Re: [hylafax-users] DTE to DCE Data Underflow

[Dmitry Bely <dbely@mail.ru>]

Andreas Grimmel <grimmel@additive-net.de> writes:

> I set up a HylaFAX Server quite successful some time ago, and it seemed
> to work REALLY fine. Then I went in vacation and when I came back, I got
> lots of people queueing in front of my office door, all of them telling
> me the same thing: Some Faxes just can't be delivered, and in the logs
> it says everytime the same thing: "DTE to DCE Data Underflow" etc etc
> etc. This does not happen all of the time, most faxes are delivered very
> well. For me it seems this is a problem with older Fax machines, I tried
> this in our office with one 5-year-old-machine and one 6-months-old-one.
> With the old one faxing crashes everytime, with the new one never.
> I'm using a SUSE 6.3 Box with a 2.2.14 Kernel running on a
> PentiumIII/550 SCSI Machine with 256 Megs of RAM, an Elsa Microlink 56K
> Pro Modem in Class 2.0 Mode and HylaFAX 4.1beta2.

Strange enough, but your logs was collected in Class 2 mode (not Class
2.0), and there was no "DTE to DCE Data Underflow" ...

> Could that be that the older Fax machines don't understand my Modem's
> talking in Class 2.0 mode ?

If the firmware is buggy, that's possible. Try Class1 -- if we find a fax
protocol-related bug in Class1 driver, we can understand what really
happens and fix the bug ourselves. I consider this as a significant
advantage :-)

> I don't have too much knowledge about fax
> protocols and phone lines and that stuff, so please excuse me if that is
> a totally dumb question where reactions normally are "damn can't you see
> this has REALLY NOTHING to do with each other???" ;-)

Jay has already explained that Class1, Class2, Class2.0 have nothing to do
with the fax protocol itself (and so fax machines compatibility). I can't
do this better anyway :-)

> I would be very happy if you could help me outta this, because ignoring
> the above problem it's just wonderful seeing the HylaFAX server work (we
> had an NT Machine before, and it crashed twice a week, hehe).

Now about your logs. I'd bet that these "Unspecified Transmit Phase B
error" were caused by incorrect Postscript, supplied to the server
(e.g. your Windows client used HP printer driver, which appends PJL
commands to the Postscript file). Under such circumstances hfaxd don't
identify the data as a Postscript, fails to invoke Ghostscript rasterizer,
but anyway queue the empty job (if my preliminary analysis was
right). faxsend realizes this too late, and terminates connection,
producing that silly "Unspecified Transmit Phase B error". I know about
this bug for about a month, but still going to fix it :-))

> Thanks to the HylaFAX devel team for just a GREAT piece of work.

Hope to hear from you soon,
Dmitry




____________________ HylaFAX(tm) Users Mailing List _______________________
 To unsub: mail -s unsubscribe hylafax-users-request@hylafax.org < /dev/null