Re: Help! Random calls disconnects
Yu Boot <yu@...
2014-10-10 06:46:58 GMT
Now we're forced to move most annoyed clients from this Yate to Asterisk
at least for testing. All the same network, hardware, clients, uplinks
etc - and no call disconnects at all. Magic??? Something is definitely
wrong with Yate and its' configuration. What can it be?
09.10.2014 20:24, Paul Chitescu пишет:
> Can you take a look at the Yate generated BYE message - in particular the
> Reason and P-RTP-Stat headers? Also check the disconnect reason in CDR if
> What is the peer of the SIP channel? What is the inbound call coming from?
> Do you do RTP forwarding or all calls' voice is proxied through Yate?
> The call termination may be caused by:
> - inbound channel hanging up (BYE received by Yate)
> - possibly some periodic reINVITE issues
> - reINVITE not received or not answered by Yate
> - remote side expecting periodic reINVITE without RFC 4028 negotiation
> - fax tone misdetection causing a rejected T.38 reINVITE
> - a hangup on RTP receive timeout configured in yrtpchan.conf
> - a timeout (around 65000 in this case) parameter set during routing
> In the last 2 cases Yate would hang up both inbound and outbound call legs
> from the middle.
> On Thu Oct 9, 2014 18:40:32 Yu Boot wrote:
>> Help me please! Just got crazy about it.
>> Yate randomly sends "BYE" to clients in the middle of conversation
>> without any visible reason. It happens few dozens times a day on random
>> clients. Network is OK, no ping loss, no delays, no overload etc.
>> Unfortunately I can't to turn on message sniffer or even tshark all (in
>> and out) SIP traffic due to huge amount of calls. All I can is tcpdump
>> client IP only for a day or two and then open dump file in Wireshark
>> http://tinypic.com/r/295uyyb/8 - it's how it looks in Wireshark on
>> tcpdump'ed file from Yate.
>> 78 IP is Yate, 38 is a client.
>> What parameters, timers, conditions, etc AT ALL can cause that Yate
>> behaviour? Yate was specially moved to quad core Xeon with 8GB RAM for
>> only 150-160 concurrent SIP sessions, machine is faaaaaar underloaded,
>> and this didn't help. Fresh installed CentOS 6.5 with all updates,
>> latest Yate 5.4.1 devel1 r5915, G729 codec from yate.hosting.lv
>> I can provide any Yate logs and configs to solve this problem, without
>> message sniffer and total "tcpdump port sip", of course.