Re: Servers just doen't work (after following the troubleshooting page)

On Tue, 27 Sep 2005, Per Hedeland wrote:
The client sends a query to ntp2.<mydomain>, but gets the response from server2.<mydomain>, i.e. presumably a different IP address on the same host. The client will not (and should not) be interested in responses coming from what it sees as "someone else". The same effect will apply to the intra-server/peer queries.

That proved to be the problem! I replaced those ntp addresses with the primary addresses of the servers' interfaces, and it works like a charm, now.

Note to self: udp communications may not work with secondary IP / IP aliases, specifically ntp won't work.

Probably if I use SNAT on outgoing packet, it will work with the secondary IP, but I feel it is too much trouble. ntp have their own way with high-availability anyway.

Has ntpd bound a socket to ntp2.<mydomain>:ntp? You can check with e.g. 'neststat -an | grep :123' - all the IP addresses used must be individually listed there. I'm not sure if ensuring this avoids the bug, but I would guess so. 

Actually this was not. All the IPs was listed once before.

Thank you!

--
                               Stephan Paul Arif Sahari Wibowo
    _____  _____  _____  _____
   /____  /____/ /____/ /____
  _____/ /      /    / _____/       http://www.arifsaha.com/
.

Relevant Pages