Summary:ASTERISK-13156: Asterisk logs: Max retries exceeded to host on IAX/peer (type = 6, subclass = 11) but these packages can't be found in a tcpdump
Reporter:Daniel Tryba (kwark)Labels:
Date Opened:2008-12-02 10:56:08.000-0600Date Closed:2011-06-07 14:03:16
Versions:Frequency of
Environment:Attachments:( 0) 20081202__bug14008.diff.txt
Description:During the last 2 months I had a couple of asterisk "crashes", each time the logs get spammed with lines like:

[Dec  1 20:18:55] WARNING[1812] chan_iax2.c: Max retries exceeded to host on IAX2/locografix-5656 (type = 6, subclass = 2, ts=4978257, seqno=230)

During the last I had the opportunity  to make a tcpdump (tshark -w /tmp/foo port 4569), but this tcpdump only contains normal iax2 traffic for this host:
pokes/pongs/acks and regreqs/regacks/acks
So the only subclass I have in the cap are 3,4,13,15 and 30.

During the previous "crash" the problem started with 1 iax peer, but over the course of 30m the max retries log entries started to appear for almost all iax peers (sadly no packetdump from this crash).

Next time I'll try to remember to dump all interfaces (2 ethernet (1 external, 1 lan) and loopback), maybe somehow asterisk is routing the traffic to the wrong interface.

Restarting asterisk fixes all problems.

This machine is the gateway to the ISDN network, the peers connect thru iax trunks.


Asterisk version is from Debian/unstable backported to Debian/stable amd64, so there are a couple of nonstandard patches applied, but the only patch that might interfere with iax is the chan-iax2-hangup-cause patch (http://patch-tracking.debian.net/patch/series/view/asterisk/1:

The cause of the problem might be in the zaptel driver used on this system. I found this iax problem when I couldn't dial a external number through the ISDN-pri connected to a TE121 with VPMADT032. Due to echo cancelling problems with the hw canceller we are running http://svn.digium.com/svn/zaptel/team/mspiceland/zaptel-1.4-adtapiupdatefor117 (rev 4420). Trying to dial outwards from iax hangs (Dial(Zap/g1/xxx) never returns), incoming calls get routed to the fallback asterisk server by the isdn provider.

So my dilemma is: should the bug category be iax of zaptel? Since I have no usefull zaptel debug info I decided for iax since the max retries log entries appear not to be correct.

BTW up until the calls started to get routed to the fallback machine, "pri debug span 1" was producing the normal debug output, suddenly it went completely silent.
Comments:By: Tilghman Lesher (tilghman) 2008-12-02 18:48:20.000-0600

This patch should, at the very least tell you why the packets aren't getting delivered.  It's possible that the frames are hanging out past the point where they should be destroyed, or it's also possible that something is wrong with the network, and a permanent failure has resulted.  In either case, we should output that to the logs and we are not currently.

Patch is against the current 1.4 SVN, which may be slightly different than

By: Joshua C. Colp (jcolp) 2009-02-11 11:17:38.000-0600

kwark: Any information from Corydon76's patch?

By: Daniel Tryba (kwark) 2009-02-12 07:06:46.000-0600

Sorry for the late reply.

I'll try to get this on the machine with the connection problems.

I'll keep you posted.

By: Daniel Tryba (kwark) 2009-02-12 07:10:45.000-0600

BTW the crashes aren't related to any specific zaptel version. Both the custom zaptel I got from Digium and the zaptel driver from Debian/unstable (with OSLEC support) will crash eventually.

By: Leif Madsen (lmadsen) 2009-05-12 20:03:19

Any update from the reporter with results after applying the patch from Tilghman?

By: Leif Madsen (lmadsen) 2009-06-10 13:05:40

Closing this issue as there has been no response from the reporter. Should you have the information requested, please reopen the issue and attach it. Thanks!

By: Bobby Hakimi (bobbymc) 2017-11-05 23:32:09.313-0600

i have the same issue with asterisk 11.25. anyone got any updates with this ?