Incomplete/failed faxes

[ThurmanB]

Hi,

So I've been working on this problem for several months to no avail. My system is on 5.2.1.

We have 15 sites (all in the same area code, on the same CM) and all of them are experiencing fax problems at every site. The problems are intermittent, but from what I can tell, they have the highest chance of failing when the other party is using a fax server like Ritefax. Both sending and receiving, partial pages and incomplete faxes are occurring, along with Comm. Errors. These usually go over a PRI, though I do have analog trunks that I've sent the faxes over to troubleshoot with the same result. If I plug directly in to the analog modem and bypass our system entirely, I have no problems. That leads me to believe that the problem is in fact on my side. They were having fax issues before I took this position, so I have no data concerning changes made before I was here.

Here are some things I've tried that didn't seem to change much:
Turned off any and all ECM
Tried Setting ip-codec-set p.1 "Audio Codec" to G.711MU, G.711A, G. 729
Tried setting ip-codec-set p.2 "Mode" to pass-through, relay, and t.38
Called PRI provider and asked what is recommended Mode/Codec, and set accordingly
Discovered Primary Sync Source was defunct, and set a new sync source (I'm wondering If I need to dig into this further)
Tried multiple brands/types of fax machines
Changed Transmission Speed of Fax machines
Changed XOIP endpoint type in station to AUTO (This was a wild guess)
And a few other things I'm sure I'm forgetting. . .

Please, if you have any knowledge to impart, don't feel like you're imposing. I'm fairly new to phone systems in general.

Thanks,
Thurman

 

[kyle555]

http://www.phonet.cz/archiv/dok_cizi/ef-lb2626.pdf

So, you're saying you've got 15 gateways with analog cards +fax machines and incoming faxes, usually from a fax server, on PRI, have problems?
All PRIs - if you have one in say slot 1v3, list measurements ds1 log 1v3 - make sure you got all 0's. Those are ISDN errors/slips. If you're on G650 cabinets, you set the sync source in CM. G350/450/430/etc you set it in the gateway CLI - show sync timing.

Now, the IP stuff is inconsequential when the fax is in the same gateway as the trunk its going in/out of. If you're central trunking thru site 1 for example and all the other gateways are going IP to that one, then yes, you'd need to consider the inter-region connectivity. If you've got slips on a PRI in a gateway with a fax in that same gateway having problems, you're hooped.

Few main points: t38 maxes at 9600bps and it a standard supported by anyone. So, the DSP hears fax tones and the analog transmission exists between the fax and the gateway and between the other gateway and the trunk/other fax. Gateway-Gateway communication is t38 and not as sensitive to jitter/packetloss. Pass through and relay are Avaya proprietary and only supported between their DSPs. Passthru goes up to 33.6kbps, relay is older than the t38 standard, but the same idea. Passthrough is faster and uses less DSPs and is more susceptible to goofups, relay is slower, steadier and uses more DSPs

If you've got no non-Avaya IP fax destinations - like your own SIP fax server or carrier IP trunks, you don't need to front-end those endpoints with t38 either - you can just have relay between all your regions to be as steady as possible. You can also do a "list trace station" on a fax making a call and see what the jitter/pktloss/negotiated fax IP codec is to get an idea if anything's funky.

My gut says PRI slips to be honest. Something about a beefy Dialogic card or VOIP stack on a fax server not playing nicely in sub-optimal conditions. I suppose a mildly comparable analogy is Avaya voicemail in its first release of being virtualized and direct IP. If I let phones from 15 sites over 15 WANs doing things to it direct, it wouldn't nicely juggle 15 audio streams with different nitty gritty jitter/packet loss stuff. Frontending that with a gateway and forcing all phones to go through 1 gateway and that 1 gateway to send predictable audio streams to the voicemail got rid of voicemail answering with "...the mailbox of" and made it answer "you've reached the mailbox of..." and not cut off audio/record choppy messages/etc. If that fax server is busy - and why wouldn't it be - answering 20 calls and yours is 21st with slips, you're the first to get the cold shoulder.

 

[ThurmanB]

Hi Kyle,

Thanks for the quick response!

Yes, I've got 15 gateways with analog cards and fax machines, and incoming faxes from fax servers on PRI have problems. The other parties' fax servers also have similar issues (Partial/incomplete) when they're sent from one of my machines.

I have G450s and G430s at every site except for the main site, which has two G650 cabinets (A PRI DS1 card in each G650).
After running " list measurements ds1 log XXXX " on the PRIs, one PRI returns all zeroes, the other has ones in a couple of places.

Unfortunately, I do not have the root password to get in to the gateways, so I will have to go on site to each gateway to change the password, then change the sync sources. I am sure they are set to the now-decomissioned source. I was unaware that most of my gateways have to have their sync-source set in the gateway CLI, so that's good to know.

I'll see if I can get this list trace soon.

Thanks again.

 

[kyle555]

OK, 1 errored second in a 15 minute interval isn't bad. And, not having it in most 15 minute intervals is great. You might be syncd right and 1 slip in a 6 hour period isn't your problem - you'd have complained that all faxes were good except this 1 between 10:12 and 10:27.

Are you PRI/SIP through your carrier? Do you know if they backhaul your PRIs over IP? Maybe a carrier in the middle is cheap. You know, you can dial 10-10-XXX of the pic code of another LD carrier. It'll cost a bit to make the call, but it certainly would point to a carrier problem between you two if Verizon LD is fine, Sprint isn't, etc. Are you faxing to a toll free? Maybe find their termination number and try that if its the case. You're probably not the problem here.

 

[jimbojimbo]

FAX over IP is a pain. Especially if you are on CM5.2.1 and you have G650 cabinets. The Medpro Boards do not support T.38 with Error Correction Mode. The only way to really get FAX over IP to work correctly/consistently is to upgrade to 7.x, install a G450 at the core and move your DS1 boards into it, and make sure you have MP160/MP120's in the gateways. You can then support V.150 and T.38 with ECM (introduced in 6.3).

Alternatively, go with pass-through and manually configure your FAX machines to negotiate a max through-put of 9600. SuperG at higher speeds will cause you issues.

 

[rejackson]

This solved my fax problems when t38 tries to meet with G3. On the codec use g711 ,enable t38 and disable modem. If modem is enabled IP fax using t38 has problems falling back to G3.