[asterisk-users] Asterisk 16.6.1: PJSIP: delayed action of core since update to 16.6.1

[Joshua C. Colp]

On Wed, Nov 20, 2019 at 5:40 AM O. Hartmann <o.hartmann at walstatt.org> wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA256
>
> Am Sat, 16 Nov 2019 07:39:08 -0400
> "Joshua C. Colp" <jcolp at sangoma.com> schrieb:
>
> > On Sat, Nov 16, 2019 at 4:07 AM O. Hartmann <ohartmann at walstatt.org>
> wrote:
> >
> > > -----BEGIN PGP SIGNED MESSAGE-----
> > > Hash: SHA256
> > >
> > >
> > > Hello,
> > >
> > > we're running a small Asterisk appliance on a PCengine APU2C4. Base
> > > operating system is
> > > FreeBSD 12-STABLE, most recent incarnation as of today.
> > >
> > > Since update of port net/asterisk16 to the latest bug fix revision
> 16.6.1,
> > > we face a severe
> > > "slowdown" of everything that the Asterisk core performs, i.e. outgoing
> > > calls are delayed ~ 20
> > > seconds and I guess incoming calls suffer the same until they gett
> patched
> > > through to an
> > > endpoint/telephone. We also register a higher load on idle asterisk
> > > process since the last
> > > update.
> > >
> > > Here is an example when calling two attached physical phones directly,
> > > which performed prior
> > > to 16.6.1 almost immediately and now takes up to 30 seconds to make the
> > > called ednpoint ring.
> > >
> > > The calling phone/endpoint sinals by callsound that it is calling, and
> the
> > > sound changes then
> > > (some kind of different octave/tune, don't know) when the asterisk core
> > > reports
> > >
> > > [Nov 15 13:21:24]   == Using SIP RTP Audio TOS bits 184
> > >
> > > (see below). It is here approx 10 seconds, but there are situations
> were
> > > it might more (as
> > > observed). the host has no further load so far!
> > >
> > > Incoming testcalls we made from wireless/mobile show the same. It
> seems,
> > > asterisk is acting as
> > > a black hole delaying device for approx 10 seconds until it decides to
> > > pass the call through
> > > to an endpoint and then it takes another 10 seconds until the endpoint
> > > starts ringing (it is
> > > in fact a group of phones ringing alltogether).
> > >
> > > I can not see anything unusual with the underlying OS or some critical
> > > debug messages from
> > > asterisk itself.
> > >
> > > Any ideas?
> > >
> >
> > Do you have a STUN server configured in rtp.conf? If you do, is it
> > reachable, does the problem go away if you remove it?
> >
>
> Is there anything wrong/buggy with the implementation of the STUN service
> in 16.6.1?
>

In that specific version? No. That code itself hasn't been touched in
years, so the problem applies to every version. Using the defined STUN
server to get a server reflexive ICE candidate is a blocking process. If
the server isn't reachable or is extremely slow then it has to wait until a
timeout occurs, causing a delay.

-- 
Joshua C. Colp
Digium - A Sangoma Company | Senior Software Developer
445 Jan Davis Drive NW - Huntsville, AL 35806 - US
Check us out at: www.sangoma.com & www.asterisk.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-users/attachments/20191120/5322d159/attachment.html>