Dhcpd response in peer
Glenn.Satchell at uniq.com.au
Sat Nov 22 12:55:31 UTC 2008
>Date: Fri, 21 Nov 2008 14:28:14 +0100
>From: Martin Hochreiter <linuxbox at wavenet.at>
>To: dhcp-users at lists.isc.org
>Subject: Dhcpd response in peer
>X-BeenThere: dhcp-users at lists.isc.org
>We have a particular problem with dhcpd:
>We are using a fail-over/load balancing configuration of the isc dhcpd.
>Master machine is running on Suse linux enterprise (10.0, dhcpd 3.0.3)
>with ldap support and is doing well.
>The second machine is a Cent OS 5.2. Cent OS 5.2 packaged dhcpd has no
>support for ldap, so i downloaded the original 3.0.5 dhcpd, patched it
>with Brian Masney's ldap patch and Ari Edelkind's paranoia patch to get
>ldap support and chroot support.
>The dhcpd daemon runs stable with querying the ldap server in its cage but
>it doesn't reply always to correctly to dhcp inquiries.
>I check the response every 30 minutes with nagios's check_dhcp with an
>to the failover, but it does not respond everytime.
>To check a possible failure of the check_dhcp script, i stopped the
>to see if the machines can get there adresses ... some do, some dont
>Nagios and the non-working machines show DHCPDISCOVER and DHCPOFFER in
>the logs of the failover peer but no DHCPACK's.
>I really have no idea where I could start to troubleshot here -
>can somebody give me a hint please?
>- if somebody is asking why I use that "old" version of dhcpd: I didn't
>get a recent
> version of dhcdp compiled with Masney's ldap patch (that we really need!)
You need to use the exact same version of dhcpd on both boxes. For
example there could be protocol diffeences in the failover protocol,
causing it to not work correctly or passing on invalid state or
This is documented in the dhcpd.conf man page in the first paragraph
under "DHCP FAILOVER":
If you wish
to use the failover protocol, make sure that both failover
peers are running the same version of the ISC DHCP server.
More information about the dhcp-users