[Asterisk-Users] Asterisk 1.2 problems

[jonc]

On Fri, 2005-12-02 at 16:32, tneuwert at formos.com wrote:
> >> After working on the problem for
> >> several days, I finally built a new box and installed Asterisk 1.2 on
> >> it. Using this new 1.2 box I no longer see the "Maximum retries
> >> exceeded on call" warnings on the console but still experience the
> >> strange behavior. Unfortunately, the errors occur randomly so I am
> >> unable to reproduce the error on demand. I turned on SIP debugging and
> >> set console logging to debug and captured an instance of the problem
> >> with the hang up not being recognized.  The details are below:
> >> 
> >> I dial in from my cell phone. My Cisco phone begins to ring. I then
> >> hang up my cell phone. Asterisk acknowledges the hang up, but the Cisco
> >> phone continues to ring. After a minute or so, or if I pickup the
> >> phone, Asterisk display the following message "That's odd...  Got a
> >> response on a call we donÂ’t know about. Cseq 102 Cmd SIP/2.0"  I've
> >> included a copy of the console output when this occurs that shows both
> >> the SIP message and the Asterisk debug output.
> > 
> > Odds are you have local network congestion -- Dropped packets or delayed 
> > packets.  Try moving your phone and asterisk server to an isolated network
> > switch - no other traffic (certainly no computers) - then test.
> > 
> > If the problems go away, then update your virus scanners and check your 
> > computers.
> > 
> > Good Luck
> > 
> > Jon Carnes
> >
> 
> Thanks for the feedback.  We will definitely try this and let you know how it goes.  Do you know if there is a way for us to observe the loss of packets?  Because of the sporadic nature of the problem, we have thought several times that the problem was solved only to have it crop back up hours later.  If we ran Ethereal on the pbx, would we be able to see that packets were being dropped?
> 
> Thanks,
> Tim

I do run Ethereal on mine when looking for real-time problems. It's
great for helping you see what is going on at the packet level, but it
is the wrong tool for measuring Latency/QOS problems.

Shadow Ping works fairly well for measuring latencies. In earlier times
I used to just run a quasi-flood ping to the offending phone (10
pings/second) and look for latency variations and dropped packets. On a
clean network with no problems there should be NO dropped packets, and
latency variations should be minimal.

You'll find some interesting problems that accompany the use of cheap
unmanaged switches (and please don't tell me you are trying to use
hubs!).

For our setups we use either Cisco 2900XL-EN or Cisco 3500 series
switches.  This come with built in VoIP detection at the port level
*and* allow us to use VLANs to separate out Voice and Data. They are
champion workhorses and using them lets us also run single wire to the
desktop (running the PC off the pass-thru switch on the back of the
phone).

Good Luck,

Jon Carnes
FeatureTel