DHCP failover?

Ryan McCain Ryan.McCain at dss.state.la.us
Thu Sep 20 05:47:39 UTC 2007


I'm reading through Release Notes of different versions of DHCP and it seems the failover code stabilizes in the 3.1 code.  Is that an accurate statement?

>>> Simon Hobson <dhcp1 at thehobsons.co.uk> 09/15/07 3:11 AM >>>
Ryan McCain wrote:

>I was reading back through older posts regarding failover and came 
>across this response to one of my posts. 
>Simon, could you clarify why using rsync would be easier? Is it 
>because the dhcp failover implementation is still buggy?

It's not to do with failover per-se, but in how to manage your 
config. I believe what many people do is something like this :

Master :
<failover stuff for master>
INCLUDE <common stuff in a separate file>

Slave :
<failover stuff for slave>
INCLUDE <common stuff in a separate file>

Apart from the failover peer declarations, the rest of the config is 
common to both servers, so it makes sense not to keep two different 
files which can get out of sync. So management becomes something like 

Edit common include file on master
Test config & restart master
Copy common file to slave
Restart slave

It's that 3rd step where rsync (or anything that works for you) comes 
in. One technique I use is for the script that generates the config 
file to use rsync to copy a config file to another machine - on the 
other machine, a cron job checks every few minutes and if a new file 
has appeared it tests it and then makes it the live config.

More information about the dhcp-users mailing list