presence cannot be send to other online users
Submitted by Anonymous on Fri, 2009-03-06 10:44
Hi!
While we were testing Tigase clusters on Imacs, all went fine.
Now we installed a cluster of 2 nodes on linux servers (not available on the internet) and sometimes we get some errors about the presence which cannot be delivered.
this is an example:
2009-03-06 09:54:29 SessionManager.walk() WARNING: Can not add packet: to=sess-man@jabber02.netnoc.eu, from=null, data=
0
, XMLNS=jabber:client to processor: msgoffline internal queue
we also sometimes get these errors from the cluster-contr:
Jabber IM with cluster-contr@netlog.com06/03/09 10:05
Disconnected cluster nodes
Cluster nodes have been disconnected (Fri Mar 06 10:05:52 CET 2009)
1. jabber01.netnoc.eu
New cluster nodes connected
Cluster nodes have been connected (Fri Mar 06 10:05:57 CET 2009)
1. jabber01.netnoc.eu
has this something to do with Tigase or with a network configurtion?
Thanks!
Cliff
- 1140 reads
Printer-friendly version- Send to friend
Sorry, we found them in the
Sorry, we found them in the tigase utils download.
I will update alle the class files again, to be shure
this one:
this one: https://svn.tigase.org/reps/tigase-server/tags/tigase-server-4.1.4/
but there are class files missing, such as tigase.util.Algorithms, tigase.util.JIDUtils, ...
we found the missing classes in other releases but they are probably outdated.
Which latest? Latest public
Which latest? Latest public release, the latest binary update or latest SVN?
Hi! we use the latest
Hi!
we use the latest version of Tigase server and use decent servers to test the clusters. We already solved one problem by increasing the java memory.
But the second problem, when the clusters suddenly disconnect and connect, not. We don't know why this occurs...
We use a load balancer on which the clients connect and every client conects on one of the servers in the custer.
What version of the Tigase
What version of the Tigase server do you use? What system it is running on? RAM, CPU, .....
It is really impossible what is going on the server from a single log line. You mention 2 separate events, one a problem with session manager processing and another with disconnectio/connection between cluster nodes. Those 2 events might be unrelated or can be related but without proper investigation I really can not tell you too much.
The log entry is printed when the packet processor internal queue is overfilled and the packet can not be processed properly. In this case this is offline messages processor. I guess it communicates with the database. If you have database connectivity problem then this might affect the whole system performance. If this is the Netlog REST API with JSON calls then it most likely doesn't offer performance good enough to cope with the traffic on the Tigase server. You may improve it with some smart caching mechanism but still, I am afraid this might be a serious bottleneck.
The second event - disconnection between cluster nodes may suggest some networking problems. Again without further investigation I can not say too much.
2009-03-06 09:54:29
2009-03-06 09:54:29 SessionManager.walk() WARNING: Can not add packet: to=sess-man@jabber02.netnoc.eu, from=null, data=<presence xmlns="jabber:client" from="joostroelandts@netlog.com/Holly" to="evalouwielou@netlog.com">
<priority>0</priority>
<c node="http://www.apple.com/ichat/caps" xmlns="http://jabber.org/protocol/caps" ext="audio video recauth maudio avcap avavail rdclient rdserver ice auxvideo mvideo" ver="616"/><x xmlns="vcard-temp:x:update"/></presence>, XMLNS=jabber:client to processor: msgoffline internal queue
Post new comment