[Asterisk-Users] PRI Woes continue

[Andrew McRory]

> On December 29, 2004 22:35 pm, Andrew Kohlsmith wrote:
> 
> > Well it is hard to go back to a specific configuration since I have 
> > used the system to test the rpm packages I compile.
>
> Yikes.

Yep. But there is only one way to "know for sure" that a new package is
working. I have had much success in 2004. So much that I figured all the
warnings on the list to "not upgrade a working system" were for the ultra
paranoid. Still think that is mostly true.

> > Nothing like using a production server for testing, eh? I have  
> > reverted to a (actually several) pre 1.0 release that worked well, 
> > changed the port, moved the PCI slot, changed out the motherboard 
> > three times, enabled and disabled onboard devices, tried several 
> > kernels, rerun the cabling from the smart jack, checked the 
> > powersupply voltages, UPS, power cabling, etc etc etc. Basic 
> > troubleshooting? yeah man.
> 
> That wasn't meant to be flip -- Perhaps I've just been bitten too many 
> times myself by doing the exact same thing you just did -- I back up my 
> config (going as far as to rsync or image the partition if I need) 
> before changing something like that on a production system... especially 
> something as important as our main telephone system.  :-)

Understood. I want to believe that each day will only bring improvements
to the code. Sure, I know that bugs can slip in to the updates but that
should be temporary if it happens at all... right? hahahahaha! Perhaps my
doctor is right and I am crazy.

> > I dont have a T100P lying around so I cant do much in the way of 
> > changing the interface. Yet. Before I commit to changing that I want 
> > to rule out any other possibilities... How can one determine without a 
> > shadow of a doubt that it is the card or otherwise? I have enabled all 
> > the debugging I can find BUT the output is foriegn to me... <shrug>
>
> Yeah -- I don't know -- I am the last to blame hardware (10 years as an
> embedded electronics designer does that to you) but failing everything 
> else it really does seem that this is the issue, does it not?

Yes it does, but I still want to think it is me. After all, I have spent
more time compiling RPMS than I have learning the dialplan. With that said
my production dialplan is the most basic you can get. I have an alternate,
more complicated dialplan that I am developing but I only switch that in
for brief periods testing during low traffic hours. My plan is to move the
complicated part of the dialplan to a secondary server that will handle
the the real work - VoIP calls, AVR, Voicemail, etc. over TDMoE or IAX2...
still thinking that one out. 

> Something else I learned the hard way -- have any criticial hardware
> available onhand, not at a distributor, even if they can ship overnight
> -- I have a story about a DS3 MUX that had both controllers die and the
> manufacturer shipped one overnight but UPS lost it...  true story.  
> It's expensive to have hardware sitting on the shelf idle but better
> that than be without phone service or whatever other critical system
> you've got.  :-)

yep I need another T400p, a spare engine for my truck and not one, but two
legal age females on hand in case my wife gets the flu, or worse. Of
course I have a 7206VXR here sitting on a shelf with a lot of pretty cards
stuffed in it just waiting for a chance to prove itself... but thats a
different story altogether.

> > Is there a way to log all communication on the D Channel? Have I 
> > missed some critical debugging reference? I'm going crosseyed looking, 
> > tweaking and trying the same things over again.
>
> pri debug span 1 will show you all q.931 traffic and intense will show 
> you the q.921 traffic too, but this seems deeper than that -- I am not a 
> telco expert but it certainly seems like something very low level is 
> buggered.  I am sorry I can't be more help.

Well you tried and I commend you for that. I wouldn't even be asking here
if I had even the slightest clue on what to do next. Perhaps Digium will
come through with some testing tomorow / err / I mean today.

I have to do whatever it takes to regain 99.999% reliability for my
dial-up customers. I would like to accomplish this with Asterisk, as a
proof of concept if for nothing else, but I will be forced to pull it out
of the chain if a resolution can't be found in the next couple days. This
has gone on toooo long and I am looking bad. oops.

Gotta get some sleep. Thanks for you comments!

Regards,

-- 
Andrew McRory - President/CTO 
Linux Systems Engineers, Inc. - http://www.linuxsys.com
Located in beautiful Tallahassee, Florida
Office  850-224-5737
Office  850-575-7213
Mobile  850-294-7567