Re: flexfax: faxsetup on Linux (bug in moduser?)

[Matthias.Apitz@SOFTCON.de]

Nico Garcia wrote:
    
    On Sun, 9 Nov 1997 Matthias.Apitz@SOFTCON.de wrote:
    
    > I'm checking the faxsetup(1M) on Linux (to detect any problems
    > and see if it works reliable). For some unknown reason my Linux
    > box (S.u.S.E.) is shipped with a user "fax" in /etc/passwd and
    > HylaFAX's faxsetup(1M) wants to modify this ("fax" must have
    > the same uid as "uucp"). But it seems that there is a bug
    > in the usermod(1M) command. It complains about the existing
    > (new) HOME of "fax" with exit code 12 but the change is done
    > in the /etc/passwd file. A workaround could be running the command
    > twice 'cuze it exits with 0 for the 2nd change which has
    > nothing to do anymore. Any comments from the Linux gurus?
    
    I think the problem is slightly different flavors of Linux have
    slightly different flavors of the adduser or usermod or whatever
    command. I think this is a completely reasonable workaround.  Some use
    passmgmt, some use useradd, some use adduser, etc.

The HylaFAX software can't work corretly if the user "fax"
in the passwd(4) has a different uid than "uucp". If there
is such wrong entry the faxsetup(1M) script should be able to
fix this. I'll let this to the Linux folks to clear what's to
do to have a working usermod(1M) command in Linux.
For the moment I'll call it twice in faxsetup.linux.
    
    > BTW: The usermod(1M) wants the flag "-m" after the "-d home"
    > while the original faxsetup script in pl1 has the "-m" as the
    > first parameter. This can easy be fixed in the script faxsetup.linux
    > (or should be fixed in faxsetup at all).
    
    faxsetup.linux would be the place to do it, I think.

Not sure. The SVR4 usermod(1M) also wants the "-m" flag after
the "-d home".

    > All other things in faxsetup(1M) work without any kind of problems.
    > The faxq(1M) is up and the hfaxd(1M) starts on connects via
    > the inetd(1M):
    
    I thought most of us were using the boot-time script to start faxq and
    hfaxd as background daemons. Is the inetd.conf technique now
    considered reliable enough for general use on low-volume sites?

Yes, that's the message of my posting. Running hfaxd(1M) from
inetd.conf(4) should work. If it does not, someone should fix
it on the OS where it causes problems. The line in Linux's
inetd.conf(4) (added by faxsetup(1M)) is:

hylafax stream  tcp     nowait  fax     /usr/local/lib/HylaFAX/sbin/hfaxd      hfaxd -I

and faxsetup(1M) sent also SIG_HUP to the inetd(1M) to let it
re-read the file inetd.conf(1M). It's working fine and there is no
reason not to use faxsetup(1M) on Linux.

On my production system I'm running hfaxd(1M) from a boot-time
script in /etc/rc2.d/  Both techniques should work.

	matthias