-1

2 nodes, active/pasive. 2 resources, a virtual ip, openLdap, and the nfs mount where openldap saves the data.

When both nodes are up, things worked fine. You could move resources away and put the active in stanby. But when i rebooted the passive node, ( with the resources in the active node), and the passive node loses conectivity, all the resources in the active where stopped by pacemaker.

I'm reading the documentation right now, but I just need a little quick tip to figure what could be hapenning here.

Im using:

  • corosync
  • pacemaker
  • RHEL 6

1 Answer 1

1

Quick answer, run this: crm configure property no-quorum-policy=ignore

Long answer: With only two nodes, you no longer have over 50% of the active nodes agreeing on something (quorum), so the remaining node shuts down as to not cause any damage. Picture this situation, both nodes have access to a shared storage device, but have lost contact with each other. They both try to do different filesystem writes, which leads to disk corruption. Pacemaker prevents this by default, by requiring at least 50% of the nodes to agree on which one is the master. See section 5.3.1 here: http://www.clusterlabs.org/doc/en-US/Pacemaker/1.1/html/Clusters_from_Scratch/ch05s03.html

1
  • Thanks! i forgot to check again on this... i thought that i would receive an email notification :S I figured out reading the documentation, and the solution was to set the no-quorum-policy to "ignore".
    – Wakaru44
    Sep 18, 2012 at 18:45

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .