Registration Problems

[jmattwood]

 (no NAT) to 80.249.108.21:5060
Retransmitting #1 (no NAT):
REGISTER sip:registrar.voip.co.uk SIP/2.0
Via: SIP/2.0/UDP 127.0.0.1:5060;branch=z9hG4bK1d795eae
From: <sip:snipped@registrar.voip.co.uk>;tag=as2dffb7b8
To: <sip:snipped@registrar.voip.co.uk>;tag=ac21416b
Call-ID: 3cdb4bff18af1ff10f02267206156bfd@192.168.1.2
CSeq: 187 REGISTER
User-Agent: Asterisk PBX
Proxy-Authorization: Digest username="snipped", realm="registrar.voip.co.uk", algorithm=MD5, uri="sip:registrar.voip.co.uk", nonce="1139072913:600cfd9dd095536730fac365722a91ca", response="de63eec95461905435fa1c0bd6f6921c", opaque=""
Expires: 120
Contact: <sip:snipped@127.0.0.1>
Event: registration
Content-Length: 0

Above is taken from CLI whilst sip debug ip is set.

Third line up from the bottom should show, I presume, my public IP address, not the loopback address of the server.

Sip.conf has the recommended externip set properly & I have not made any changes that would cause this, as far as I'm aware. It is, however, causing outbound calls to be silent & inbound to be completely unobtainable, with one provider only.

[supa]

I have a very simillar problem:

-- Got SIP response 500 "Internal Server Error" back from 147.135.20.128
Destroying call '78a1df8376af856b5be47d08526ba340@127.0.0.1'
REGISTER 12 headers, 0 lines
Reliably Transmitting (no NAT) to 147.135.20.128:5060:
REGISTER sip:sip.broadvoice.com SIP/2.0
Via: SIP/2.0/UDP 69.162.210.234:5060;branch=z9hG4bK2d58f86d;rport
From: <sip:7169541991@sip.boradvoice.com>;tag=as0bdc8804
To: <sip:7169541991@sip.boradvoice.com>
Call-ID: 78a1df8376af856b5be47d08526ba340@127.0.0.1
CSeq: 171 REGISTER
User-Agent: Asterisk PBX
Max-Forwards: 70
Expires: 120

[jmattwood]

The original ITSP I mentioned has forced a fix, which has got me up & running again, although they say it is a temporary fix.

Quote from thei mail: "i'll temporarily "fix" your issue by forcing 127.0.0/24 contact
addresses to be re-written to the source ip + port the REGISTER came
from on our end rather than just blank outright rejecting it. I can't
keep it there forever though as it indicates a configuration error that
can lead to other problems, which obviously makes it harder to diagnose
issues when the cause if the SIP UA thinking it's local contact address
is 127.0.0.1"

Unfortunatley, it's started happening, now with a second ITSP.

to 217.10.79.219:5060
Retransmitting #3 (no NAT):
REGISTER sip:sipgate.co.uk SIP/2.0
Via: SIP/2.0/UDP 127.0.0.1:5060;branch=z9hG4bK2a308723
From: <sip:SNIPPED@sipgate.co.uk>;tag=as2e81f810
To: <sip:SNIPPED@sipgate.co.uk>
Call-ID: 59cc6eee45ddee167edce25c7dbb4495@127.0.0.1
CSeq: 124 REGISTER
User-Agent: Asterisk PBX
Authorization: Digest username="SNIPPED", realm="sipgate.co.uk",
algorithm=MD5,
uri="sip:sipgate.co.uk", nonce="43e7989e0d1e8ad0e5db4ae73c025bf8d9a5bb30",
response="15ba591ac679b5346904eb3e170ec384", opaque=""
Expires: 120
Contact: <sip:SNIPPED@127.0.0.1>
Event: registration
Content-Length: 0

Now, although I've fiddled & fiddled and am now seeing my external IP address in the headers, it's gone to one way audio. Pah!

[supa]

I changed the "Bind Address", in my SIP.conf, and that got rid of the 127 address. Im going to update and then but the PBX out on the DMZ to see if it's just a stupid natting issue.

ill let you knw

[jmattwood]

What did you change it from/to?

[chandave]

This usually happens when Asterisk cannot determine the address of the Network Interface Card that it should use to send out packets.

This might work so long as you don't have 2 network interface cards:

1. At the linux command line, issue the following: /sbin/ifconfig eth0
2. Look for the IP address next to the "inet addr:" line. It will be the IP address of the Network Interface Card. Let's assume it says: inet addr: 192.168.0.2
3. Edit your /etc/asterisk/sip.conf and change/add line in the [general] section the line:
   bindaddr=192.168.0.2
4. Restart your asterisk instance.

This will force your asterisk server to listen for connections only on the Network Interface with the IP addresss of 192.168.0.2. It also has the side effect that it can only use that interface to send out packets. If you are behind a NAT, then you will also need to fill in your externip= and localnet= entries to reflect your public network address and private network addressing scheme.

See ya...

d.c.

[jmattwood]

Thanks for the explicit suggestions, once again.

It had been suggested that I used the address of the asterisk network card in bindaddr. Having it set as you suggest appeared to cause the 127.0.0.1 address to be sent to the ITSP (Sipgate). Putting it back to 0.0.0.0 seemed to fix it.........I could equally be mistaken, though.

[chandave]

Actually, I just noticed something after reading one of Ward Mundy's old Nerd Vittles postings. Is the hostname of your machine resolving to 127.0.0.1?

I noticed that Ward suggested that you place your machine's Fully Qualified Domain Name along side of the 127.0.0.1 entry of your /etc/hosts (http://mundy.org/blog/index.php?m=200508). As a result, it looks like your Asterisk box might be resolving your hostname to 127.0.0.1 when DNS does not resolve fast enough.

See ya...

d.c.

[jmattwood]

Yeah, I have been considering exactly the same thing. What would it break if I removed 127.0.0.1 from the hosts file?

Anyway, I've since bought a new router & installed the DD-WRT (voip) firmware. I must say I'm very impressed. Everything works much better now, registrations included, even though I have removed my externip=my.dyndns.org from sip.conf, turned off all port forwarding & don't have my asterisk in the DMZ.

[mberlant]

It will not break. In fact, it should fix your problem. The FQDN should go in your sip.conf file, so that Asterisk can resolve it (whenever sip.conf gets reloaded) and can put the current IP address into SIP packets. Having your hosts file resolve your own FQDN to 127.0.0.1 might make some timers happy, but it will confuse any SIP server that needs the real IP address to be in the payload of the Register packet (and not just in the "return address" of the header).