got a question / problem related to NTP and use of virtual IPs. Here's the background:

1 server, RedHat 2.4.9, 2
ethernet interfaces: eth1, eth3
virtual interfaces: eth1:1, eth3:1

All interfaces working just fine for things like pings, ftps, telnets, and DNS.

NTP working just fine on eth1, eth3, but won't return answers to queries to the IPs for eth1:1 and eth3:1. tcpdump shows the NTP queries hitting eth1:1 and eth3:1, but nothing is returned, just a timeout from the host originating the NTP query

Is there some "issue" with NTP on Linux and the use of virtual IPs? I've done this in the past on HP-UX, BSDI, and Solaris...

bp3434

I assume by "NTP" you mean "network time".

If the problem's with a client querying using a virtual IP: have you tried pointing to a different NTP server? For example, to "time.windows.com"?

If the problem's with an NTP server you have configured on a virtual IP address: does it work if you configure your hostnames so that the NTP service points to the "real" address? Have you tried downloading and installing a different version of the NTP daemon?

In either case, have you tried taking an Ethereal trace, and does it show anything interesting?

I honestly can't imagine any reason NTP should care whether one or both endpoints are a "real" or a "virtual" IP address - it should work either way.

Sorry I can't be of more help .. PSM