I recently replaced a site that was using 1.4.[mumble] with hylafax/iaxmodem. They have an RBS T1 and were using about half of their 50 DID numbers for “fax to email”. This all broke with the new system 🙁
The original chan_dahdi.conf had no mention of “faxdetect”, so I assume it was operating with whatever is the default. Off?
The new box originally had “faxdetect=no”, and I found that all my test faxes failed with negotiation errors. When I finally tried
“faxdetect=incoming” test faxes from another machine running hylafax went through fine, and I thought I was done.
The following week the customer reported that inbound faxes weren’t working. When I looked at the log, I saw lots of these:
chan_dahdi.c: — Redirecting DAHDI/24-1 to fax extension
Which in my FreePBX setup eventually goes to a “no service” message and hangs up. I’ve never defined a “fax” extension and don’t really know what that is about. Turns out that any fax machine that calls ends up following this path. If my other hylafax server calls, it follows the normal path and gets answered by my pool of iaxmodems. I don’t really understand the difference between the two types of calls, first of all.
So it seems from this experience and a recent thread on -users that enabling faxdetection in chan_dahdi sets up some additional buffering that at least in my case, in 1.8, seems to be required (without it all inbound faxes fail from my hylafax server with negotiation problems). Unfortunately for me, this also seems to bypass normal DID handling and sends calls to an undefined “fax extension”.
Can anyone shed some light?