chan_iax2.c: Max retries exceeded

There’s no recent entries about this error:

WARNING[25290] chan_iax2.c: Max retries exceeded to host 172.16.x.x on IAX2/trunk_elastix05 (type = 6, subclass = 11, ts=40013, seqno=15)

And the few posts about this didn’t find a solution other than keep reloading asterisk or rebooting the server (a case even changed the router)

We have two Asteriks servers comunicating via IAX2 trunk, the work fine for a few days (usually 4 or 5) and then the trunk collapses and starts dropping the calls (current and new), and the asterisk process goes to 100% CPU consumption. I read somewhere at that SIP trunks won’t scale very good for more than 100 calls and IAX trunks at 300 to 400 calls; since we are having 320 to 360 simultaneous calls I thought we could try to balance the calls among 3 trunks (implemented via dialplan).

It worked for about 4 or 5 days and again the warning and the 100% CPU load; neither asterisk or iax module reload fix the problem, only rebooting the server. I’m still trying to troubleshoot this in order to get to the root of the problem and fix this as soon as possible.

We are using Centos 5.6 and Asterisk 1.8 on Server_A and Centos 6.5 and Asterisk 11 on Server_B (the one bugging out), in the meantime we are rolling back to Asterisk 1.8 where it seemed to work just fine, so I think that maybe there is something in the Asterisk manager that loops and boost the CPU consumption internally denying the service, unless there’s some weird bug when you connect 2 different versions of Asterisk.

Any tips or guides will be really appreaciated, also I will post my findings I get to the root of this issue.