[asterisk-dev] delay in *

[Clod Patry]

hi guys,
Since 2 days, i noticed something really weird happening in my whole *.
When im calling someone, when im talking, there's _no_ delay, all
communications seems really fluide.
Im using MeetMe or MOH too, there's no problem.

But when im leaving that system idling with active meetme and/or MOH, when
im comin' back, all chans are still ups, but with delay.
I'll explain a bit more here:
All RTPs packets are comin' but it's like they're so slow, so if im saying
HEY on phone A, it takes upto easily 1 sec to hear the HEY on phone B.
I checked the CPU, everything is okay too.
But if i hangup, then call back, everything is like no problem never
occured, even without restart/reload.

When I hanguped all phones, this is what happen in CLI:

[Feb 23 09:04:32] DEBUG[402]: channel.c:3233 ast_channel_masquerade:
Planning to masquerade channel SIP/86-1017f778 into the structure of
AsyncGoto/SIP/86-1017f778
[Feb 23 09:04:32] DEBUG[402]: channel.c:3246 ast_channel_masquerade: Done
planning to masquerade channel SIP/86-1017f778 into the structure of
AsyncGoto/SIP/86-1017f778
[Feb 23 09:04:32] DEBUG[402]: channel.c:3611 ast_do_masquerade: Released
clone lock on 'AsyncGoto/SIP/86-1017f778<ZOMBIE>'
    -- Executing [92 at gondor:1] MeetMe("SIP/86-1017f778", "1234") in new
stack
[Feb 23 09:04:32] DEBUG[408]: rtp.c:2769 bridge_native_loop: Oooh, something
is weird, backing out
[Feb 23 09:04:32] DEBUG[408]: channel.c:4027 ast_channel_bridge: Bridge
stops because we're zombie or need a soft hangup: c0=SIP/203-10179728,
c1=AsyncGoto/SIP/86-1017f778<ZOMBIE>, flags: No,No,Yes,Yes
[Feb 23 09:04:32] DEBUG[408]: channel.c:4128 ast_channel_bridge: Bridge
stops bridging channels SIP/203-10179728 and
AsyncGoto/SIP/86-1017f778<ZOMBIE>
  == Spawn extension (gondor, 86, 1) exited non-zero on 'SIP/203-10179728'
[Feb 23 09:04:32] DEBUG[442]: app_meetme.c:1294 conf_run: Placed channel
SIP/86-1017f778 in ZAP conf 1023
  == Spawn extension (gondor, 92, 1) exited non-zero on 'SIP/203-10185990'
  == Spawn extension (gondor, 92, 1) exited non-zero on 'SIP/202-10161ef8'


I dont like to see the ZOMBIE stuff, nor the "Oooh, something is weird,
backing out".

Based on rtp.c code:
                /* Check if anything changed */
                if ((c0->tech_pvt != pvt0) ||
                    (c1->tech_pvt != pvt1) ||
                    (c0->masq || c0->masqr || c1->masq || c1->masqr)) {
has to happen to see my last debug msg, but any reason why it happen
exactly? How come the flow pass in that if? One channel (c0 or c1) has to
have a masq, no? Since the whole pbx was idling my my little lab setup, I
dont understand.


Any suggestions on that?
how to debug exactly?
Can a zombie chan complety _lag_ the pbx?

-- 
Clod Patry
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.digium.com/pipermail/asterisk-dev/attachments/20070223/4a9de046/attachment.htm