Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am trying to provide HA capabilities for our load balancers (nginx, pgpool and haproxy optionally) using pacemaker/corosync. I have defined three primitives:

1 for the failover IP,

1 for lsb::nginx

1 for lsb::pgpool

and I assume that I can add lsb::haproxy, although I have not tested this scenario.

and grouped them together. My issue is that when for example pgpool goes down, the cluster will switch to the other node, including nginx. From my understanding of the documentation, the failover IP can only be included in one group, making this an all or nothing situation (either all the services work or if one fails, a failover will take place for all three of them). Is it possible to group them together so for example node1 can provide nginx and node2 pgpool and haproxy, using the same virtual IP? Is it even possible? I am using Debian Wheezy with all the packages coming from its repositories (haproxy comes from backports). Thanks for any replies.

share|improve this question
up vote 2 down vote accepted

I think the easiest way out is to have 2 fail-over IPs. One for nginx and the other for pgpool + haproxy pair. That way it will work the way you intend it to. nginx can be configured to run from first machine and pgpool + haproxy pair from the second one.

share|improve this answer
so 3 fail-over IPs for each of the service – thanasisk Jun 24 '13 at 11:53
yes, but if you intend to run pgpool and haproxy from the same node, then you need only 2 failover ips. – n3rV3 Jun 24 '13 at 11:57

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.