4

I am trying to set up a Ceph storage cluster using the quick start guide found here: http://docs.ceph.com/docs/master/start/quick-ceph-deploy/

When I try to deploy a manager daemon using this command:

    ceph-deploy mgr create enickel7

I get this error:

    [ceph_deploy.mgr][ERROR ] OSError: [Errno 2] No such file or directory: '/var/lib/ceph/mgr/ceph-enickel7'
    [ceph_deploy][ERROR ] GenericError: Failed to create 1 MGRs

(enickel7 is the name of the node I'm using - the Ceph documentation calls the nodes node1, node2, and node3.) I tried to manually create the directory /var/lib/ceph/mgr, then ran the command again. Then I got this error:

    [enickel7][ERROR ] Error EACCES: access denied
    [enickel7][ERROR ] exit code from command was: 13
    [ceph_deploy.mgr][ERROR ] could not create mgr
    [ceph_deploy][ERROR ] GenericError: Failed to create 1 MGRs

Does anyone know what this error means, or how to fix it? ceph-deploy definitely has sudo permissions, and the mgr directory has the same permissions as other directories in /var/lib/ceph.

Thank you for your time!

Lillian Zhang
  • 41
  • 1
  • 3

5 Answers5

4

It's because your ceph version is not Luminous >=12.2.0, you must use ceph-deploy to install ceph as the document said, the default version installed by ceph-deploy is 10.2.10 Jewel for now.

If you want to create a manager daemon process, you need to upgrade your ceph to Luminous 12.2.1. The doc is here: http://docs.ceph.com/docs/master/release-notes/#v12-2-1-luminous

Michael Meepo
  • 51
  • 5
0

I just ran into this same issue on ubuntu 16.04 trying to deploy kraken with ceph-deploy version 1.5.39.

Ceph-deploy automatically created the directories for me but they were not owned correctly. It looks like the keyring it created in /var/lib/ceph/bootstrap-mgr along with that directory is owned by root. I chowned it to ceph. and that got me past that error.

In your case I would guess that the directory is owned by your user instead of "ceph". I hope this helps.

Lookcrabs
  • 43
  • 6
0

please test a below command:

chown ceph:ceph /var/lib/ceph

and

what ceph version used?

please use a latest version (mimic 13.2) and ceph-deploy 2

iraj norouzi
  • 111
  • 8
-1

Faced the same issue. As Michael Meepo said it was version problem. On admin node I registered the ceph repo for luminous & installed ceph-deploy. But when I tried to use it ceph-deploy installed the default version (Jewel) on remote node. To install specific version you should ask for it:

ceph-deploy install master --release luminous
Silk0vsky
  • 941
  • 1
  • 18
  • 34
-1

To use the ceph-deploy version matching your distribution's, as from https://github.com/ceph/ceph-deploy page, use ceph repositories. For instance, as Debian stretch provides Jewel (Ceph v. 10), use the following repository: http://ceph.com/debian-jewel by creating a /etc/apt/source.list.d/ceph-deploy.list file containing:

deb http://download.ceph.com/debian-jewel/ stretch main

Install the keys:

wget -q -O- 'https://download.ceph.com/keys/release.asc' | sudo apt-key add -

Then proceed with 

apt-get install ceph-deploy

From there it should work as expected.

wazoox
  • 1,384
  • 16
  • 27