Many thanks for the many people who replied. Bottom line is that you can
not -HUP inetd for a change to the telnet daemon if you are telnetted into
the machine. You have to have no telnet connections active in order for
it to work. After killing all shells except for console, it still
wouldn't work because
*.telnet 9919 BOUND
kept showing via "netstat -a | grep telnet". Arthur Darren Dunham hit the
nail on the head so I've included his respond. Bottom line, I had to
reboot to get tcpd and in.telnetd working correctly. :-/
-- James
P.S. One wonders why the manpage for netstat (Solaris 2.6) doesn't
mention the parameter "BOUND" at all........hmmm....
---------- Forwarded message ----------
Date: Tue, 6 Apr 1999 14:47:36 -0700 (PDT)
From: Arthur Darren Dunham <add@netcom.com>
To: James Ford <jford@tusc.net>
Subject: Re: in.telnetd broke
>
> I have just installed tcp_wrappers (v7.6) on my box. I have successfully
> configured imapd to deny some addresses, allow others, etc. I have now
> attempted to do the same for "in.telnetd" but now can't telnet into the
> box.
>
> (Note: I am still telneted into the machine in question....new telnet
> connections fail.)
>
> Currently /etc/hosts.allow and /etc/hosts.deny do not have an entry for
> in.telnetd....meaning (I assume) it's not a tcp_wrapper problem. The
> error message I'm receiving is:
>
> Apr 6 12:51:57 ns1 inetd[4998]: telnet/tcp: bind: Address already in use
>
> Below are the lines from the /etc/inetd.conf file. Note: in testing two
> of the "telnet" lines are commented out thus making only one line active.
Make sure you restart inetd with 'kill -HUP <pid>'. I've had lots of
problems with Solaris when I actually 'kill' and then restart inetd. It
sometimes doesn't restart (and even when it does, it enherits your
environemnt which can cause lots of problems).
<***** two key paragraphs --James>
If you run 'netstat -a | grep telnet' you'll probably find something
bound to the incoming telnet port. I used to get some that were in the
stat 'BOUND'. The netstat man page doesn't even mention that as a
possible stat.
When this happened, we always had to reboot the machine. If that's too
drastic, you *may* try doing an ifconfig down/up on the interface. I
can't remember if we did that and it worked.
<***** /two key paragraphs --James>
I used to be in the habit of killing inetd from SunOS 4.x experience
where the HUP would not always remove services. Solaris 2.x seems to be
fine in that aspect. I've stopped killing it for service changes.
-- Darren Dunham Unix System Administrator Got some Dr. Pepper?
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:13:17 CDT