[asterisk-bugs] [JIRA] (ASTERISK-20453) res_xmpp.c: JABBER: socket read error, afterwards outgoing connections never get answered

Thu Mar 7 10:22:01 CST 2013

    [ https://issues.asterisk.org/jira/browse/ASTERISK-20453?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=203971#comment-203971 ] 

Aaron Chan commented on ASTERISK-20453:
---------------------------------------

i'm on 11.2.1 and i've been seeing this issue pretty frequently. It happens when:

- I answer a call on the Motif channel too fast. I mitigated this by issuing a Wait(2) before Answer() whenever i receive a google voice call

- Just happens randomly. Last night, after Asterisk was running for about 7 hours, I got "JABBER: socket read error" on the log even when i wasn't placing or receiving any calls. afterwards i was unable to place/receive calls through google voice.

In attempting to figure out why this happens on the new Motif channel but not on the old GTalk channel in 1.8, i read the code on res_xmpp.c and tried to compare it with res_jabber.c.

On res_xmpp.c's xmpp_client_thread() function, the following code was placed before the xmpp_client_reconnect() function:

               if (client->state == XMPP_STATE_DISCONNECTING) {
                        break;
                }

this, i assume, would break out of the infinite loop and terminate the thread.

however, on res_jabber.c's aji_recv_loop() function, the same code was placed AFTER the aji_reconnect() function.

i suspect this was the cause... does anyone know why such a change was implemented in res_xmpp.c? if we move that code snippet so it's run after xmpp_client_receive(), will that solve the problem?

> res_xmpp.c: JABBER: socket read error, afterwards outgoing connections never get answered
> -----------------------------------------------------------------------------------------
>
>                 Key: ASTERISK-20453
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-20453
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Channels/chan_motif, Resources/res_xmpp
>    Affects Versions: 11.0.0-beta1
>         Environment: Linux lucid 32bit running under openvz
>            Reporter: Andrey Petrov
>            Assignee: Andrey Petrov
>            Severity: Critical
>
> Hello,
> I have been using asterisk with google voice already for a 1+ years.
> Recently I decided to switch to new asterisk11beta1, because of all new goodies coming with new Motif.
> I switched about 2 weeks ago and everything worked pretty well.
> However just a few days ago, XMPP connections seem to start ending abruptly:
> {noformat}
> [Sep 18 04:47:56] WARNING[8510] res_xmpp.c: JABBER: socket read error
> [Sep 18 07:17:00] WARNING[8510] res_xmpp.c: JABBER: socket read error
> [Sep 18 07:20:22] WARNING[8510] res_xmpp.c: JABBER: socket read error
> [Sep 18 07:21:12] WARNING[8510] res_xmpp.c: JABBER: socket read error
> {noformat}
> Interestingly enough, this error first started occuring pretty seldom, perhaps 1 a day. So in the begginning I discarded it as mundane tcp error. However, the frequency somehow increased and now it happens every other minute. It also happened to me during actual call several times. I tried setting keepalive=yes in xmpp.conf but to no avail. I tried different transports for Motif and it didnt help either.
> If I connected with asterisk -r,
> "xmpp show connections" seems to show connection is alive. And it is alive, as I am receiving status notifications.
> However, outgoing calls (and perhaps incoming, but I am not sure) placed through Motif never get answered.
> I am getting something like this in the log when placing a call:
> {noformat}
> [Sep 18 08:05:54] VERBOSE[8542][C-00000000] netsock2.c:   == Using SIP RTP CoS mark 5
> [Sep 18 08:05:54] VERBOSE[9012][C-00000000] pbx.c:     -- Executing [9093900003 at outbound:1] MixMonitor("SIP/android-00000000", "to-9093900003-Tue Sep 18 08:05:54 2012.wav,b") in new stack
> [Sep 18 08:05:54] VERBOSE[9013][C-00000000] app_mixmonitor.c:   == Begin MixMonitor Recording SIP/android-00000000
> [Sep 18 08:05:54] VERBOSE[9012][C-00000000] pbx.c:     -- Executing [9093900003 at outbound:2] Dial("SIP/android-00000000", "Motif/google/9093900003 at voice.google.com,,Tr") in new stack
> [Sep 18 08:05:54] VERBOSE[9012][C-00000000] app_dial.c:     -- Called Motif/google/9093900003 at voice.google.com
> [Sep 18 08:05:55] VERBOSE[9012][C-00000000] app_dial.c:     -- Motif/9093900003 at voice.google.com-a3a3 is proceeding passing it to SIP/android-00000000
> [Sep 18 08:06:10] VERBOSE[9012][C-00000000] pbx.c:   == Spawn extension (outbound, 9093900003, 2) exited non-zero on 'SIP/android-00000000'
> [Sep 18 08:06:11] VERBOSE[9013][C-00000000] app_mixmonitor.c:   == MixMonitor close filestream (mixed)
> [Sep 18 08:06:11] VERBOSE[9013][C-00000000] app_mixmonitor.c:   == End MixMonitor Recording SIP/android-00000000
> {noformat}
> This is a critical issue to me, because it forced me to downgrade back to 1.8.
> Additional information:
> I tried trunk, branches/11, but I can not confirm or deny if it is reproducible or not, because with these versions once call is established the voice is never heard both ways for unknown reason.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.asterisk.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira