Sip Internet down bug
Submitted by SwiftNets on Wed, 07/23/2008 - 3:32am.
Hi Everyone
I'm sure I can remember someone mentioning a workaround for the internet down - sip dying bug where you run a sip trunk around to the box itself so something can register, but can't find it anywhere.
I have searched and searched, can anyone enlighten me as to where this may be, or a way to implement it as installing bind has not helped in this system.
Is there someother work around I can try, other than making them fix their internet >:(?
Michael
There is no effective workaround
Wed, 07/23/2008 - 5:21am
Unfortunately, here are your options:
1. Don't use SIP. (kind of the main reason I installed trixbox)
2. Use DNS and lose your entire phone system whenever the Internet goes down. Including your backup Zap lines and internal intercom calls! EVERYTHING!
3. Hard code the IP addresses of your sip providers in your hosts file. Now you will lose those SIP trunks whenever your SIP provider changes the IP address of their server, until you manually go in and fix it.
I opted for number 3. I was at 2 for a year or more, but it was intolerable, everyday getting up and wondering if the phones work today...
I am ASTOUNDED that the folks working on Asterisk don't consider this an absolute #1 top priority. It has been a documented bug in Asterisk for longer than i can remember...for years.
Some people have suggested installing Bind on your server to act as a local dns cache. This does help. It extends the window from immediate failure to failure after 1 hour (when the cache expires). Better than nothing I guess.
I'm not complaining, mind you. This system is free and it rocks. But these are the facts.
Thanks for that, I've gone
Fri, 07/25/2008 - 5:08am
Thanks for that, I've gone option 3 as well.
I just thought I could remember someone mentioning a work around where if Asterisk could sort out at least one trunk, it would calm down and let things happen.
I'm not too fussed about the providers changing IP addresses, I always have at least two in each install, so it's unlikely that both would change their IP before I have a chance to change them in trixbox, especially now I actually work for one of them ;)
Michael
Forgot to mention the box
Sat, 07/26/2008 - 12:48am
DNS
Sat, 07/26/2008 - 3:09am
Ok this is NOT an Asterisk thing to fix.
I myself with 100 of installs have never had this issue that was not fixed by a local DNS cache, I have fixed this issue for MANY folks over the past few years, without fail.
so if a DNS box is not working for you, as it does for the 99.9999 rest of us, you need to look at your setup.
If this was a "bug" without solution / work around Asterisk would be deployed around the world and be in production at so many places.
A good rule of thumb about issues is to dig up as much info as you can about what you think is a bug, if you see that the amount of posting about your issue is thin at best, and most post point to the same fix, then you can rest in the fact that the "fix" does in fact work, and that YOUR problem is not the same as others.
DIG DEEPER IN YOUR SETUP.
One must look for the common ground in failing setups, not the common ground in working setups.
I have seen off the shelf routers with SPI (stateful packet inspection) stop a SIP trunk form registering, if not being the cause of the trunk to drop, the CHEAPO router was sure it was a DDOS attack or something so it would reset itself and auto block sip traffic, solution was to disable SPI.
Thanks Bubba
Although I do
Sat, 07/26/2008 - 7:06pm
Thanks Bubba
Although I do agree with you, I have use various different modem+routers/gateways in this particular setup, none of which had any problem transporting SIP when it was available, and the box still struggles with DNS.
I haven't been there to time and don't have an hour spare at the moment to spend at this clients office to test, but I suspect Basildene is right above when he mentions it only gives them an hour until the cache expires. When I quickly tested the local caching dns install I put in at this office, the workaround did work, but I didn't stay there an hour. I realised the cache would expire eventually, but wasn't sure when. I don't doubt there is a setting to extend that further, but that has other undesirable consequences.
Is it possible that most/some of the other '99.9999%' have just never had their internet down for longer than an hour (or their caches expiry) so those admins don't know if it has "fixed" the bug or just delayed. I know after i installed bind, I just pulled the plug out of the modem and watched for about 5 - 10 mins. All seemed ok so plugged it back in and left. As far as I knew that was sorted out, as that was what everyone was suggesting here on the forum.
So imagine my surprise when they told me the phones were down again. They had no internet for a week, this obviously went over the hour buffer so they were back to their old nothing working again.
As I mentioned above, I hardcoded the ip addresses in to the trunks and within a minute, the phones were all available again and they could make and receive calls on their POTS line.
I know I will not bother with local DNS caching again, nor trying to use FQDNs in any sip trunk, not until * work out this bug. It's straight IPs for me
Michael
Member Since:
2007-04-16