Re: flexfax: Hylafax sending hung

[David Woolley <david@djwhome.demon.co.uk>]

> 
> I am just beginning to use Hylafax as an outgoing-only fax server. My
> installation is on SCO 5 using Hylafax version hylafax-v4.0pl2 compiled
> under gcc.

Oops.  I'm so used to people saying they are using Red Hat Linux that
I only gave the SCO answer as an aside!  For SCO it is easy to get
fairly detailed information on the state of the hung process.  I think
the ps on 3.2v5.0.x of SCO Unix does interpret wchan and SCO wchan is
the conventional wchan, which is the address of a data block representing
the resource being waited for.

You should also use crash.  Do 'p ! grep fax' to find the process table
slot for the hung process, then do t on that process to get the kernel
stack traceback, which will show you exactly where it was waiting.  Also
use tty to get the status of the serial port, assuming a standard one.
If you don't have an interpreted wchan, you can use ts or ds to interpret
it (there is another command to find exactly which entry in a kernel table
is involved).  If you can catch it before it hangs, you might also want
to attach the trace command to it.

NB when you raise this with SCO or on comp.unix.sco.misc, remember that
there is no such thing as SCO 5, and even more so, there is no such thing
as SCO 3.  You need to specify the full version (the latest is 
SCO Unix 3.2v5.0.5) and identify all service packs (Support Level Supplements
in SCO terms) and third party drivers that have been added.  (The three
in what is popularly called SCO 3 is the 3 in 3.2, so what is popularly
called SCO 5 is really SCO 3 by this definition!  All are based on 
AT&T System V R 3.)