Whats the correct way to restart inetd

[amreek bansal]

Hi,

Can anyone help.. I am getting a error when telneting to my HP-UX server, (I'm trying to telnet to port 3872 on HP-UX server from aremote server, I get connection refusted. I'ce checked firewall, The rules are in palce and the port 3872 is enabled, it doesnt look like its firewall. I've checked /etc/inetd.conf and found the line with telnet commented out. So I have un commented and would like to know the correct way to restart inetd. And hoping that will fix it.

Thanks.

[Ganesan R]

Hi,

Just run #inetd -c to reread the configuration file. that's enough. No need to restart inetd daemon.

[Torsten.]

"port 3872 is enabled"

What process should answer port 3872 requests?

This is normally used by oracle.

[amreek bansal]

Yes I have tried inetd -c, it probably has re read the file, but hasnt fixed th etelent connection refused issue.

[amreek bansal]

yes should be oracle using port 3872

[Ganesan R]

Hi Amreek,

By default telnet will listen on port no 23. If you want to configure to listen on some other port comment out the existing line on /etc/inetd.conf and /etc/services and put the new entries like this...

#vi /etc/inetd.conf

newtelnet stream tcp nowait root /usr/lbin/telnetd telnetd

#vi /etc/services

newtelnet 3872/tcp # new telnet with port no 3872

then reread the inetd daemon by #inetd -c

Make sure that port 3872 is not used by any other service or application.

[Fabio Ettore]

Hi,

in my opinion port 3872 is not in listen currently, you can check that by

netstat -an|grep 3872

You're getting connection refused not for telnet issues, the reason is that port is not on listen. Just for a try, issue

netstat -an

and

telnet

HTH.

Best regards,
Fabio

[Torsten.]

I'm not a oracle expert at all, but I guess this certain application is not using inetd.

[Anoop P_2]

There are some cases where you have to restart inetd. A bug that caused a memory leak causing ssh to fail as inetd approaches its maximum allowable memory allocation for example.

The correct way to restart would be:

/sbhin/init.d/inetd stop; /sbin/init.d/inetd start

in single line, yes, to minimize the time when it remains down. And then ensure it started.

It is critical to start inetd immediately upon stopping it because many services are dependent on it. Particularly when you have clustered nodes, a failure of inetd beyond time out periods can spell disaster.

Regards,
Anoop

[Torsten.]

I don't know anything about this, but based on what I have found there is an apache daemon "behind" this port; likely not started by inetd but running "static".

So you may test with telnet, but modify the inetd.conf will create other problems and isn't a solution.