Re: [hylafax-users] Serial latency

["Jay R. Ashworth" <jra@baylink.com>]

On Wed, Jun 27, 2001 at 09:56:23AM +0100, Tim Allen wrote:
> I'm running Hylafax 4.1rc2 under RH 7.1 on a 486 100MHz (16550AF UARTS) with
> a Multitech MT1932ZDXK Class 2 modem, RTSCTS handshaking (cabling
> double-checked) and 38400 Baud between PC and modem.

ATTENTION EVERYBODY:  See that?  That's how a problem report *should*
start.  Thanks, Tim.  :-)

> Clearly the machine isn't exactly cutting edge and once the machine comes
> under load when receiving a fax I start losing chunks of the incoming page.
> Can anyone advise on the likely success of any of the following ideas:
> 
> Drop the modem comms speed to 19200 or 9600 Baud.
> 
> Run setserial at startup with the low_latency option.
> 
> Modify the UART fifo interrupt threshold. At some point during the
> installation I recall a message stating this was set to 8 but I don't know
> what to do to change this.

If you're running in class 2 and dropping incoming data, the problem is
almost certainly interrupt latency, and if your 550 is configured
correctly, I'm shocked; *shocked*, I tell you.  :-)

I would recommend putting a break out box on the cable, and msking sure
that the handshake lines are actually toggling, and making sure that
the modem is actually configured for that (&K?).

Yeah, I would try backing the modem rate off to 19.2... and I'd also
look into increasing memory, if you don't have at *least* 32MB; 64M
would be better.  Swapping will *kill everything* on a Unix box; if
you're swapping, you don't have enough RAM.

RAM's cheap.  ;-)

Cheers,
-- jra
-- 
Jay R. Ashworth                                                jra@baylink.com
Member of the Technical Staff     Baylink
The Suncoast Freenet         The Things I Think
Tampa Bay, Florida        http://baylink.pitas.com             +1 727 804 5015

   OS X: Because making Unix user-friendly was easier than debugging Windows


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