[Pacemaker] benefits of cman?

[Andrew Beekhof]

On Sat, May 19, 2012 at 4:26 AM, Matthew O'Connor <matt at ecsorl.com> wrote:
> On 5/17/2012 7:40 PM, Andrew Beekhof wrote:
>> cman will not become required. In fact, depending on your
>> distro+version, it is being phased out and the membership+quorum
>> services it had will be provided directly by corosync 2.0. This change
>> is starting in fedora and its derivatives with f-17. However, for
>> distros+versions that have traditionally shipped RHCS (fedora prior to
>> f-17 and existing RHEL/CentOS/EPEL/... releases) it /is/ recommended
>> to keep using cman, since it is likely to have had more testing, and
>> to layer Pacemaker on top. This is particularly so if you want to use
>> cluster file systems such as GFS2, since the utilities they need only
>> understand cman. Does this help?
> Yes, thank you, that's very good to know!  I'm not certain where
> Canonical is headed with their support...I'm considering switching to
> either CentOS or openSUSE, so long as they have good kvm/qemu support.
> Canonical's most recent Ubuntu Server release has made it difficult to
> use OCFS2 without CMAN, while their previous releases seemed to lack a
> great deal of CMAN support with regard to Corosync+Pacemaker+OCFS2
> (perhaps they didn't compile everything in?).  I was thinking that it
> would be nice to not have to convert the OCFS2 cluster stack from pcmk
> to cman in the near future, and just go with what will be supported for
> the long-term.  Not sure which direction I'll go at this point...still a
> little uncertain if OCFS2 would benefit from cman.