dhcp failover

[Glenn Satchell]

>Date: Wed, 5 Jul 2006 11:00:21 +0400
>From: "Tony Aldo" <tony.aldo at gmail.com>
>To: dhcp-users at isc.org
>Subject: dhcp failover
>
>Hi, I have a 2 node cluster using the Linux HA project.
>I am not using any of DHCP's built in failover features but instead have
>linux ha manage dhcp by keeping one node running dhcp at a time.
>My question is, how bad is it not to mirror the leases databases between the
>two servers?
>
>Thanks
>Tony

As stated in other mails this is pretty bad.

A simple cron job to copy it every 5 minutes or so would probably be
satisfactory for anything but the most critical networks.

This is the issue with using something like linux ha - there needs to
be a way to keep stable storage between the master and the standby
system. A shared drive from a SAN can work well, but thi sis
expensive.

In the long run, dhcp failover works very well pretty much all of the
time, and allows the load to be shared across two servers, so you don't
have spare hardware just sitting there.

regards,
-glenn