>The Digium cards actually are sharing IRQs with other devices -- the >installer mentioned it could be an issue initially, but when he saw that >the devices that the cards were sharing with were the network card and >the video card, he said to just try and see if it works first. Sounds >like some of the problems that we're having could be related to this, >and it's probably the first thing I should try changing.
Definitely sort out the IRQ problems first. Absolutely, the TDM cards require a distinct IRQ. Disable every onboard device if it isn't needed including USB, sound card, parallel port, serial port, etc (to give you more IRQ choices). Run /usr/src/zaptel/zttest before and after and see if it makes any difference. If it does (i.e. ZTTEST yields ~98.X% before and ~99.X% after) then right there that's going to help. I have used these statements in /etc/rc.d/rc.local to good effect (YMMV) setpci -v -s 01:04:0 latency_timer=ff <--Specific statement for Digium TDM04, you will have to do lspci -vv for the pci vendor id of your specific Digium card setpci -v -s XX:XX:X latency_timer=0 <-- Replace XX:XX:X with the PCI Vendor ID of every other card that lspci -vv recognizes. What the setpci statements do is allow the Digium cards to hog the bus more than any other card. It is controversial as to whether this actually does anything; some guys claim not, however, it worked for me. >If the sound quality is poor, I'll try hooking up the phones to >a new network card. Um, yup. In fact, I'd ditch *any* embedded NIC unless it was eepro100 or 3com (not very common these days) 'cause embedded these days is pretty much crapola. >One additional question -- are VoIP lines generally easier to get going >w/ good sound quality than POTS lines? Yes and no. They are different animals. Asterisk bridges the two, but the kind of latency/ echos / bad call quality etc issues are on the same order of magnitude for PSTN and VoIP. They just require different methodologies and training to troubleshoot. It's easy to set up a crappy VoIP link. >Another thing I was wondering is >whether we could get hunting to work properly with a mix of VoIP and >POTS. I'll call the phone company tomorrow, but if anyone has tried >anything like this, I'd like to hear about it. This is all dialplan stuff. A crude way to do it is to set your extensions.conf to Dial(ZAP/g0/xxxxxxxx) to use your PSTN lines and to Dial(IAX2/foo:[EMAIL PROTECTED]) as the next priority, so if all of your ZAP channels are busy, it will dial IAX/SIP after the ZAP priority fails. Problem there, is for inbound calls. In this scenario, if you have your inbound number that rings to a ZAP channel, a caller will ring busy if the zap channel is in use. If that works for you, fine. If not, you could work with this scenario by having all of your inbound numbers come in thru a VoIP provider and all your outbound thru ZAP with a failover option to IAX2/your provider if the zap channels were all busy. If you want to get fancy, some providers will failover to a PSTN number of your choice if your server is not responding. The nice thing in this scenario is that an IAX or SIP provider changes the definition of what a "line" is so theoretically you can have hundreds of simultaneous inbound "lines" if you want, you just pay per-minute charge X number of simultaneous inbound "lines". This functionality is per-provider, they all have different policies. >Oh, great. ;-) For next time, does anyone have recommendations for a >particular motherboard or a particular type of motherboard? Yup. Intel chip. I know people will say I'm trolling, but I wouldn't use an AMD for an Asterisk box. Workstation, yes. I run one myself, they work fine. Note I said Chip not Chipset. I've had good luck with older Intel chipsets but from what I understand some other guys have had problems with the new Intel chipsets. I totally trust Asus though. What I would do were I you is Asus with no onboard stuff as much as possible with a P4 chip. >This is for an office -- I figured that running hardware RAID would be >the most likely to avoid downtime if a hard drive failed. How do most >people handle this? You don't. You do not use hardware RAID on an Asterisk box, and if you do, you try it beforehand. Interrupts. All about the interrupts. As a point of reference, our extremely expensive Mitel ICP 3300 has a 20 gig Maxtor in it. That's all. Same hard drive as a friggin Xbox! Strategies for recovering that I would look at if you want to try sans RAID is to do your Asterisk config, then ghost it onto a second HD, then install the 2nd hard drive as a warm spare with an IDE drive carrier (this is a tray that is removable like a SCSI hot swap, except not hot) and every night you cron a 'dd' command to dupe the primary HDD onto the spare (dd application note: make damn sure that the spare drive is the exact same model. dd will not compensate if your drive has different geometry). Primary fails, turn off the box, yank out the primary, move the spare to the primary, turn on box. Worst case scenario is that you loose X hours of voice mails and call logging where X is the number of hours since the last DD command. One last comment: TDM04 cards have to be reinitialized periodically otherwise they sound like crap. And by reinitialized, I mean warm boot. Maybe Digium fixed this, but mine, I reboot the boxes every night. No more problems. hth, keep the faith. Once it working good you will never go back and laugh at everything else. _______________________________________________ Asterisk-Users mailing list Asterisk-Users@lists.digium.com http://lists.digium.com/mailman/listinfo/asterisk-users To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users