Ceph

Hello,

Following the Quick installation guide, I ran into some issues which I think should be addressed:

[PREFLIGHT CHECKLIST]
http://ceph.com/docs/master/start/quick-start-preflight/

a) At no point do the instructions indicate which Distro version/architecture they were tested to work on. I tried Ubuntu 13.10 x86/x86_64, 12.04 x86/x86_64, and ONLY Ubuntu 12.04 x86_64 actually works as expected. I can only imagine that folks using non-ubuntu distros would have an even harder time with the instructions.

b) It's not clear which user on the Admin Node should be used to execute the 'ceph-deploy' commands. In my case, I used a 'cephadm' account for all Admin Node procedures, and I found that if the user on the Admin Node is not configured to run sudo with NOPASSWD, then ceph-deploy commands have a tendency to fail miserably. Perhaps this should be clarified in the docs.

[STORAGE CLUSTER QUICK START]
http://ceph.com/docs/master/start/quick-ceph-deploy/

CREATE CLUSTER:

Step 5: 'sudo mkdir /tmp/osd0' - OSD's probably shouldn't be created under /tmp (even for testing), because this directory has a tendency of being purged periodically on some systems. I would recommend using /var/local/ceph/osd0, or some other directory.

Step 7: 'ceph health' should be 'sudo ceph health', otherwise the command results in an error as follows:

cephadm@ceph-node1:~/cluster$ ceph health
2013-10-29 20:41:30.538219 7fe164e27700 -1 monclient(hunting): ERROR: missing keyring, cannot use cephx for authentication
2013-10-29 20:41:30.538645 7fe164e27700 0 librados: client.admin initialization error (2) No such file or directory
Error connecting to cluster: ObjectNotFound

ADDING MONITORS:

a) 'ceph-deploy mon create ceph-node2 ceph-node3' - This command generates an error and doesn't actually work.
As described here: http://www.mail-archive.com/ceph-users@lists.ceph.com/msg05093.html

The following additional steps must be taken prior to this step:
- Add pubblic_network definition to ceph.conf on admin host
- Push updated ceph.conf to other nodes using:
'ceph-deploy --overwrite-conf config push ceph-node1 ceph-node2'

b) 'ceph quorum_status' - the output of this command is a one-liner which isn't very useful to a novice. Recommend changing to 'ceph quorum_status --format json-pretty' so one can try to make some sense of the output.