Converting an existing cluster to cephadm

Cephadm allows you to convert an existing Ceph cluster that has been deployed with ceph-deploy, ceph-ansible, DeepSea, or similar tools.

Limitations

  • Cephadm only works with BlueStore OSDs. If there are FileStore OSDs in your cluster you cannot manage them.

Preparation

  1. Get the cephadm command line tool on each host in the existing cluster. See Install cephadm.

  2. Prepare each host for use by cephadm:

    1. # cephadm prepare-host
  3. Determine which Ceph version you will use. You can use any Octopus (15.2.z) release or later. For example, docker.io/ceph/ceph:v15.2.0. The default will be the latest stable release, but if you are upgrading from an earlier release at the same time be sure to refer to the upgrade notes for any special steps to take while upgrading.

    The image is passed to cephadm with:

    1. # cephadm --image $IMAGE <rest of command goes here>
  4. Cephadm can provide a list of all Ceph daemons on the current host:

    1. # cephadm ls

    Before starting, you should see that all existing daemons have a style of legacy in the resulting output. As the adoption process progresses, adopted daemons will appear as style cephadm:v1.

Adoption process

  1. Ensure the ceph configuration is migrated to use the cluster config database. If the /etc/ceph/ceph.conf is identical on each host, then on one host:

    1. # ceph config assimilate-conf -i /etc/ceph/ceph.conf

    If there are config variations on each host, you may need to repeat this command on each host. You can view the cluster’s configuration to confirm that it is complete with:

    1. # ceph config dump
  2. Adopt each monitor:

    1. # cephadm adopt --style legacy --name mon.<hostname>

    Each legacy monitor should stop, quickly restart as a cephadm container, and rejoin the quorum.

  3. Adopt each manager:

    1. # cephadm adopt --style legacy --name mgr.<hostname>
  4. Enable cephadm:

    1. # ceph mgr module enable cephadm
    2. # ceph orch set backend cephadm
  5. Generate an SSH key:

    1. # ceph cephadm generate-key
    2. # ceph cephadm get-pub-key > ~/ceph.pub
  6. Install the cluster SSH key on each host in the cluster:

    1. # ssh-copy-id -f -i ~/ceph.pub root@<host>

    Note

    It is also possible to import an existing ssh key. See ssh errors in the troubleshooting document for instructions describing how to import existing ssh keys.

  7. Tell cephadm which hosts to manage:

    1. # ceph orch host add <hostname> [ip-address]

    This will perform a cephadm check-host on each host before adding it to ensure it is working. The IP address argument is only required if DNS does not allow you to connect to each host by its short name.

  8. Verify that the adopted monitor and manager daemons are visible:

    1. # ceph orch ps
  9. Adopt all OSDs in the cluster:

    1. # cephadm adopt --style legacy --name <name>

    For example:

    1. # cephadm adopt --style legacy --name osd.1
    2. # cephadm adopt --style legacy --name osd.2
  10. Redeploy MDS daemons by telling cephadm how many daemons to run for each file system. You can list file systems by name with ceph fs ls. Run the following command on the master nodes:

    1. # ceph orch apply mds <fs-name> [--placement=<placement>]

    For example, in a cluster with a single file system called foo:

    1. # ceph fs ls
    2. name: foo, metadata pool: foo_metadata, data pools: [foo_data ]
    3. # ceph orch apply mds foo 2

    Wait for the new MDS daemons to start with:

    1. # ceph orch ps --daemon-type mds

    Finally, stop and remove the legacy MDS daemons:

    1. # systemctl stop ceph-mds.target
    2. # rm -rf /var/lib/ceph/mds/ceph-*
  11. Redeploy RGW daemons. Cephadm manages RGW daemons by zone. For each zone, deploy new RGW daemons with cephadm:

    1. # ceph orch apply rgw <realm> <zone> [--subcluster=<subcluster>] [--port=<port>] [--ssl] [--placement=<placement>]

    where <placement> can be a simple daemon count, or a list of specific hosts (see Placement Specification).

    Once the daemons have started and you have confirmed they are functioning, stop and remove the old legacy daemons:

    1. # systemctl stop ceph-rgw.target
    2. # rm -rf /var/lib/ceph/radosgw/ceph-*

    For adopting single-site systems without a realm, see also Migrating a Single Site System to Multi-Site.

  12. Check the ceph health detail output for cephadm warnings about stray cluster daemons or hosts that are not yet managed.