UNIX for Advanced & Expert Users

I am working on a Unixware 7.1.1 system that seems to have suddenly started having huge telnet login slowdowns.

I am theorizing it might be due to processor load an I wanted to know what INIT should be using on average in terms of processor power. For me it is at least at 15%.

I also want to know what to do if init IS using too much cpu power.

Thanks in Advance,

1. Check that reverse DNS lookups are working first, especially the IPs of the telnet clients.

2. Why are you still using telnet?

OpenServer versions of SCO often suffer from this problem. It is usually caused by "ghost" telnet sessions. That is users have cancelled a telnet session without logging off, (by clicking on the X in the top right corner of the screen.)
You can see that you will get a different list of users from the 'who' command than from 'ps -leaf |grep ttyp'.
The fix is to kill the processes that do not have an active terminal associated with it.

-I checked the reverse DNS checks: Actually when I was first looking at the machine I noticed resolv.conf was renamed to something else. I decided to replace it with resolv.conf and I wonder if that may have caused the machine to start doing DNS lookups (and failing somehow).

-I checked for ghost sessions and didn't find anything.

-I am using telnet because that is what the server is configured on. It is really a legacy setup with unixware 7.1.1. I am not mandated to touch anything other than create a failover replica for this system.

-I also get an error in the log files (happens continuously) like "/var/adm/utmp being rewritten by /var/adm/utmpx due to size mismatch". This is a new error message which happened close to this new problem. I am tempted to erase the files and "touch" new ones but I am worried it will prevent people from logginin.

-I was thinking about modifying inetd to have telnet NOT go through tcp wrappers.

Thanks Again!

How can I actually find out WHAT init is doing strance, truss or something?

init problems are usually caused by a bad /etc/inittab. So I would check that file first.

I would now try pinging the telnet client, and/or run traceroute back to the client. I think that the host now has no way to find the client.
What does the resolv.conf file look like?
Is ndc running on the server?
Are the clients on the same subnet?