Can someone please explain to me how fence_vmware_soap work?

Question

I was able to set up fence_vmware_soap in my cluster and I know it is used to prevent data corruption so that two nodes do not write to a shared storage (luns in my case) at the same time. The fence will make sure the unhealthy node is completely down before the active node is able to take over and write to the shared disk.

I will like to know what happens and how the one node in the cluster knows the other node is unhealthy before the unhealthy node kills itself using fence_vmware_soap agent.

I will really appreciate an answer explaining it in a very simple way because this is my first time setting up an nfs cluster (active /passive)

Answer 1

I know this thread is kinda old, but:

First try to reach your vmware cluster is available:

# fence_vmware_soap -a my_host_ip -l my_user -p my_pw --ssl -z -v -o list

I don't know how to do it with pacemaker, but the solution without is to change following in your cluster.conf:

<clusternode name="n1" nodeid="2" votes="1">
  <fence>
   <method name="1">
      <device name="vmwarefence" port="rhel5rhcs-node1"
           uuid="xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxx"/>
   </method>
  </fence>
</clusternode>
</clusternodes>
<cman expected_votes="1" two_node="1"/>
<fencedevices>
  <fencedevice agent="fence_vmware_soap" ipaddr="xxx.xxx.xxx.xxx"
     login="root" name="vmwarefence" passwd="pwd" ssl="1"/>
</fencedevices>

You can afterwards check the cluster status by

# clustat