rlogin stop working

[The_Cina]

Hello,

Today on a HP-UX 11iv3 the rlogin daemon suddenly stopped working.
I had to restart the inetd daemon make it work again.

This server is a ServiceGuard cluster node.

No modification on the config files has been done and on the syslog no messages appeared. I'm trying to figure out what it could be the root cause.. any suggestion?

Thanks in advance

[Tim Nelson]

Did you review the /var/adm/syslog/syslog.log for any error messages ?

[The_Cina]

Thanks Nelson,

I did reviewed the syslog, but i didn't see any error. Service Guard kept querying the the host, and at a certain point there were no more connections entries on the syslog.

Then, I tested manually and I have notice the rlogin was not working properly, and I had to restart the inetd to make it working again...

[TwoProc]

That's actually quite common. And funny in a wry sort of way.

I've seen that problem with inetd, no kidding, since HPUX 8.x! It was much more common back then for HPUX 8.x and 9.x. But, I've seen it from time to time on HPUX 10 and 11.0 and 11.1.

2 things to remember
1) if youv'e got a console hooked up, that won't be affected because that is run by getty from inittab.
2) Way back when I used to see it often, I would just keep a login open on my computer that I could "su" to root as needed. That way, I had my connection regardless of inetd being up or down, and I could just restart it.

Of course, nowadays I see the problem extremely rarely, in which years and years go by without seeing it again.

[The_Cina]

I feel better now that I share the problem with someone else... but.. what's the bet solution..?

We have a critical application which use rlogin.

I would propose to switch to ssh, having the feeling that it would be more stable.

But.. is there any way to know why the service crashed? I have found no logged information so far.

Thank you,

[TwoProc]

Well, log a call with HP on it. It's probably a known issue. If it's that critical for you, it's pretty simple to write a cron job that checks every 5 minutes or so to see if it's up, and if it's not, then bring it back up.

Re: rlogin. Yeah, you need to get away from that. I've not had any trouble what so ever on sshd. It runs independent of inetd. But mainly, from a security standpoint - you need to quit using rlogin and telnet. That is, don't use the unencrypted login programs any longer.

[TwoProc]

You can start inetd with logging:

inetd -l

You'd have to make in the change in the init startup file.

If you want to just do it now from the command line:

$> inetd -l -c

The above will force inetd to reread the inetd config files, and start everything up with logging.

-- short term solution - you *could* just create a cron job to restart inetd every 5 minutes or so, by the running the command above. This would restart it if already running (won't affect anything already running), and if it is dead already - then it will just start it up.

Maybe turning on logging will tell you more.