On 07/21/2009 02:46 PM, Bela Ban wrote:
* Sometimes, a webapp on node2 fails over to node3 rather than
node1
Can someone look into this ?
I have found when it happens... Once you remove a node (clean shutdown)
mod_cluster will remove the information about the node so it have no way
to find the domain, that is some work to fix it (need another shared
area with removed nodes, just JVMRoute | domain).
There is a JIRA for it: MODCLUSTER-82
Cheers
Jean-Frederic