HylaFAX The world's
most advanced open source fax server
|
|
[
Date Prev][
Date Next][
Thread Prev][
Thread Next]
[
Date Index]
[
Thread Index]
Re: FW: Will 4.1b3 or 4.1 solve "no response to MPS"?
At 02:20 PM 10/31/99 +0100, Giulio Orsero wrote:
>On Sun, 31 Oct 1999 16:24:53 +1100, hai scritto:
>
> >stty < /dev/<modem device file>
>$ stty < /dev/ttyS2
>speed 38400 baud; line = 0;
>min = 127; time = 1;
>-brkint -icrnl -imaxbel
>-opost -onlcr
>-isig -icanon -iexten -echo -echoe -echok -echoctl -echoke
Try that again with the '-a' param ie stty -a < /dev/ttyS2, you need to
make sure ixon, ixany ixoff are set correctly.
Are you running faxgetty on the line? The speed is also set quite high for
an xon/xoff connection it might be worth swapping to hardware flow control.
> >Compare to the hylafax modem configuration file, inspect also the log files
> >of the failures.
>Hylafax force xonxoff
>
>ModemType: Class1 # use this to supply a hint
>ModemRate: 38400
>ModemATCmdDelay: 300
>ModemFlowControl: xonxoff # XON/XOFF flow control assumed
>#ModemFlowControl: rtscts # XON/XOFF flow control assumed
....
>My rate is about 5%.
>But you have to consider that you live in Australia, I live in Italy; we
>use to keep old equipment until it dies...
That is too high. There is a configuration or hardware problem, the number
of machines that actually ignore the last EOP is minuscule, look again at
the figures i gave yesterday:
Modem: Combined Total
Total faxes sent: 82173
Error No Rate/1000
------------------------------------------------------------------------------
No response to MPS * 47 0.6
No response to EOP * 333 4.1
71% percent of the documents sent are 1 page(based on successful sends from
xferlog).
A MPS failure implies the receiving fax does not have control of the
session for whatever reason - whether hylafax waits or ignores the EOP
reply matters not one bit.
Therefore 29% of uncontrolled sessions will result in a MPS failure, 71% of
uncontrolled sessions will result in a EOP failure.
=> roughly 115 of the EOP failures are due to uncontrolled sessions.
=> only 218 out of 82 173 or less than 3 of every 1000 fax
machines ignore the EOP reply.
That figure will only go down as that old equipment is replaced, the
uncontrolled session problem will *not*.
> >ie the errorstats ouput of a server i operate:
>How do I get this output?
http://www.trump.net.au/~rjc/hylafax/index.php
> >Show a log! Preferably of both hylafax failing and other software
> >succeeding with the same hardware setup.
>Well, Il try to see if MSFax has logging capabilities.
>For efax see:
>http://www.hylafax.org/archive/1999-05/msg00143.php
That case looks very much like misconfigured flow control - look at efax it
sent the EOP and got a MCF in reply, the receiving fax is behaving correctly.
By comparison look at the hylafax log there's quite large pause after end
of page is sent and telltale ^S^Q characters in the log. Just afterwards
then after that a bunch of weird empty lines as the modem and computer have
trouble communicating with each other.
>In the list archive there are examples of class2.0 modems.
>
>This is Hylafax, It has the ^Q^S you already told someone about...
>==========
> Oct 19 12:03:10.36: [19840]: --> [2:^S^Q]
...
> Oct 19 12:03:20.06: [19840]: --> [0:]
...
> Oct 19 12:03:24.89: [19840]: --> [0:]
...
The log is very similar to the previous example, flow control characters
and empty lines suggest problems communicating with the modem.
> >...but would you want that kind of software?
>Yes, I'd like the option to tell hylafax to consider such a fax a
>success.
>
>If a customer receives a bad quality fax , he calls me and he asks me to
>resend it. If it receives, many times, 3 or more copies of a fax or of
>the 1st page of a fax, he calls me and he is angry with me.
That will not work - if the receiver only received junk from the top of the
page they will have no idea who sent it(some fax machines print a report
page on failed receive but most don't). Meanwhile you will think it was a
success....i hope you don't ever try to fax anything important.
- Robert