Re: [cobalt-users] tried Postgres -- screwed up SSH (need help)

[josh <josh@xxxxxxxxxxxxxxxxxx>]

On Thu, May 16, 2002 at 04:55:21AM -0400, Gerald Waugh wrote:
> On Tuesday 14 May 2002 12:06 pm, Tim Dunn wrote:
> >
> > Disagree.  Most sshd installations I've seen/supported are started out
> > of /etc/[rc.d]/init.d
> >
> 
> I agree!, I meant to say it runs under tcp wrappers, and you can use 
> /etc/hosts.deny and allow
> The original poster was locked out of his system from fooling with those 
> files.

The main gripe I have is that one common way of securing machines is
to stop inetd and tcp wrappers. Then to run only independent
secure daemons. While I'm happy the feature is here, I still think it odd.

As the original poster I should mention that the problem was that
traditionally tcp wrappers is used to control processes started by
inetd. I think I might still be missing some configuration issue. Now
that I've added the line tcpwrappers still complains about it.

[josh josh]$ /usr/sbin/tcpdchk
warning: /etc/hosts.allow, line 8: sshd: no such process name in /etc/inetd.conf


It works now. (Actually, most of what I'm doing is working now -- but
I wouldn't want to dramatically change to tone of the list by posting
success stories.)

It would also have been nice if I had seen the posts
that RaQs use a 115Kbs serial connection by default. I was able to
connect a vt510 to my RaQ, but I've never been able to watch the start
up messages - because I didn't know the right speed. It will autobaud
and eventually match. (Yes I could check /etc/inittab -- but only
after it was running again. Same way I could see the startup messages
if I took it down.) 

BTW, for whatever reason, even with the answerback
message set to vt510 on the vt510, and a vt510 entry in /etc/termcap
the $TERM variable is always undefined which upsets commands like vi
and man.

-- 
Josh Kuperman                       
josh@xxxxxxxxxxxxxxxxxx