okd/playbooks/openshift-glusterfs

OpenShift GlusterFS Playbooks

These playbooks are intended to enable the use of GlusterFS volumes by pods in OpenShift. While they try to provide a sane set of defaults they do cover a variety of scenarios and configurations, so read carefully. :)

Playbook: config.yml

This is the main playbook that integrates GlusterFS into a new or existing OpenShift cluster. It will also, if specified, configure a hosted Docker registry with GlusterFS backend storage.

This playbook requires the glusterfs group to exist in the Ansible inventory file. The hosts in this group are the nodes of the GlusterFS cluster.

• If this is a newly configured cluster each host must have a glusterfs_devices variable defined, each of which must be a list of block storage devices intended for use only by the GlusterFS cluster. If this is also an external GlusterFS cluster, you must specify openshift_storage_glusterfs_is_native=False. If the cluster is to be managed by an external heketi service you must also specify openshift_storage_glusterfs_heketi_is_native=False and openshift_storage_glusterfs_heketi_url=<URL> with the URL to the heketi service. All these variables are specified in [OSEv3:vars],
• If this is an existing cluster you do not need to specify a list of block devices but you must specify the following variables in [OSEv3:vars]:
  • openshift_storage_glusterfs_is_missing=False
  • openshift_storage_glusterfs_heketi_is_missing=False

By default, pods for a native GlusterFS cluster will be created in the default namespace. To change this, specify openshift_storage_glusterfs_namespace=<other namespace> in [OSEv3:vars].

To configure the deployment of a Docker registry with GlusterFS backend storage, specify openshift_hosted_registry_storage_kind=glusterfs in [OSEv3:vars]. To create a separate GlusterFS cluster for use only by the registry, specify a glusterfs_registry group that is populated as the glusterfs is with the nodes for the separate cluster. If no glusterfs_registry group is specified, the cluster defined by the glusterfs group will be used.

To swap an existing hosted registry's backend storage for a GlusterFS volume, specify openshift_hosted_registry_storage_glusterfs_swap=True. To additoinally copy any existing contents from an existing hosted registry, specify openshift_hosted_registry_storage_glusterfs_swapcopy=True.

NOTE: For each namespace that is to have access to GlusterFS volumes an Enpoints resource pointing to the GlusterFS cluster nodes and a corresponding Service resource must be created. If dynamic provisioning using StorageClasses is configure, these resources are created automatically in the namespaces that require them. This playbook also takes care of creating these resources in the namespaces used for deployment.

An example of a minimal inventory file:

[OSEv3:children]
masters
nodes
glusterfs

[OSEv3:vars]
ansible_ssh_user=root
deployment_type=origin

[masters]
master

[nodes]
node0
node1
node2

[glusterfs]
node0 glusterfs_devices='[ "/dev/sdb" ]'
node1 glusterfs_devices='[ "/dev/sdb", "/dev/sdc" ]'
node2 glusterfs_devices='[ "/dev/sdd" ]'

Playbook: registry.yml

This playbook is intended for admins who want to deploy a hosted Docker registry with GlusterFS backend storage on an existing OpenShift cluster. It has all the same requirements and behaviors as config.yml.

Role: openshift_storage_glusterfs

The bulk of the work is done by the openshift_storage_glusterfs role. This role can handle the deployment of GlusterFS (if it is to be hosted on the OpenShift cluster), the registration of GlusterFS nodes (hosted or standalone), and (if specified) integration as backend storage for a hosted Docker registry.

See the documentation in the role's directory for further details.

Role: openshift_hosted

The openshift_hosted role recognizes glusterfs as a possible storage backend for a hosted docker registry. It will also, if configured, handle the swap of an existing registry's backend storage to a GlusterFS volume.