topic Re: Address already in use in Operating System - HP-UX

Address already in use

Kurt Henning — Tue, 28 Nov 2000 21:26:45 GMT

I'm trying to load the time protocol daemon (xntpd)and keep getting the following message in the syslog.

Nov 28 14:55:23 b30-ud01 xntpd[28676]: bind() fd 16, family 2, port 123, addr c001ffff, in_classd=0 flags=0 fails: Address already in use

I've tried reconfiguring inetd, checked /etc/services, and looked for another process using port 123 with lsof and netstat. So far, no luck. I'm not sure what address c001ffff is referring to. Any insights?

Re: Address already in use

Rick Garland — Tue, 28 Nov 2000 21:58:49 GMT

You possibly have another ntp process running.
Check the processes and grep for ntp. There should only be one on them.

Re: Address already in use

Kurt Henning — Tue, 28 Nov 2000 23:26:38 GMT

Nope. That was the first thing I looked for.

Re: Address already in use

James R. Ferguson — Wed, 29 Nov 2000 02:22:03 GMT

Kurt:

Make sure you don't have 'ntpdate' running. It uses the same port as 'xntpd'.

...JRF...

Re: Address already in use

James R. Ferguson — Wed, 29 Nov 2000 13:09:17 GMT

Kurt:

Seems that if you can't find reference to the port with 'netstat -an', 'lsof', and if neither 'xntpd' nor 'ntpdate' is running, then I would guess that the socket is in some limbo state from a previous instance that is no longer viable. Does a reboot cure the problem?

...JRF...

Re: Address already in use

Lasse Knudsen — Wed, 29 Nov 2000 15:32:25 GMT

Hmmm . It seems like you have checked all there is.

I can only comment it to further add that address c001ffff is a hex IP-address - in this case 192.1.255.255 which is a bit strange. I do not think it is the address of your machine - but maybe a broadcast address.

Do you by any chance have a subnet mask of 255.255.0.0 on a class C-net ?
That could give some strange errs.

/Allan

Re: Address already in use

Kurt Henning — Wed, 29 Nov 2000 20:12:05 GMT

No "ntp" anything is running according to ps -ef | grep ntp. The primary heartbeat network for Service Guard is 192.1.0.0, so I suppose that's where the 192.1.255.255 is coming from, but I don't understand why it's trying to bind to a broadcast address and I'm not sure how to force it to try a different address that actually corresponds to an interface in the machine.

The xntpd daemon loads without a hitch on the sister box to this one. Lsof indicates that ntp is bound to every IP address, and every logical address, on that host.

I still need to try a reboot. It's just a matter of when I can pry the system away from the users.