Summary:ASTERISK-17243: Cannot originate DAHDI PRI calls after upgrade
Reporter:John Covert (jcovert)Labels:
Date Opened:2011-01-14 00:59:21.000-0600Date Closed:2011-06-07 14:00:43
Versions:Frequency of
Environment:Attachments:( 0) chan_dahdi.conf
Description:    -- Launching dial(DAHDI/r1/15074554436) on SIP/covert.org-00000000
[14-Jan-2011 01:33:57.09] WARNING[12861]: chan_dahdi.c:5004 dahdi_confmute: DAHDI confmute(0) failed on channel 1: Invalid argument
   -- Couldn't call r1/15074554436
[14-Jan-2011 01:33:57.09] WARNING[12861]: chan_dahdi.c:5004 dahdi_confmute: DAHDI confmute(0) failed on channel 1: Invalid argument
[14-Jan-2011 01:33:57.09] WARNING[12861]: chan_dahdi.c:4947 restore_gains: Unable to restore gains: Invalid argument
[14-Jan-2011 01:33:57.09] WARNING[12861]: chan_dahdi.c:4630 reset_conf: Failed to reset conferencing on channel 1: Invalid argument
   -- Hungup 'DAHDI/1-1'
 == Everyone is busy/congested at this time (0:0/0/0)

Downgraded back to and calls complete again.



chan_dahdi.conf attached

Comments:By: John Covert (jcovert) 2011-01-14 01:04:36.000-0600

Object lesson (as I told my client when the hardware arrived long ago):

Buy real Digium PRI cards.  Any savings from lower priced cards from other
vendors will be lost in staff or consultant time and fees on future upgrades.

That message worked today for a different client who sent back the competitive cards and ordered Digium replacements.

By: Leif Madsen (lmadsen) 2011-01-17 08:30:48.000-0600

Is there really enough information to go on here? I see chan_dahdi.conf, but beyond that, I'm not sure where the developer should be looking. Perhaps I'm missing something obvious though.

By: John Covert (jcovert) 2011-01-19 10:43:39.000-0600

The above output is the result of:

asterisk3*CLI> originate sip/37@covert.org application dial DAHDI/r1/15074554436

so there's not a complicated situation here.

It works in  libpri, dahdi, and wanpipe are unchanged when downgrading to

Looking at the code, we have just called

 ioctl(p->subs[SUB_REAL].dfd, DAHDI_CONFMUTE, &x);

which apparently returns with invalid argument.  This appears to me to be a call down into the driver.  I haven't spent enough time in the chan_dahdi code to know why we would be calling confmute as a result of a simple dial, but we do.  I suppose we're making sure that we are unmuted at the beginning of the call.  A diff shows significant changes with confmute between the two versions.

My comments about buying real Digium PRI cards are, of course, personal, based on my past contracting relationship with Digium, as well as a "keep it simple" mentality: why bring in another set of drivers which make it take extra steps to upgrade and add another support organization into the mix when there is a problem.

I have ticket #[XXXX]02625 open with the PRI card vendor, who says, "lets see if we can change your opinion about that," and offered to work with me.  He also says, "Now the issue you have here should not be [our product] related as it appears to be more of a Asterisk/libpri issue. But we will help you get to the bottom of this and get everything sorted out for you."  So I'm about to contact him now.

Production system, but fortunately not active late at night; we can only test then.

By: John Covert (jcovert) 2011-01-19 11:16:20.000-0600

Other vendor believes dahdi-linux-complete-2.2.1+2.2.1 is simply not compatible with asterisk-1.8, and that I will need to upgrade both dahdi and their drivers.

Is there a matrix published somewhere?

By: Leif Madsen (lmadsen) 2011-01-19 14:50:48.000-0600

There is no matrix, but 2.4.0+2.4.0 is generally considered the version of DAHDI that you should be running.

By: John Covert (jcovert) 2011-01-21 08:46:12.000-0600

Unfortunately my client has decided that upgrading DAHDI and wanpipe is too risky.  Of all their systems, this one has been stable on (possibly because it has the lowest volume, serves a single customer at the end of a 47 channel NAS PRI, has no ldap or gtalk or skype), so they have decided to leave it behind the rest of the systems in the group.

Please close this.