Hi there. Is any way to prevent device from reservation during fencing? I have 
an shared qfs in cluster wnvironment and some clients outside a cluster. When 
one node going down, second own all available device include qfs shared target, 
that must always available to client. After that, clients stop accessing qfs 
with reservation conflict message and i got halted network. I not found any 
methods to stop fencing for specified device, so just remove it from 
did_instances with checksum correction. This cause some small bug, but cluster 
still operational. Any idea how to do it correctly?
--

This message posted from opensolaris.org


Reply via email to