Re: [hylafax-users] Timeouts depending on disk drive!

["Uwe Dippel" <udippel@xxxxxxxxx>]

So far I lacked the time to start compiling HylaFAX+ 5.2.7 on OpenBSD.
I will do it as soon as I get the time.

In the meantime, I have created a complete testcase, and can by now
time out the modem by reducing the speed of the storage. This will
reproducibly make T.30 T2 timeouts increase.
Currently, I am running one testcase on Geode 500 MHz, with 512 MB of
RAM, and essentially nothing else on that machine. Firstly, there is
no difference between serial and USB modems.
With USB hard disk, everything runs excellent. With USB flash, we have
some trouble, but with a 'renice' to -10 it gets better. The slowest
storage is CF, as internal HDD. In all cases the timeouts happen after
some pages (4-5) of the same document. It goes through, though, even
on CF, when I renice the faxgetty to -20.

Overall, the system loads at receiving a fax with this low-power system are:
1-2% user 10% system 1-2% IRQs 85% idle.
At the same time, the CPU usage is essentially the same as the
faxgetty process: 10%. So it is not a lack of general system resources
that we see.

Also, I use a purely internal phone, so noise is essentially no
problem. I as well looked at the transmitted partial pages (before it
breaks off), and there is no noisiness. Pages simply break off and are
received as partial pages (like the log that I sent).

Of course, using high quality modems that don't offload processing to
the host will fare better in a case like the underlying one, when a
condition of postponement or a race condition seems to occur at some
delay at accessing the persistent storage medium.

Just as an intermediate information on the progress,

Uwe


____________________ HylaFAX(tm) Users Mailing List _______________________
  To subscribe/unsubscribe, click http://lists.hylafax.org/cgi-bin/lsg2.cgi
 On UNIX: mail -s unsubscribe hylafax-users-request@xxxxxxxxxxx < /dev/null
  *To learn about commercial HylaFAX(tm) support, mail sales@xxxxxxxxx*