I get an issue similar as described in TID 10096378.

The scenario could be descibed as the following:
Most people loose connection to the cluster resource while some still do
have the connection. When the resource is taken offline an online again
the connection will restart working. However a couple of minutes later
the game is starting over.

The procedure with XK3 does no more make sense because we started now
to give the trustees directly on the partition (to be no more dependant
of the external references). By the way the volume is all the time
acessible via the physical node. The volume doesn't become unresponsive
by accessing it this way.

The environment is Netware 6.5 SP3.