[plug] configuring talk

Greg Gamble gregg at maths.uwa.edu.au
Fri Jul 16 18:04:25 WST 1999


Hi,

  I've been wrestling with talk ... trying to make it work. What am I
(not doing)/(doing wrong)? I can't seem to get rid of the following
diagnostic that appears when one tries to execute talk:

  [Couldn't bind to control socket : Cannot assign requested address (99).
   Press any key...]

We have a Debian 2.1 Linux 2.2.9 system ... 

First I ran

  update-inetd --enable talk

which updated /etc/inetd.conf to read:

talk  dgram   udp    wait    nobody.tty  /usr/sbin/tcpd /usr/sbin/in.talkd

At first, I thought the user.group entry: nobody.tty might be the
problem ... finger works with user nobody (so that should be ok). 
I added all the users into the tty group ... still didn't work.
I removed .tty so that the entry looked like finger's ... still didn't work.

Yes ... I did run:

  /etc/init.d/netbase restart

to restart the inetd daemon each time I modified the inetd.conf file.

I feel the diagnostic is referriing to the second parameter on the 
inetd.conf line: dgram   ... i.e. my guess is that something about
the dgram socket is wrong ... how do I fix it? or diagnose further?

On another machine (with an old Slackware Linux) we did things 
a little differently, and it works there. AFAI can see the difference 
is that we have commented out the `talk' line in /etc/inetd.conf
and enabled talk via /etc/hosts.allow (adding a line starting with
`in.talkd: ' followed by a list of IP nos. I tried that also on
the Debian machine ... same message when talk is run and whinges about
other services not starting correctly when the daemon was restarted.

Any clues? Perhaps I've ~RTRFM (= not read the right fine manual).

  Regards,
  Greg Gamble <gregg at maths.uwa.edu.au>


More information about the plug mailing list