentation/install- guide.html