I forgot to mention you can change the ceph-init script in the packaged
version too if you don't want to compile ceph from source. It's a very
easy modification and it should work fine afterwards.
Mark
On 11/12/2013 07:56 AM, Mark Nelson wrote:
FYI, I created a bug for this last week:
http://tracker.ceph.com/issues/6720
Mark
On 11/12/2013 06:22 AM, Dave (Bob) wrote:
On 12/11/2013 07:22, Wido den Hollander wrote:
On 11/11/2013 06:51 PM, Dave (Bob) wrote:
The utility mkcephfs seemed to work, it was very simple to use and
apparently effective.
It has been deprecated in favour of something called ceph-deploy, which
does not work for me.
I've ignored the deprecation messages until now, but in going from
70 to
72 I find that mkcephfs has finally gone.
I have tried ceph-deploy, and it seems to be tied in to specific
'distributions' in some way.
It is unuseable for me at present, because it reports:
[ceph_deploy][ERROR ] UnsupportedPlatform: Platform is not supported:
I therefore need to go back to first principles, but the documentation
seems to have dropped descriptions of driving ceph without smoke and
mirrors.
The direct approach may be more laborious, but at least it would not
depend on anything except ceph itself.
I myself am not a very big fan of ceph-deploy as well. Most
installations I do are done by bootstrapping the monitors and osds
manually.
I have some homebrew scripts for this, but I mainly use Puppet to make
sure all the packages and configuration is present on the nodes and
afterwards it's just a matter of adding the OSDs and formatting their
disks once.
The guide to bootstrapping a monitor:
http://eu.ceph.com/docs/master/dev/mon-bootstrap/
When the monitor cluster is running you can start generating cephx
keys for the OSDs and add them to the cluster:
http://eu.ceph.com/docs/master/rados/operations/add-or-rm-osds/
I don't know if the docs are 100% correct. I've done this so many
times that I do a lot of things without even reading the docs, so
there might be a typo in it somewhere. If so, report it so it can be
fixed.
Where I think that ceph-deploy works for a lot of people I fully
understand that some people just want to manually bootstrap a Ceph
cluster from scratch.
Wido
Maybe I need to step back a version or two, set up my cluster with
mkcephfs, then switch back to the latest to use it.
I'll search the documentation again.
--
To unsubscribe from this list: send the line "unsubscribe
ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Thank you very much Wido.
David
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
