CEPH Monitor: 'client did not provide supported auth type'

[Nicolas Dey]

Hello,
I'm deploying CEPH 10.2.4-1~bpo80+1 on PVE 4.4.19-1-pve.
As you must already know, CEPH services are running as user 'ceph'. In the logs, I can see many lines like:

…
2016-12-08 10:45:07.055520 7fb9fd0c3700  1 mon.0@0(leader).auth v79 client did not provide supported auth type
2016-12-08 10:45:17.103639 7fb9fd0c3700  1 mon.0@0(leader).auth v79 client did not provide supported auth type
…

Therefore, I tried:

# su - ceph -s /bin/bash -c "ceph auth list"
2016-12-08 10:47:37.865420 7fb4b2940700 -1 auth: unable to find a keyring on /etc/pve/priv/ceph.client.admin.keyring: (13) Permission denied
2016-12-08 10:47:37.865427 7fb4b2940700  0 librados: client.admin initialization error (13) Permission denied
Error connecting to cluster: PermissionDeniedError
#

I've added user 'ceph' to the group 'www-data' to be able to read file '/etc/pve/priv/ceph.conf', but currently permission of '/etc/pve/priv/ceph.client.admin.keyring' are 0600 so only 'root' can read it. I would like to know how I could modify these permissions to 06400 without reinstalling CEPH or restarting the servers… It is always possible, but I'm sure there is a way to achieve this action gracefully. I have search through the Internet and the forum with

Of course, maybe I am going in the wrong direction, please feel free to tell me.

Best regards,
-- Nicolas D.

 

[Nicolas Dey]

Apparently, this is not what causes the root issue: I've manage to use a copy of that file in '/etc/ceph/' with the correct permissions on each node and I still get the errors:

…
2016-12-08 14:11:02.074272 7fd7ca85a710  1 mon.0@0(leader).auth v88 client did not provide supported auth type
2016-12-08 14:11:02.180787 7fd7ca85a710  1 mon.0@0(leader).auth v88 client did not provide supported auth type
…

I will edit title accordingly.
Thanks,
-- Nicolas D.

 

[Nicolas Dey]

I still have these messages, but I believe they are not harmful. I've fixed two things to have Proxmox see Ceph storage:

• chown ceph:ceph /dev/sdX2 - partition with the journal
• service pvedaemon restart - on each node

 

[Nicolas Dey]

I've checked the certificates as well, and I don't have this message anymore. I believe this message is directly related to certs.
Thanks.