On Tue, Sep 08, 2009 at 12:19:58PM +0200, Christoph Lechner wrote: > Andrew Beekhof wrote: > > The pre-reqs for a functional OCFS2 cluster (using the userspace stack) are: > > * pacemaker 1.0 > > * kernel >=2.6.26 > > * ocfs2-tools >= 1.4.1 > > * openais > > > > Heartbeat was only a possibility for SLES10 kernels which had this > > capability patched in. > > The patches were not accepted upstream and have since been deprecated. > > The chances of getting them to apply on a recent kernel are minimal. > So, bad news... > > Another question. Is there a risk of data loss/corruption in my case if > I use the (broken) Filesystem OCF script from 2.1.3? > I'd say no, given the fact that the servers don't have shared storage, > because it's a Primary/Slave DRBD configuration. DRBD makes sure that > only one host can write at the same time.
If you lose the replication connection, that is not exactly true. Ok, it still does not corrupt data. But data sets can diverge, which may do even more harm. > Actually it seems to me that we don't need OCFS because the idea is that > all hosts have simultaneous write access to the shared medium -- like > iSCSI or Fibre Channel. In our case only one host has write access at > the same time to a non-shared medium. > > OCFS2 on the disk is the legacy of the previous DRBD setup with two > primaries. So should we migrate to ext3? We postponed it as it would > induce quite a long downtime. What exactly is it that you are trying to achieve? -- : Lars Ellenberg : LINBIT | Your Way to High Availability : DRBD/HA support and consulting http://www.linbit.com DRBD® and LINBIT® are registered trademarks of LINBIT, Austria. _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
