[asterisk-users] TDM 400 hardware(?) issue

[Steve Totaro]

On Mon, Dec 21, 2009 at 8:23 PM, Darrick Hartman
<dhartman at djhsolutions.com>wrote:

> On 12/21/2009 05:34 PM, Steve Totaro wrote:
> >
> >
> > On Mon, Dec 21, 2009 at 6:06 PM, Greg Woods <greg at gregandeva.net
> > <mailto:greg at gregandeva.net>> wrote:
> >
> >     I am having to abandon asterisk after having used it for 2.5 years
> due
> >     to this problem. Every couple of days (sometimes more often,
> sometimes
> >     less), the machine will lock up because the TDM board or the Dahdi
> >     driver goes south. /var/log/messages starts filling up with repeated
> >     messages:
> >
> >     kernel: TDM PCI Master abort
> >
> >     The card I have is:
> >
> >     *CLI> dahdi show status
> >     Description                              Alarms     IRQ        bpviol
> >     CRC4
> >     Wildcard TDM400P REV I Board 5           OK         0          0
> >     0
> >
> >
> >     I asked about this once before and I am asking again in desperation,
> as
> >     I have had to shut off my asterisk server, take all the VOIP phones
> out
> >     of service, and go back to the bad old days of a single cordless
> phone
> >     base with a couple of handsets and a crappy old WalMart answering
> >     machine. This sucks.
> >
> >     Last time I asked, the most helpful answer I got was privately,
> saying
> >     that either my card is bad, or I might need a different motherboard
> (but
> >     there is no way I know of to know which motherboards would work and
> >     which would not; the last thing I would want is to go through the
> >     expense and major hassle of swapping motherboards only to find out
> that
> >     the problem is not fixed).
> >
> >     The only decent diagnostic I have is that if I catch it soon enough,
> >     before the system totally locks up, then stopping asterisk,
> restarting
> >     dahdi, and starting asterisk gets things working again (until the
> next
> >     incident). Also, I can go into asterisk -r and do "dahdi show status"
> >     and the card doesn't have any alarms; the output is the same as
> above,
> >     even as the PCI Master abort messages are spewing into the syslog.
> >
> >     If my Wildcard TDM board is bad, is there anything I can do about it,
> or
> >     am I just S.O.L. after this much time? The blasted card costs as much
> as
> >     a new machine; either way I can't afford it right now. I don't want
> to
> >     abandon asterisk as it has so many nice features, but I am running
> low
> >     on alternatives at the moment.
> >
> >     --Greg
>
> Greg,
>
> Why don't you contact Digium tech support?  They should be able to help
> you narrow down the problem.  Cards do go bad from time to time.
>

Time to time, I would say with great regularity, especially since most
people don't put surge protection on their POTS lines.


>
> Now on to Steve's reply...
>
> > How many lines are you talking about?  In light of your budget issues, I
> > would switch to quality SIP provider and have my numbers ported.
> >
> > That would most likely be cheaper in the long and short run, and more
> > reliable depending on the vendor and your internet connection.
>
> I agree, especially for a small office or home.  You can set up most SIP
> providers to failover to your cell phone if there is a problem with the
> SIP connection.  Do this and you won't need a hardware card.
>
> > Other options are going back to old versions of Asterisk.  What version
> > are you running?  What was wrong with the version from 1.5 years ago?
> > Maybe your card likes being a Zap device, rather than a DAHDI.
>
> Seriously?  This makes no sense at all.  Even early TDM400p cards will
> work with dahdi, usually better than they did with zaptel, but no worse.
>  The version of Asterisk and zaptel from 1.5 years ago is likely full
> of bugs that have been fixed by recent versions.  Are you still driving
> your 1978 Ford Pinto?  Someone from Digium made a post to that effect a
> week or so ago.  If you're running version 1.4, you should be running
> 1.4.28 with the latest version of dahdi.
>

Do you believe everything you hear/read.  One of the major Digium dev guys
once said that his code was not "Beta" anymore, and therefore bug free....
I still get a kick out of that one.

Yes, seriously.  The OP made it sound like a new problem.  Obviously he was
not running DAHDI 2.5 years ago.  When cards and/or modules go bad, they
generally completely die.  Newer code introduces new bugs.

Revert to the version that worked for you previously.


>
>
> You could make a cron job to reboot the machine at midnight, daily.
>
> Bandaid/ducttape?  The only thing this may 'solve' is a memory leak.
> It's really hiding the underlying problem.
>

If it works then it works.


>
>
> I have a box full of Digium cards with all sorts of modules, I could
> > "sell" you what you need for the price of postage but I really think SIP
> > is your silver bullet.
>
> That would be nice of you, but he should find out the problem before
> throwing more hardware at the issue.
>

What can I say, I am a nice guy.  I will not be deploying any of these cards
except for timing and a empty card with no modules does just fine.  Anyways,
it is hard to find any new server boxen with the older PCI slots.

Above and beyond that, I know people are hurting out there and I would like
to help.


> Darrick
> --
> Darrick Hartman
> DJH Solutions, LLC
> http://www.djhsolutions.com
>
>
Thanks,
Steve Totaro
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.digium.com/pipermail/asterisk-users/attachments/20091221/b65eef66/attachment.htm