On Fri, Jun 24, 2005 at 11:27:17AM -0500, Nate Carlson wrote: > Hey all, > > OK, I'm trying to get GULM working, and running into some initial issues - > just want to try to clarify a few things to make sure I'm understanding > properly. :) > > I'm basing things on the document at: > http://sources.redhat.com/cluster/gulm/gulmusage.txt > > I'm running the current CVS STABLE branch of code, on a 2.6.12 kernel. > > Now, for the questions: > > 1) With gulm, do I still need to join the cluster with 'cman_tool join' > and 'fence_tool join'? > > 2) Do I need to list all nodes in the cluster.conf, or just lock servers? > I realize I need to list lock servers separately in the <gulm> section, > and lock servers need to be listed as <clusternode>'s, but do the other > boxes also need to be listed? > > 2a) If other boxes do need to be listed, is it possible to add/remove > boxes without having to restart cluster services on all the other > boxes? > > I think that's about it for now, I'm sure more will come up. :) > > ------------------------------------------------------------------------ > | nate carlson | natecars@xxxxxxxxxxxxxxx | http://www.natecarlson.com | > | depriving some poor village of its idiot since 1981 | > ------------------------------------------------------------------------ > > -- > > Linux-cluster@xxxxxxxxxx > http://www.redhat.com/mailman/listinfo/linux-cluster > > You'd rather not use cman with gulm. It works better without it, as it can manage cluster membership, fencing, etc. Gulm use odd number of master servers and local servers for ecah node. So you must run lock_gulmd -s master_server1, master_server2, master_server3,etc -n tablename --name nodename. Cman is not needed. -- Linux-cluster@xxxxxxxxxx http://www.redhat.com/mailman/listinfo/linux-cluster
