RedHat Linux High Availability add-on behaviour

[A.W.R]

HI,

We are testing the RedHat High Availability add-on software with the Resilient storage add-on. We are having a slight issue:-

We have created a 2 node RedHat Linux cluster using virtual machines.

The cluster forms correctly, and the Global File system is fenced.

If we now reboot the one system, the system leaves the cluster correctly, but it does not rejoin the cluster on boot up.

If they reset the node and boot, it joins the cluster correctly.

If anybody can advise on why this should happen we would really appreaciate it.

Thanks

Andrew

[Matti_Kurkela]

Which version of RHEL? 5.x? 6.x?

What kind of virtual machines? VMware? RHEV? Other?

Which fencing method are you using?

Have you verified that multicast works between the nodes?

Here are some tools and procedures for testing:

https://access.redhat.com/site/articles/22304

With some switches (e.g. Cisco), IGMP snooping will work correctly only if there is a source of IGMP queries in the network segment/VLAN. If there is no source of IGMP queries, cluster can form if two nodes are started simultaneously, but if one node is stopped and then restarted, it will fail to rejoin the cluster because the forwarding of multicast traffic between the switches will not be restarted.

RedHat document:

https://access.redhat.com/site/node/216463/40/0

Cisco document:

http://www.cisco.com/en/US/products/hw/switches/ps708/products_tech_note09186a008059a9df.shtml