flexfax: "No response to MPS or EOP" - AN answer not THE answer

[Neal Cartwright <neal@evesham.demon.co.uk>]

A contribution for anyone sweating over the ever-popular "No response to
MPS or EOP" problem. This is the fix to my problem, it _may_ help you.

The symptoms were:
Multiple but often incomplete page delivery when the images submitted
for sending were whole, depending on fax sent & destination machine.
Email reports "requeued" or "failed" with "No response to MPS or EOP..."
& a log trace.

The system was: 
RedHat5.2/i386, hylafax-4.0pl1-2, Hayes Accura 56K (class 1) modem - but
I suspect none of that matters.

First point is that to the best of my understanding the "No response to
MPS or EOP" message just means "they didn't say they received it" and
could have any number of causes. The cause in my case was flow
control...

The chase:
The clue was a line in the failure message log trace - marked "clue
here" below 
...
May 31 19:46:48.50: [19655]: REMOTE best rate 9600 bit/s
May 31 19:46:48.50: [19655]: REMOTE max page width 1728 pixels in 215 mm
May 31 19:46:48.50: [19655]: REMOTE max unlimited page length 
May 31 19:46:48.50: [19655]: REMOTE best vres 7.7 line/mm
May 31 19:46:48.50: [19655]: REMOTE best format 2-D MMR
May 31 19:46:48.50: [19655]: REMOTE supports T.30 Annex A, ECM
May 31 19:46:48.50: [19655]: REMOTE best 20 ms, 10 ms/scanline
May 31 19:46:48.50: [19655]: USE 9600 bit/s
May 31 19:46:48.50: [19655]: USE 20 ms, 10 ms/scanline
May 31 19:46:48.50: [19655]: SEND file "docq/doc6.ps;40"
May 31 19:46:48.50: [19655]: USE page width 1728 pixels in 215 mm
May 31 19:46:48.50: [19655]: USE unlimited page length 
May 31 19:46:48.50: [19655]: USE 3.85 line/mm
May 31 19:46:48.50: [19655]: USE 2-D MR
May 31 19:46:48.50: [19655]: SEND training at v.29 9600 bit/s
May 31 19:46:48.50: [19655]: <-- [9:AT+FTH=3\r]
May 31 19:46:48.55: [19655]: --> [7:CONNECT]
May 31 19:46:50.33: [19655]: --> [7:CONNECT]
May 31 19:46:50.77: [19655]: --> [2:OK]
May 31 19:46:50.77: [19655]: DELAY 75 ms
May 31 19:46:50.86: [19655]: <-- [10:AT+FTM=96\r]
May 31 19:46:50.92: [19655]: --> [7:CONNECT]
May 31 19:46:52.87: [19655]: --> [2:]
May 31 19:46:52.87: [19655]: --> [2:OK]
May 31 19:46:52.87: [19655]: <-- [9:AT+FRH=3\r]
May 31 19:46:53.71: [19655]: --> [7:CONNECT]
May 31 19:46:54.59: [19655]: --> [2:OK]
May 31 19:46:54.59: [19655]: TRAINING succeeded
May 31 19:46:54.59: [19655]: <-- [10:AT+FTM=96\r]
May 31 19:46:54.65: [19655]: --> [7:CONNECT]
May 31 19:46:54.65: [19655]: SEND begin page
May 31 19:47:01.64: [19655]: SENT 27402 bytes of data
May 31 19:47:05.63: [19655]: SENT 16382 bytes of data
May 31 19:47:05.63: [19655]: SEND 2D RTC
May 31 19:47:05.63: [19655]: SEND end page
May 31 19:47:16.38: [19655]: --> [2:]    <<<<<<<======================
CLUE HERE
May 31 19:47:16.38: [19655]: --> [2:OK]
May 31 19:47:16.38: [19655]: DELAY 95 ms
May 31 19:47:16.49: [19655]: SEND send EOP (no more pages or documents)
...

The clue was that 10 or 11 second gap followed by --> [2:] or -->
[2:^S^Q]
This is Xon-Xoff modem flow control not working 
(Credit to Robert Colquhoun <rjc@trump.net.au> 1st Nov 99)

My /var/spool/fax/etc/config.modem had the line 
ModemFlowControl:      xonxoff         # XON/XOFF flow control assumed
but "stty -a < /dev/modem" 
returned (amongst other settings) -ixon (i.e. no Xon-Xoff on this serial
port)

My fix (others possible:
Turn on hardware handshaking (i.e. match hylafax settings to modem
settings)
Comment out the existing ModemFlowControl line in config.modem, add
instead
"ModemFlowControl:       rtscts          # RTS/CTS flow control"

Neal