Projects tigase _server server-core Issues #343
Problem in connecting to Tigase XMPP on EC2 HOST from an external Spark Client (#343)
Won't Fix
Girish Prabhu opened 1 decade ago

I just setup Tigase 5.2.0 on an Amazon EC2 instance. The setup by itself went well, but when I try and connect from an external SPARK Client, I am not able to login to the server (even-though I see the request coming...it's just not able to response to it).

Now, I noticed one thing, when I setup the server (and i used the text base console method), the HOST NAME in the configuration files is the INTERNAL HOST NAME on EC2. (eg: ip-172-xx-x-xxx.us-west-1.compute.internal). I see that all the entries in the conf files are using this as the hostname! Note that the host also has an external IP address (which is obviously what am using on my SPARK client). I am wondering if Tigase is seeing a mismatch because of the external IP and the INTERNAL VHOST name for that domain.

Here's the log entry when the SPARK CLIENT makes the request to EXTERNAL EC2 IP:5222

2014-06-30 01:31:34.119 [ConnectionOpenThread] ConnectionManager$ConnectionListenerImpl.accept() FINEST:* Accept called for service: null@null*

2014-06-30 01:31:34.120 [ConnectionOpenThread] ConnectionManager.serviceStarted() FINER: [[c2s]] Connection started: null, type: accept, Socket: nullSocket[addr=/24.5.80.184,port=61563,localport=5222], jid: null

2014-06-30 01:31:34.167 [pool-15-thread-1] ClientConnectionManager.xmppStreamOpened() FINER: Stream opened: {xmlns:stream=http://etherx.jabber.org/streams, to=54.193.XXX.XX, xmlns=jabber:client, version=1.0}

2014-06-30 01:31:34.168 [pool-15-thread-1] ConnectionManager.serviceStopped() FINER: [[c2s]] Connection stopped: c2s@ip-172-xx-x-xxx.us-west-1.compute.internal/172-xx-x-xxx_5222_24.5.80.184_61563, type: accept, Socket: c2s@ip-172-xx-x-xxx.us-west-1.compute.internal/172-xx-x-xxx_24.5.80.184_61563 Socket[unconnected], jid: null

How do I get this running?. Is this because of the mis-match between the EC2 INTERNAL domain name and the external Elastic IP (which is what the SPARK client is using?).

Thanks in advance for the help.

Artur Hefczyc commented 1 decade ago

Please use our forums to ask questions about your Tigase installation: https://projects.tigase.org/projects/tigase-server/boards

Girish Prabhu wrote:

I just setup Tigase 5.2.0 on an Amazon EC2 instance. The setup by itself went well, but when I try and connect from an external SPARK Client, I am not able to login to the server (even-though I see the request coming...it's just not able to response to it).

Now, I noticed one thing, when I setup the server (and i used the text base console method), the HOST NAME in the configuration files is the INTERNAL HOST NAME on EC2. (eg: ip-172-xx-x-xxx.us-west-1.compute.internal). I see that all the entries in the conf files are using this as the hostname! Note that the host also has an external IP address (which is obviously what am using on my SPARK client). I am wondering if Tigase is seeing a mismatch because of the external IP and the INTERNAL VHOST name for that domain.

There is no mismatch. The internal hostname of the VM is used by Tigase for it's internal communication between components only. And also for communication between cluster nodes if the Tigase runs in a cluster mode. So there should be no problem with this.

Here's the log entry when the SPARK CLIENT makes the request to EXTERNAL EC2 IP:5222

2014-06-30 01:31:34.119 [ConnectionOpenThread] ConnectionManager$ConnectionListenerImpl.accept() FINEST:* Accept called for service: null@null*

2014-06-30 01:31:34.120 [ConnectionOpenThread] ConnectionManager.serviceStarted() FINER: [[c2s]] Connection started: null, type: accept, Socket: nullSocket[addr=/24.5.80.184,port=61563,localport=5222], jid: null

2014-06-30 01:31:34.167 [pool-15-thread-1] ClientConnectionManager.xmppStreamOpened() FINER: Stream opened: {xmlns:stream=http://etherx.jabber.org/streams, to=54.193.XXX.XX, xmlns=jabber:client, version=1.0}

2014-06-30 01:31:34.168 [pool-15-thread-1] ConnectionManager.serviceStopped() FINER: [[c2s]] Connection stopped: c2s@ip-172-xx-x-xxx.us-west-1.compute.internal/172-xx-x-xxx_5222_24.5.80.184_61563, type: accept, Socket: c2s@ip-172-xx-x-xxx.us-west-1.compute.internal/172-xx-x-xxx_24.5.80.184_61563 Socket[unconnected], jid: null

How do I get this running?. Is this because of the mis-match between the EC2 INTERNAL domain name and the external Elastic IP (which is what the SPARK client is using?).

What VHost have you set in the Tigase configuration? Looks like the problem comes form the fact that you used the server IP address instead of the VHost in Spark. Most likely the IP address is not set as the server VHost, therefore Tigase rejects connections for IP address as domain.

Referenced from commit 11 months ago
#343 - Ad-hoc command to load/show the server configuration file.
git-svn-id: file:///home/svn/repos/tigase-server/trunk@2818 7d282ba1-3ae6-0310-8f9b-c9008a0864d2
andrzej.wojcik committed 1 decade ago
issue 1 of 1
Type
Bug
Priority
Major
Assignee
RedmineID
2043
Version
tigase-server-5.2.1
Issue Votes (0)
Watchers (0)
Reference
tigase/_server/server-core#343
Please wait...
Page is in error, reload to recover