[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

[Linux-cluster] wait states?



Hi all,

I've setup a simple two node cluster, and did some testing on it, but I'm 
having problems interpreting the results. Unfortunately I couldn't find any 
documentation answering my questions, so I'll post them here.

I don't want to do any complicated stuff, just run CLVM properly to serve 
logical volumes as iSCSI targets. The iSCSI target software should run on both 
nodes, independent of the cluster stack. The cluster is needed only because of 
CLVM. I tried to avoid using fencing as much as possible since I don't really 
see the need for it. My cluster.conf looks like this:

<cluster name="iscsigw" config_version="5">
  <cman two_node="2" expected_votes="1" />
  <clusternodes>
    <clusternode name="iscsigw1" nodeid="1">
      <fence />
      <unfence />
    </clusternode>
    <clusternode name="iscsigw2" nodeid="2">
      <fence />
      <unfence />
    </clusternode>
  </clusternodes>
  <fencedevices />
  <rm />
</cluster>

To test things, I broke the connection between the nodes for a while, and then 
restored it. I expected the cluster to return to normal state, but it didn't. 
The main difference between the nodes is the "wait state", about what I could 
hardly find any documentation. On one node it's "messages", on the other it's 
"quorum". Could you explain what these mean and how to return the cluster into 
normal state?

Thanks,
-- 
cc


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]