[Linux-HA] new resources insist on trying to start on the wrong
optimalix at gmail.com
Fri Apr 18 13:38:58 MDT 2008
On a cluster running a couple of Xen VM's, i'm adding another two. The VM's
run out of disk files placed on top of a DRBD fs, so i configure constraints
to have the resources run where the fs resource is active, which in turn
starts where the drbd master is. This is working just fine for the already
running VM's. I then try to add the new contraints for the new VM's, then
it's resources. What i then see is that the new VM's try to start on the
secondary node, and obviously stay unmanaged. At first i thought maybe this
could happen because the cluster is symmetric (still, shouldn't), so then i
try to force their migration with crm_resource -M -H, but they just won't
move to the active node. I then have to delete and cleanup them.
Any hints on this?
Btw, this is (now) 2.1.3 (built from srpm) running on CentOS 5.1, but same
thing was happening with 2.1.2.
More information about the Linux-HA