internet down, can't dial other extensions.

from what i understand (ive had this problem before) it is normal behaviour and if i remember correctly happens when you use sip trunks.

Someone posted a workaround for this - but i havent had time to try it yet.

It is a pain though - when the internet goes down, we cant monitor the door.

Here is some info on our dns setup.

I checked a Polycom's phone. I forget the exact wording, but the dns settings were for 127.0.0.1 then 192.168.1.2 . the 2-nd is our router.

Trixbox is on 192.168.1.17, and does not have a dns server running.

I figure that path to solving this would involve :
- having Trixbox running a dns server
- have the Polycom phones get dns / routing info from Trixbox instead of the firewall [192.168.1.2 ]

Any comments on those assumptions?

Pushing 127.0.0.1 as a DNS server to the phone is pointless as that tells the phone to look at itself for DNS lookups and the last I checked, the phones don't have DNS servers running on them.

In this description (http://www.voip-info.org/wiki/view/Asterisk+SIP+srvlookup) it says that Asterisk can get into a bizarre state when DNS lookups fail on registers which can cause a condition that will eventually lead to all handsets failing to register. The solution is to run DNS on the phone system and point the devices' DNS server to the phone system.

When the net connection was down there were 2 problems:
1- extension to extension calls did not work for local phones. that is fixed.
2- calls would not go out over POTS lines.

# 1 is fixed [as of now for me] by setting up bind on Trixbox and having phones get their dns from bind. here are my notes: [ note our trixbox is 192.168.1.17] .

webmin and bind were already installed.

- from cli: yum update bind # to use latest bind rpm.

- webmin : servers > bind > forwarding and transfers enter i/p addresses of your isp dns.

- cli: set bind to start at boot: chkconfig --levels 12345 named on

- cli: /etc/resolv.conf : put this line at top: nameserver 127.0.0.1

restart trixbox and make sure networking and the phone system work.

- phones: I run a dhcp on our main linux [debian] server . I edited this line in dhcpd.conf :
option domain-name-servers 192.168.1.17,192.168.1.2, 208.67.222.222;
The point is to have your dhcp server tell the dhcp clients to use trixbox 1-st for dns.
power cycled the phones and checked their setting to make sure trixbox was listed 1-st for
DNS Server .

unplug the ethernet cable from the cable modem.
dial an extension.
works for me as of now.
------------------------------------------
Next I need to be able to use SIP phones to dial out on a POTS line when the cable is unplugged from the cable modem.

These notes are a work in progress.

I think I have that fixed.
I'll restart the system and the phones a few times to see if it stays fixed.

The test I'll do is
- disconnect the internet connection.
- restart trixbox and the phones.
- test extension to extension dialing.

Have had the same situation here... tried enabling bind, tried enabling bind in caching mode with no result. I haven't had time to look closer but I'm assuming that most of these providers have their TTL set very low in DNS such that even in caching mode the cached data is expiring.

From what I've read it is certainly an asterisk issue, and has been around for a long long time (really rather surprised that more people haven't been faced with it's consequences). To me it certainly seems like a pretty major issue, and something that could bring even a major installation to it's knees in very short order.

Really hesitate to remove the fqdn's because then if the provider decides to change IP's you are stuck making hand changes once things just stop working all of a sudden (and in the case of one provider, one of their fqdn's resolves to 4 different IP's).

Was able to wrangle a workaround by changing the SIP configuration to limit the number of registration retries and the time between attempts. The end result of this is that your internal phones stop working, only until the trunks stop trying to re-register. The downside to this plan is that you would have to set up some sort of watchdog to reset sip because if during the normal course of events one of your SIP trunks went down for long enough to exceed that timeout, it would not automatically reregister.

The workaround would probably be a better solution for someone running mostly POTS lines with perhaps only one or two SIP lines as "backup" (ironic that your backup failing makes your primary stop working :-)

Until I can come up with a better solution I've elected to leave things as-is, armed with the knowledge that a quick edit to the SIP configuration can make the problem go away if it's needed in an outage.

I think the biggest problem with this issue is that it slaps you out of nowhere. It first hit me when I was quite literally on the way out the door (late already) for a business trip. Absolutely *fantastic* way to raise your blood pressure :-)

Nobody expects internal phones to stop working when the internet connection goes down... even if you think of a DNS issue you then dismiss it... because none of your phones have a stitch of configuration on them that would require a DNS lookup so that couldn't possibly be the problem. Last thing you think of is that the entire SIP stack is going to take a nosedive.

Cheers,
>>>>> Mike

installing bind appears to have worked for our server. After pulling the enternet from the wireless modem and did a amportal restart. Prior to installing bind this would cause everything to break, now everything continues on as normal, apart from the 4 sip trunks obviously not being registered.

On a testing computer installed Trixbox 2.4.2 and upgraded rpm's and Freepbx to latest. Then configured trunks , outbound routes and added extensions. I did not restore a backup from my production system - I wanted a clean install. Both systems have a Sangoma 200 pci card for pots lines.

Outbound route is set to voip then pots trunk.

On my test system, when there is not an internet connection, calls try to go out on voip, then after about 20 secs go out on the pots trunk.

