Hello,
I tried the ceph packages from jessie, too. After some time penetrating
Google I think I found the solution. This will probably work for all package
sources.
You have to create an empty marker file named 'sysvinit' in the directories
below /var/lib/ceph/XXX. Then everythink works fine.
ceph-deploy will create those files for you depending on the init
system for the monitors:
$ ls /var/lib/ceph/mon/ceph-node1/
done keyring store.db upstart
$ cat /etc/issue
Ubuntu 12.04 LTS \n \l
And the same would happen for OSDs (upstart file is there):
$ ls /var/lib/ceph/osd/ceph-0/
activate.monmap active ceph_fsid current fsid journal keyring
magic ready store_version superblock upstart whoami
The output in ceph-deploy that tells you what init system will be used is this:
[ceph_deploy.osd][INFO ] Distro info: Ubuntu 12.04 precise
[ceph_deploy.osd][DEBUG ] activating host node1 disk /home/vagrant/foo
[ceph_deploy.osd][DEBUG ] will use init type: upstart
Can you share your ceph-deploy output as you are deploying your OSDs?
Sorry for the late reply.
I deleted my ceph setup I created with ceph-deploy. So I cannot give you
the info you requested. Perhaps I will create another setup using some
virtual maschines ...
I found some time to work on my 'manually' created setup (Wheezy, Ceph
Firefly repositories). Everything works fine and after manually creating
the empty 'sysvinit' files in the data dirs of the daemons, I am able to
start/stop the daemons on the same host. In my opinion the sysvinit
files should be mentioned in the manual install section of the
documentation.
One problem remains:
At the moment my ceph.conf file does not contain any configuration for
the single daemons, just a [global] section. The docs mention the
allhosts switch -a that doesn't work for me. I tried to add daemon
specific sections in the config file (host entries), but that made no
difference. Is the -a switch still supported for sysvinit? What entries
do I have to put into the config to remotely start/stop services?
It would be nice to start/stop the ceph cluster from a single host.
Thanks in advance.
Dieter
_______________________________________________
ceph-users mailing list
ceph-users@lists.ceph.com
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com