Hi Gary, This may not have anything to do with your problem and you probably already checked it, but sometimes PCI boards share IRQs and I've heard (maybe even from you) that the Digium TDM boards don't like to share IRQs.
cat /proc/interrupts Just a thought... G Gary G. Hendershot wrote: > Since upgrading our production systems to 0.4.3 (tags) I have started > experiencing some objectionable behavior on one phone in particular ... As > it happens, its my boss' phone so the issue has been elevated to a rather > high priority ... Google gave me some clues but none have provided any joy > ... I am hoping that maybe my description of the problem will ring a bell > with someone who has run into similar ... Am looking for ideas on how to > solve this one ... Am hoping its just some obscure setting that I missed ... > > > this is the infrastructure ... > > Astlinux server is a Shuttle Clone AMD 1.5GHz box with a 2xFXO/2xFXS Digium > card in it ... the machine has LOTS of RAM (512MB) ... boots off a 256mb CF > and uses a 1gb USB thumb disk for KD ... 2 x FXO supports access to office > PSTN lines and we also have a SIP DID with an Internet based provider ... 2 > x FXS ports are not currently in play ... server operates in server only > mode and sits behind a firewall that supports 1 to 1 NAT ... the Astlinux > server is visible directly from the Internet ... can even be pinged ... > > The LAN is VERY strong with all ports switched at 1gb ... some of our > network devices (the Cisco phone and the Astlinux server included) are > limited by their 100mb network cards but most traffic is handled at 1gb ... > there is no QoS or VLAN in play ... > > There are only 8 workstations, 12 phones and 4 servers on the LAN ... it is > rare when more than 4 people are in the office during business hours as we > are normally at client sites ... So LAN activity is usually pretty light ... > Our worst case scenario is when we have our weekly conference calls with > various clients ... At these times, there may be as many as six simultaneous > sessions > > The internet connection is a full T1 ... we do host mail and Astlinux on our > network but our web site and other Internet based services are hosted by > third parties ... bottom line is that neither the LAN nor the Internet > connection seem to represent any sort of bottleneck that would cause what I > am about to describe ... > > > This is the scenario ... > > Am fighting two problems > > 1) on inbound PSTN calls, there is a LOT of echo on the boss' Cisco 7960 w/ > v8.2 SIP that seems to persist for a lot longer than it did back when I used > vanilla Asterisk 1.09 or Astlinux v0.3 ... issue exists when using the > handset but is MUCH more noticeable when using the speaker phone ... makes > first few seconds of the speaker phone pretty nasty ... preferred codec is > set to ULaw ... > > there is NO echo when calls originate from our SIP DID or another SIP > extension on the LAN ... have tried many different setting combinations for > ZAP echo training and TxGAIN/RxGAIN and have not managed to find a sweet > spot ... keep in mind that I had good working settings for this prior to > implementing 0.4.0 and newer ... > > 2) on PSTN calls either placed to or originated from the Cisco phone the > sound is "choppy" ... sounds like chunks of the RTP feed are being lost ... > if I were a betting man, I would guess that there is some sort of "silence > suppression" in the loop that is causing the problem but I am not sure of > this and have no idea how to find out ... again, this is a LOT more > noticeable when using the speaker phone than the handset but it is a factor > with either ... and again, I had working settings that did not produce this > choppy sound prior to 0.4.0 and newer ... > > > Summary ... > > this system was rock solid stable without any issues with vanilla Asterisk > 1.09 run against CentOS 4.2 ... it was rock solid stable when I switched to > Astlinux 0.3 a little less than a year ago ... with both setups, there was a > slightly noticeable echo on PSTN calls that seemed to go away before the end > of the first sentence but that was about it ... there was NEVER any of this > choppy sound nonsense ... > > the intensity of the echo problem increased and the choppy sound started > when I went to Astlinux 0.4.0 ... I was patient and have continued to update > through 0.4.2 ... however, since going to 0.4.3 about a month or so ago, it > has gotten a LOT worse to the point where I am back to messing around with > ZAPPATA.CONF settings and taking wild stabs at killing the gremlin .... > > I have v0.4.4 on my test bench system right now and it seems to work well in > the lab ... I am testing it with Aastra and GranStream phones with good > results ... but I don't have another Cisco phone I can play with on my test > system ... I am a bit concerned that implementing 0.4.4 on my production > system might make the problem even worse as this was my experience with > going from 0.4.2 to 0.4.3 ... > > Obvious question is, what the heck has changed recently in Asterisk that > might contribute to this problem ??? and is there anything unique to > Astlinux that might be the source of the problem ??? maybe there are NEW > settings in Asterisk that escaped me ??? > > I have found a lot of "OLD traffic" that discusses similar issues with > Asterisk that appear to have resolved shortly after the release of v1.2 ... > seems there were issues discussed that were unique to the Cisco > implementation of SIP that were fixed with some kind of patch about that > time ... I wonder if the code base used in Astlinux might be missing this > "Cisco SIP unique" patch ?!?!? the patch discussed had something to do with > RTP timing with off hand references to the way Cisco SIP handles silence > suppression ... (bug report 5374) > > any ideas or feedback on this one would be appreciated ... if I cant figure > out any other way to beat it, I can revert to 0.3.0 which seemed to work > fine ... but I really do not want to do this as I am looking forward to some > of the new features being developed for the current version ... I have asked > the boss if he would be willing to switch off to one of the Aastra 480i > phones (which do not seem to have these problems) and he looked at me like > like I was crazy ... seems he really likes that darn Cisco ... > > if posting snippets of some sort of diagnostic logging will help, please let > me know ... this issue is a big enough problem for me that I will jump > though whatever hoops are required to solve it ... > > thanks in advance ... > > G.Hendershot > > > _______________________________________________ > Astlinux-users mailing list > [email protected] > http://lists.kriscompanies.com/mailman/listinfo/astlinux-users > > Donations to support AstLinux are graciously accepted via PayPal to [EMAIL > PROTECTED] -- ============================= Gene Cooper Sonora Communications, Inc. 5531 N. Oracle Road Tucson, AZ 85704 (520) 293-8461 x101 (520) 888-4060 fax [EMAIL PROTECTED] _______________________________________________ Astlinux-users mailing list [email protected] http://lists.kriscompanies.com/mailman/listinfo/astlinux-users Donations to support AstLinux are graciously accepted via PayPal to [EMAIL PROTECTED]
