Skip to main content

Ceph via Rook (technical preview)

warning

This is a technical preview and not recommended for production use yet.

In OSISM it is also possible to integrate and use existing Ceph clusters. It is not necessary to deploy Ceph with OSISM. If Ceph is deployed with OSISM, it should be noted that OSISM does not claim to provide all possible features of Ceph. Ceph provided with OSISM is intended to provide the storage for Glance, Nova, Cinder and Manila. In a specific way that has been implemented by OSISM for years. It should be checked in advance whether the way in OSISM the Ceph deployment and the provided features are sufficient. If this is not the case, it is recommended to deploy Ceph in a different way directly and independently of OSISM. For possible open source projects, please refer to cephadm and Rook.

  1. Deploy services.

When using rook, all services are deployed via a single helm chart and at the same time. This could be altered by passing custom CRDs. See Rook Configuration Guide.

  • Install Kubernetes Cluster

  • Deploy Rook Operator Helm Chart

    osism apply rook-operator

  • Deploy complete Rook Ceph Cluster

    osism apply rook

  • Copy ceph keyrings to kolla directories (if deploying OpenStack)

    osism apply rook-fetch-keys

  1. Get ceph keyrings. This places the necessary keys in /opt/configuration.

    osism apply rook-fetch-keys

    After run, these keys must be permanently added to the configuration repository via Git.

    environments/infrastructure/files/ceph/ceph.client.admin.keyring
    environments/kolla/files/overlays/gnocchi/ceph.client.gnocchi.keyring
    environments/kolla/files/overlays/nova/ceph.client.cinder.keyring
    environments/kolla/files/overlays/nova/ceph.client.nova.keyring
    environments/kolla/files/overlays/cinder/cinder-backup/ceph.client.cinder.keyring
    environments/kolla/files/overlays/cinder/cinder-backup/ceph.client.cinder-backup.keyring
    environments/kolla/files/overlays/cinder/cinder-volume/ceph.client.cinder.keyring
    environments/kolla/files/overlays/manila/ceph.client.manila.keyring
    environments/kolla/files/overlays/glance/ceph.client.glance.keyring

    You can also overwrite the rook_cephclients parameter to skip these keys.

    environments/rook/configuration.yml
    rook_cephclients: {}

  2. A Ceph client (a wrapper on the manager for entering the rook toolbox) can be deployed.

    osism apply cephclient

    You have to make sure the correct Configuration Options for the Rook Ceph Client Wrapper are net.

  3. After getting the Ceph Keyrings, the OpenStack Deployment can optionally be done.

RGW service​

Deployment of the Ceph RGW Service is enabled by default in rook. This is done by creating a default CephObjectStore CRD. How the Ceph RGW service can be deployed and integrated into OpenStack is described here.

info

OpenStack integration between Keystone/Swift and Rook is currently missing upstream in Rook. Please have a look at #1027 to get the current status of the integration in OSISM.

Change node labels​

In case you decided to move workloads to different nodes and changed the inventory groups e.g. like this:

inventory/20-roles
[rook-mds:children]
ceph-control

[rook-mgr:children]
ceph-control

[rook-mon:children]
ceph-control

[rook-osd:children]
ceph-resource

[rook-rgw:children]
ceph-control

You can apply the changes running:

osism apply rook-change-labels

This will remove all labels and apply the changed inventory groups as labels. After those steps are done it will trigger the rescheduling of the components so they get deployed on the adjusted nodes.

Cleanup​

warning

This will permanently delete all your data in the Ceph Cluster. Be sure you know what you are doing before proceeding.

If you want to cleanup/delete the whole cluster, you can do that by enabling rook_cleanup.

environments/rook/configuration.yml
rook_cleanup: true

And running the rook-cleanup role.

osism apply rook-cleanup