Crossing failover peers

[Jay Ford]

On Tue, 23 Nov 2010, Nicolas Ecarnot wrote:
> To make it short, in two sites, we have two servers, A & B.
> On A we are using a failover context called A-to-B, also used on B.
> This is working very well.
>
> On B, we are trying to use a failover context called B-to-A, and to use it on 
> A, in different shared-networks and pools.
>
> When restarting, my log files get full of blocking failover errors that lead 
> me to investigate.
>
> According to what I've read (and for some parts from what I've experienced), 
> I'm understanding the following:
> - I'm using V3.0.5-RedHat version (under RHEL5.3) of ISC DHCP, and at this 
> stage, failover support was not as brilliant as it is now.

That's a pretty crusty old version (as RedHat annoyingly seems to like to
latch onto for ISC-originated code), so I wouldn't be surprised if a lot of
stuff is not quite right compared to current ISC releases.

> - Though I tried to use different port numbers amongst my different failover 
> contexts, this does not help much.
> - From 3.1.1, the DHCP daemon manages much better the way failover 
> communication works between peers.
>
> Can anyone tell me if all this is true?
>
> And eventually, if upgrading all my DHCP daemon is worth the pain to expect 
> getting the failover peers crossing available?

I just tried a very similar thing with ISC 4.1.1-P1 (via Debian package
isc-dhcp-server) & had results similar to yours.  I concluded that the code
was not written to support multiple failover sessions between the same pair
of dhcpd instances and/or IP addresses, even if using different port numbers.
I then took to heart the following text from the dhcpd.conf man page:

     The failover protocol defines a primary server role and a secondary
     server role.  There are some differences in how primaries and secondaries
     act, but most of the differences simply have to do with providing a way
     for each peer to behave in the opposite way from the other.  So one
     server must be configured as primary, and the other must be configured as
     secondary, and it doesn't matter too much which one is which.

& quit trying to be too clever.

Your case of wanting differing failover behavior for different pools handled
by a pair of servers is an interesting one.  It seems not possible with
current ISC DHCP code, but I'd be happy to be told otherwise.

________________________________________________________________________
Jay Ford, Network Engineering Group, Information Technology Services
University of Iowa, Iowa City, IA 52242
email: jay-ford at uiowa.edu, phone: 319-335-5555, fax: 319-335-2951