[quagga-users 12975] Re: Quagga HA with pacemaker

James Andrewartha jandrewartha at ccgs.wa.edu.au
Mon Oct 8 09:36:09 BST 2012

On 14/02/12 10:31, David Coulson wrote:
> On 2/13/12 9:24 PM, James Andrewartha wrote:
>> The ones in Debian squeeze don't implement status.
> RedHat ones appear to.
> [root at dresproddns01 ~]# /etc/init.d/bgpd status
> bgpd (pid 1113858) is running...

Usefully Debian now has /lib/lsb/init-functions which means I could
implement a status check easily:

status_of_proc -p `pidfile watchquagga` watchquagga Quagga && exit 0 ||
exit $?

>> Could you provide some more information about your setup?
> We run BGP into our core from two Linux clusters (different datacenters) 
> and advertise /32s originating from clone resources in pacemaker, or 
> simply from a single resource which fails over between cluster members. 
> We use a lot of BGP knobs, such as localpref and prepend in Quagga to 
> set route preferences across our network. We don't manipulate the BGP 
> sessions via pacemaker - We just advertise or withdraw IPs from the 
> network using it.

I've got quagga being controlled by pacemaker, however zebra appears to
leave behind IPs defined in zebra.conf when it's stopped, which is
suboptimal. Is this the expected behaviour?

Perhaps I should manage the gateway IPs with pacemaker and just have
quagga manage BGP sessions.

James Andrewartha
Network & Projects Engineer
Christ Church Grammar School
Claremont, Western Australia
Ph. (08) 9442 1757
Mob. 0424 160 877

More information about the Quagga-users mailing list