On Wed, May 30, 2012 at 6:08 PM, Kaleb S. KEITHLEY <kkeithle(a)redhat.com> wrote:
On 05/30/2012 12:44 PM, Richard W.M. Jones wrote:
>
>
> To be honest it's a pain in the neck to deal with such packages, and
> unless there's an overwhelming need, I can't recommend it. Â Does any
> user really need to parallel install both versions of glusterfs?
No, and in fact that would not work. (And it's not the problem we're trying
to solve.)
If glusterfs-3.2.x + HekaFS is installed, we essentially want to avoid ever
updating to glusterfs-3.3.x because HekaFS is not compatible with it.
One way I can solve this is to never ship glusterfs-3.3.x on f16 and f17
(and EPEL el6).
That works as it's generally recognised that there shouldn't be major
version upgrades and breakages within a release especially one that
breaks things.
I'd be perfectly happy saying we will never ship glusterfs-3.3.x
on f16 and
f17, but the reality is that there probably are people who want it.
So you can always do a fedorapeople repository for those that want to
experiment.
Along the same lines, I'd be happy saying we will not ship HekaFS
in f18
once glusterfs-3.3.x is out, but there are probably people who want
glusterfs-3.2.x, with or without HekaFS.
Those people are then free to stick with Fedora 16, Fedora 17 or EL6.
And FWIW, doing nothing doesn't resolve the glusterfs in EPEL
versus
glusterfs in the RHS Channel issue.
That's a different story entirely, and why would you want gluster in
EPEL when it's already in RHEL? What's the difference?
Peter