Pacemaker: Groupping unrelated resources

Posted on

Hi everyone. I need your suggestions.

Let's assume that I have 20 in-house developed applications. None of them are related one with another, but they share purpose.

For example, I have 10 applications that provides one type of service, and 10 applications that provides another type of service.

Using pacemaker, if I create 20 resources, the management is a little more heavy.

I would like to group these resources, but as they are not related, the failure of any of them should not impact the others.

The problem with grouping for example Group1:Service1_instance1-10 and Group2:Service2_instance1-10, is that if any of the resources fail, the others (below) will be restarted. If the first instance in the group is restarted, all of them will be restarted.

Grouping simplifies the administration of the cluster.

What I would like to do is, to have groups, to manage easily the cluster, but to avoid the others resources restart in the group in case of one resource failure.

Any suggestions?

Thank you in advance.

Responses