On Mon, 2007-11-19 at 09:26 +0100, Reiner Rottmann wrote: > Hi! > > Say you want to configure a cluster node by using a quorum disk so that it > survives when the particular node runs a certain cluster service. > > Such a scenario is even stated in the man page QDisk(8): > ---%------------------------------------------------------------------------------------------------------------ > For example, a user may have a service running on one node, and that node must > always be the master in the event of a network partition. > ---%------------------------------------------------------------------------------------------------------------ > > Then in my opinion you have a chicken/egg dilemma because you might need the > additional vote of the quorum disk to get a quorate cluster (which again may > run the rgmanager that is needed for the heuristic). Survival after a failure is not the same thing as trying to clean-start in a broken environment. > How would one solve this dilemma in a straightforward way? (a) Boot both nodes ? (b) Manual intervention. :) More roundabout ways include using heuristics which record state. Admittedly we could make this a bit easier if we handle it more gracefully inside qdiskd. -- Lon -- Linux-cluster mailing list Linux-cluster@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/linux-cluster
