>>> Florian Haas <[email protected]> schrieb am 09.11.2011 um 14:01 in >>> Nachricht <[email protected]>: > On 2011-11-09 13:36, Ulrich Windl wrote: > > Hi! > > > > I tried to co-locate a ocfs clone with a drbd ms-clone, and I tried to > co-locate a CTDB clone with the ocfs clone also. The idea was that the OCFS > filesystem is where the DRBD is, and the CTDB is where the filesystem is. So > actually that is a transitive colocation like: CTDB -> OCFS -> DRDB > > > > I guess CRM can't handle that even if CTDB is to be started before OCFS, > and OCFS before CTDB. Syslog has messages like: > > notice: clone_rsc_colocation_rh: Cannot pair prm_DLM:0 with instance of > msc_drbd_r0 > > notice: clone_rsc_colocation_rh: Cannot pair prm_ctdb:0 with instance of > cln_ocfs > > notice: clone_rsc_colocation_rh: Cannot pair prm_ctdb:1 with instance of > cln_ocfs > > > > <rsc_colocation id="col_ocfs_on_drbd_r0" rsc="cln_ocfs" score="INFINITY" > with-rsc="msc_drbd_r0"/> > > <rsc_colocation id="col_ctdb_ocfs" rsc="cln_ctdb" score="INFINITY" > with-rsc="cln_ocfs"/> > > > > It would be easier to maintain if resources would not require a multi-level > colocation (like depending CTDB on DRBD and depending OCFS on DRBD). > > > > What is the easiest solution? > > Group them all, then clone the group.
That would not work the general case, because you only need the ocfs framework once, the drbd framework less often, and the filesytem more often. Once you are having two DRBD devices ot two OCFS filesystems that won't work well. Ulrich _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
