relay server crashing

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Complile the relay server from libjingle 0.6.1
2. make the following script to re-spawning the server if it's crashing

#!/bin/bash 

echo "Running relay"

until libjingle-0.6.1/talk/build/opt/staging/relayserver ip_here:5000 
ip_here:5001; do
    echo "relay server crashed with exit code $?.  Respawning.." >&2
    sleep 1
done

3. running the script: nohup ./run_relay.sh >>relay_out.txt &

What is the expected output? What do you see instead?

The server is crashing very frequently. Into the output file I receive:

Listening internally at 10.11.1.225:5000
Listening externally at 10.11.1.225:5001
Segmentation fault
relay server crashed with exit code 139.  Respawning..

What version of the product are you using? On what operating system?

Ubuntu 11.04
libjingle 0.6.1 from SVN

If I can provide any other informations to help with this issue let me know.

Original issue reported on code.google.com by silviu.cpp@gmail.com on 14 Oct 2011 at 1:52

[GoogleCodeExporter]

Still no solution for this?

Original comment by silviu.cpp@gmail.com on 9 Nov 2011 at 9:10

[GoogleCodeExporter]

I can't seem to reproduce what you're seeing.  Have you retested with 0.6.3 or 
the latest out of SVN?  Is using the same address for internal and external the 
right usage?  Have you tried with different addresses?  The usage help for 
relayserver is sorely lacking.  Does it crash immediately?  Is anyone 
connecting to that relay server?  I can run relayserver on Fedora 14 and Ubuntu 
11.04 with similar args (same address, ports 5000 and 5001), and it isn't 
crashing, but I also don't have anyone connecting to it.

Original comment by bprodo...@connectify.me on 10 Nov 2011 at 12:30

[GoogleCodeExporter]

Hi,

I tested with libjingle 0.6.1 because I have not found any changes into to 
newer versions regarding the relay server. I will update this days to the last 
version from the repository and I will test again. 

NOTE: The server doesn't crash if no connections are made through it. It crash 
when there are active clients (based also on libjingle) which use that relay 
server. 

And when it's crashing( during a call ) the quality of call become very very 
poor since it switch to another server.

Original comment by silviu.cpp@gmail.com on 10 Nov 2011 at 7:39

[GoogleCodeExporter]

Can you possibly get a stacktrace?  These instructions apply to any distro, 
really:  http://fedoraproject.org/wiki/StackTraces#gdb

Original comment by bprodo...@connectify.me on 11 Nov 2011 at 3:23

[GoogleCodeExporter]

Hi should I run the debug version ? because on the release on I get the 
attached output which is not so clear .

The server was crashed over the night without any client being connected. 

Do you need any other information?

Original comment by silviu.cpp@gmail.com on 18 Nov 2011 at 6:48

Attachments:

• crash.png

[GoogleCodeExporter]

Hello guys,

I have reproduce the bug and you can find the entire stack attached.

Can I provide you any other useful information?

Original comment by silviu.cpp@gmail.com on 18 Nov 2011 at 11:18

Attachments:

• bug_relay.png

[GoogleCodeExporter]

Looks like it's trying to delete non-existent connection record from map.

Original comment by Vadim.Ye...@gmail.com on 20 Nov 2011 at 8:05

[GoogleCodeExporter]

Original comment by juberti@google.com on 21 Dec 2011 at 11:15

[GoogleCodeExporter]

I have reproduced the bug too with the relay server compiled from libjingle 
0.6.3. I have attached the dmp and pdb files.
Steps to reproduce:
1. use call.exe to connect 2 clients to the presence server
2. use call command from client 1 to call client 2
3. the transfer begins using relay server (p2p connection fails) and after 
about 21MB of data transferred the relay server crashes.

If you need any other information, please let me know.

Original comment by sori...@gmail.com on 12 Jan 2012 at 1:08

Attachments:

• dump_relay.zip

[GoogleCodeExporter]

Original comment by jun...@google.com on 26 Mar 2012 at 6:39

[GoogleCodeExporter]

Was this issue ever fixed?

Original comment by mepting0...@gmail.com on 23 Mar 2013 at 6:13

[GoogleCodeExporter]

Relayserver is deprecated. Use turnserver instead.

Original comment by juberti@google.com on 1 Jun 2013 at 4:28

• Changed state: WontFix

[GoogleCodeExporter]

Relayserver is deprecated. Use turnserver instead.

Original comment by juberti@google.com on 1 Jun 2013 at 4:29