Hi.
I'm trying to create a BlueStore OSD, but it seems something wrong somewhere.
My test environment is Ubuntu 16.04 + Ceph 10.2.2
(45107e21c568dd033c2f0a3107dec8f0b0e58374)
What i did:
I added the following section to my ceph.conf (i don't know how to
create BlueStore OSD on different devices):
[osd.0]
host = ceph-node-1
osd data = /var/lib/ceph/osd/ceph-0
bluestore block path = /dev/sda2
bluestore block db path = /dev/loop0
After that i did (this stage is OK):
ceph-deploy -v osd prepare --bluestore ceph-node-1:/dev/sda
After that i tried:
ceph-deploy -v osd activate ceph-node-1:/dev/sda1
On this stage i got several different errors:
1. EPERM: /dev/loop0 belongs root, need to chown it to ceph:ceph. I
think this is not a bug, but the documentation has to have a note
about this (big red font :-) )
2. EEXIST: 'prepare' stage creates a filesystem on the partition 1 and
creates required files on it. Also at this stage 'block' symlink is
created. Removing of this symlink helps me 'activate' the OSD. Bug?
3. EACCESS: before i had osd.0. I removed it, stopped the
corresponding daemon, removed the OSD from CRUSH map. But 'activate'
stage did not work. I googled the solution: "ceph auth del osd.0". Why
the removing of OSD does not clean CRUSH, Auth, etc. ?
Why it so complex?
The separate question:
ceph-deploy creates very simple config (without osd and mon sections).
How Ceph will start after reboot?
Thanks.
--
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
