asterisk users - Nov 2010 - UNREACHABLE/Lagged happening on "bulk" register/subscribe

Dears Friends,

I currently have 16 Cisco SPA525g phones with a SPA500s (Attendant
Console) connected to each phone.
All of the 16 phones, have their Attendant Console configured the same
way, where they are subscribing to each of the 16 phones.

When I power on the switch, where all the phones are connected to, I
then get 16 registers, and 256 subscriptions (16 * 16) happening at
the same time.

What often happens, is that most of the peers is getting UNREACHABLE
or Lagged.... When I try to call during this time, I get a timeout...

After some minutes the Asterisk box is back in business, but at the
next round of re-subscribe, I happens again..


I have tried to reproduce this in a test environment, and I only
manage to reproduce this by adding some network delay on the asterisk
box it self..


Any ideas on where to start debugging?

I'm running on Asterisk 1.4, with realtime users, with cache and
qualify turned on.


Best regards
Glenn Larsen

Am 04.11.10 13:14, schrieb Glenn O Larsen:
> What often happens, is that most of the peers is getting UNREACHABLE
> or Lagged.... When I try to call during this time, I get a timeout...
> 
> Any ideas on where to start debugging?
> 
> I'm running on Asterisk 1.4, with realtime users, with cache and
> qualify turned on.
> 
> 
> Best regards
> Glenn Larsen
> 
Hello Glenn,

you should try the latest 1.4 svn version. There was a part of code for
the subscribe handling, which tooks very long to be proceeded and in
this time no other sip paket (like the answer for options) is read.

this cause the high timerange you can see on qualify. the problem is
that asterisk was not able to handle this amount of subscribes just in
time. with my patch this problem should not exists anymore.

best regards


stefan