So when the internet will be down for more than a little while disable the voip trunks on the FreePbx screen.

There needs to be a way setup to monitor the internet connection and auto disable voip trunks. Then re-enable voip trunks when the internet is back up. Maybe it should be on a bounty page.

At least for now I can document and have office staff disable voip trunks as needed.

see my post above "Almost Solved"

A new clean install works ok without using bind or caching-nameserver

If I restore a backup from my production system to the test system then the test system does not work when the internet is down.

So we are hand entering all of our setup from the production system to the test one. After each step I retest the new system . the test is disconnect the internet connection , and see that calls out make it to the pots trunk.

wonderful, work for me!
Hongbo, I've follow your indications, and internal calls are possible without internet connection.
Thank's a lot.
Even though, when start trixbox some utilities start slowly:sendmail, sm-client, smartd and then appear:
could not reliably determine the server's fully qualified domain name using 127.0.0.1 for servername.
when internet connection returs, voip external calls have no audio, internal have audio.

It's seem work now!
I will do other tests, and then you say.
But I have this other problem, without resolution. I've an IP dynamic internet adsl connection, and I use Linksys SPA942 IP phone. When my connection goes down, and then returns, I can call, but there is no sound. This only in external telephone calls, because those internal everything worked ok. Have you any idea?
Thank's

my situation is:
router draytek 2700v
adsl connection with dynamic IP
I've open 5060 tcp port on router, disable all dos defence, but when connection go down and then reconnect, my isp ip change and I've no audio in external calls.
This is active channels during a call test:

Peer User/ANR Call ID Seq (Tx/Rx) Format Hold Last Message
83.211.227.21 3201909210 67f6f8b14d6 00103/00000 0x4 (ulaw) No Tx: INVITE
192.168.1.12 201 3d0c9c7e-78 00101/00102 0x4 (ulaw) No Rx: INVITE
2 active SIP channels
3201909210 is my phone cell, 201 is my spa942
I've also rebooted my router, and rebooted to factory default my spa942, but nothing.
When reboot trixbox, everything works fine

You need to forward UDP 5060 and UDP 10001 - 20000 (as defined in rtp.conf) from router to trixbox. If you can ring but there is no sound, most likely RTP ports are not forwarded.

did you use to install your system?

trixbox 2.4.2 works for me when the net is down. see my post above.

earlier versions did not . could be do to something I mis-configured or what I do not know.

The problem of the missing voice on external calls always happens when change the IP address assigned by the provider;with a simple reset of trixbox everything works well.
I have tried many settings, but always the problem. I think we should reset trixbox whenever lose the dsl signal.

Here is my solution using BIND DNS server (via Webmin).
1) First go in Webmin and under BIN DNS Server select "Setup nameserver for internal non-internet use only".
2) Create a master zone : trixbox1.localdomain
3) go in "Forward and Transfer" and set IP address of two (or more) Internet DNS server (public DNS Server as OpensDNS Server : 208.67.222.222, 208.67.220.220 or those provided from your ISP) and set "Lookup directly if forwarders cannot?" to YES - no default - (this is the most important thing!)
4) Go in Trixbox -> System -> Network and set as Primary DNS the internal IP Address of your TB server.
In this way all works fine even when Internet goes down.
Bye.
Ing. Cosimo MERCURO
Mercuro for Business

Something else is setup wrong. There is no earth shattering bug. We have systems with SIP and IAX trunks that have never been connected to the Internet. Some do have Zap channels in them also.

I would like to find the common issue with your type of problem so that we can better help users experiencing this issue.

Scott

does the problem occur when using IAX2 trunk?

About IAX2, no, we don't have any.

About manually coding the IP address, sure, i can do that with the hosts file. Don't even need bind. But, what's going to happen when a SIP provider changes an IP address for one of their servers?

After scouring for hours and trying various solutions, I still find myself with this problem at 2 sites (at least) Did I miss something? Is there a fix for this from the Trixbox side available? This seems to be a rather common problem killing systems!

I HAVE to use DNS names and I have set up a Bind server on my * box, although I am not 100% sure that it was done correctly. (The box resolves DNS names though....?) I have pointed my DNS on my SIP phones (AAstra, SNOM, Grandstream & Sipura) to the * box, but with no luck. If I disconnect the internet, my whole system goes down in seconds, only to work 100% when internet is restored. Extremely frustrating!!!!!

Can somebody with good Linux + Trixbox knowledge not write a foolproof article on how to permanently fix this problem?

This is a known problem with Asterisk and the best thing to do is to use a local DNS server. If that is not working then something you have it not setup properly.

We are just living with the reality that the phones go down frequently. I have a POTS line that rings when the trixbox is dead.

i am getting same issue, and not able to understand what happens to my asterisk. One more thing i am using local DNS server, and that server is forward to opendns. but still same issue, but when i switch my internet link every thing again comes back and work fine.

any other solution please, if any update is available ?

Naveed Razaq | www.beyondweblogs.com | www.apnimarzi.com | www.techno-soft.com

There is an Elastix tutorial at http://www.elastixconnection.com/index.php?option=com_content&tas... which addresses this issue, it is quick to install, easy to set up and cures the problem.

Hope this helps.