Hello,
at the moment I'm trying to create a small test cluster with Ceph
Firefly. I followed the documentation and everything works as expected.
But when I try to start the Ceph daemons on the individual OSD-machines
using the init scripts nothing happens.No daemon is started. No error
message on the console. No entries in the log files.
You're not telling us which version of Debian and which packages you used
(debian or inktank ones).
You're also not telling us which instructions you followed, the
ceph-deploy ones or the manual ones.
Sorry. I'm using Wheezy + Backports for Kernel 3.14 + Inktank Firefly
repositories (ceph + extra). I tried both path - the one with
ceph-deploy and the manual ones. With nearly the same result. If I do it
the manual way nothing happens on the OSDs. If I use ceph-deploy after
each reboot I must activate the OSDs again then the init script works
even for the OSDs.
If I manually start the daemons on the command line the OSDs come up as
expected and the cluster is healthy. My configuration file looks like
the one in the documentation.
That all said, I'm seeing the same thing probably on one of my clusters
running Jessie (ceph packages also from Jessie), the mon daemons come up
fine at boot time, the OSDs not and with no trace anywhere as to why.
Manually running the init script works as expected.
On my side running the init scripts manually doesn't work when I use the
manual instructions. I execute the daemons (ceph-osd) directly using the
id and the path to the ceph.conf file as parameters without using the
init scripts.
Thanks, Dieter
_______________________________________________
ceph-users mailing list
[email protected]
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
