2008/4/21, Andrew Beekhof <[EMAIL PROTECTED]>: > > On Apr 21, 2008, at 12:12 PM, Xinwei Hu wrote: > > > > Hi all, > > > > We are about to update mgmtd to support pacemaker, here are some > > issues we want to clarify. > > > > I hope you mean "support openais", since it already works with pacemaker > which is just the cluster manager :-) Yeah. Thanks for correction ;) > > > 1. mgmtd used to get a list of all nodes from hb. This is about to be > > changed. But should we get it from crm ? Or from ais ? The later one's > > better at first glance. > > > > Have a look at how the CIB process does it (eg. cib_init() ). > There is a bunch of stuff in libcrmcluster.so that hides most of the > details of the chosen stack. Good. > > 2. mgmtd use to get a list of hb parameters: > > apiauth, auto_failback, baud, debug, debugfile, deadping, > > deadtime, hbversion, hopfudge, > > initdead, keepalive, logfacility, logfile, msgfmt, > > nice_failback, node, normalpoll, > > stonith, udpport, warntime, watchdog, cluster > > I suppose all of these are not exist in pacemaker anymore. Will be > > safe to remove them all ? > Please don't remove them. > We still want the GUI to be able to support heartbeat-based clusters. ifdef them. Sorry for misleading. > > > 3. Type of node. For hb, there are normal node and ping node. What > > about pacemaker ? > > > There wont be any ping nodes in openais clusters. OK. > > _______________________________________________ > Pacemaker mailing list > [email protected] > http://list.clusterlabs.org/mailman/listinfo/pacemaker >
_______________________________________________ Pacemaker mailing list [email protected] http://list.clusterlabs.org/mailman/listinfo/pacemaker
