asterisk users - Apr 2006 - Why is the internet connection important to LAN and PSTN calls?

Out internet connection was out this morning. It seems that the SIP
extensions on our LAN were affected. Behavior like:

Call comes in over POTS to a TDM400P, there is a delay then before the Cisco
79[46]0's start to ring.
If we were lucky enough to get a call through, then we could not transfer
the call, or place the call on hold, or park the call.
Outbound calls seemed to have a delay between the time they were dialed at
the SIP phone and when they were connected.

I know this has been brought up before, in fact there is a bit of a
discussion going on now about DNS SRV (in sip.conf, set srvlookup=no, or put
all the phone ip's on /etc/hosts). But what is really causing the issue
here? Yes, it is DNS, or something related to DNS, but why does that have
anything to do with * trying to make a phone ring on the LAN?

I would think that by using qualify=yes for any outbound voip trunks we
avoid an issue of trying to call out is the net is down, but why are any
operations on the LAN affected?


Sincerely,

Brent A. Torrenga
brent.torrenga@torrenga.com

Torrenga Engineering, Inc.
907 Ridge Road
Munster, Indiana 46321-1771

+1 219 836 8918 x325 Voice
+1 219 836 1138 Facsimile
www.torrenga.com

because, a this time, the sip stack doesn't have asynchronous DNS... so ALL
the sip code is stucked waiting timeouts for DNS queries (that are long
timeouts).
When you try to dial a LAN device, the sip code is trying to resolve your
voISP service providers' addresses.
We workaround this putting all external sip peers in a separate file, say
"sip_peers.conf", included in sip.conf with "#include
filename".
Then, a daemon on the box try to resolve well-known addresses on well-known
DNS servers on the net, every 5 minutes. If the demon fails ALL the
well-known DNS queries, it assumes no internet connection is available: then
it renames sip_peers.conf, and ask asterisk a "sip reload". So all
external
sip references are out, and sip still continue working for internal phones.
Needless to say, when connection come up again, the daemon do the opposite
thing.

hope this helps


2006/4/11, Brent Torrenga <lists@torrenga.com>:
>
> Out internet connection was out this morning. It seems that the SIP
> extensions on our LAN were affected. Behavior like:
>
> Call comes in over POTS to a TDM400P, there is a delay then before the
> Cisco
> 79[46]0's start to ring.
> If we were lucky enough to get a call through, then we could not transfer
> the call, or place the call on hold, or park the call.
> Outbound calls seemed to have a delay between the time they were dialed at
> the SIP phone and when they were connected.
>
> I know this has been brought up before, in fact there is a bit of a
> discussion going on now about DNS SRV (in sip.conf, set srvlookup=no, or
> put
> all the phone ip's on /etc/hosts). But what is really causing the issue
> here? Yes, it is DNS, or something related to DNS, but why does that have
> anything to do with * trying to make a phone ring on the LAN?
>
> I would think that by using qualify=yes for any outbound voip trunks we
> avoid an issue of trying to call out is the net is down, but why are any
> operations on the LAN affected?
>
>
> Sincerely,
>
> Brent A. Torrenga
> brent.torrenga@torrenga.com
>
> Torrenga Engineering, Inc.
> 907 Ridge Road
> Munster, Indiana 46321-1771
>
> +1 219 836 8918 x325 Voice
> +1 219 836 1138 Facsimile
> www.torrenga.com
>
> _______________________________________________
> --Bandwidth and Colocation provided by Easynews.com --
>
> Asterisk-Users mailing list
> To UNSUBSCRIBE or update options visit:
>   http://lists.digium.com/mailman/listinfo/asterisk-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.digium.com/pipermail/asterisk-users/attachments/20060411/5dfc4455/attachment.htm

11 apr 2006 kl. 16.05 skrev Brent Torrenga:
> Out internet connection was out this morning. It seems that the SIP
> extensions on our LAN were affected. Behavior like:
>
> Call comes in over POTS to a TDM400P, there is a delay then before  
> the Cisco
> 79[46]0's start to ring.
> If we were lucky enough to get a call through, then we could not  
> transfer
> the call, or place the call on hold, or park the call.
> Outbound calls seemed to have a delay between the time they were  
> dialed at
> the SIP phone and when they were connected.
>
> I know this has been brought up before, in fact there is a bit of a
> discussion going on now about DNS SRV (in sip.conf, set  
> srvlookup=no, or put
> all the phone ip's on /etc/hosts). But what is really causing the  
> issue
> here? Yes, it is DNS, or something related to DNS, but why does  
> that have
> anything to do with * trying to make a phone ring on the LAN?
The SRVLOOKUP setting has nothing to do with this, Asterisk will send  
DNS
queries anyway. I just answered a similar question in another mail,  
so check that.

If DNS does not work on your local network, Asterisk will lock up.

/O

>
> If DNS does not work on your local network, Asterisk will lock up.
Out of curiosity - the async implementation you mentioned in the other
thread - will it replace gethostbyname with something smarter or just
run things in a different thread asynchronously?

Thanks,
Cristi

12 apr 2006 kl. 09.08 skrev Cristian Draghici:
>>
>> If DNS does not work on your local network, Asterisk will lock up.
>
> Out of curiosity - the async implementation you mentioned in the other
> thread - will it replace gethostbyname with something smarter or just
> run things in a different thread asynchronously?
I am not personally involved in the details, but as far as I know, it  
will
replace gethostbyname with something smarter.

/O