Custom Block Storage Back End Deployment Guide
A Guide to Deploying a Custom Block Storage Back End in a Red Hat OpenStack Platform Overcloud
The Red Hat OpenStack Platform Director is a toolset for installing and managing a complete OpenStack environment. It is based primarily on the OpenStack project TripleO (OpenStack-on-OpenStack). The Director’s primary objective is to fully orchestrate a functional, Enterprise-grade OpenStack deployment with minimal manual configuration. It helps address many of the issues inherent in manually configuring individual OpenStack components.
The end-result OpenStack deployment provided by the Director is called the Overcloud. The Overcloud houses all the components that provide services to end users, including Block Storage. This document provides guidance on how to deploy custom back ends to the Overcloud’s Block Storage service.
This document aims to leverage an administrator’s knowledge in manually configuring the Block Storage service. In a test deployment of OpenStack (for example, through Packstack), configuring this service involves editing its host node’s /etc/cinder/cinder.conf. Most of the Block Storage settings in that file are documented in better detail elsewhere; in this document, we discuss how to apply those same settings to the Overcloud in order to attach a custom back end.
This procedure has been tested successfully in limited use cases. Ensure that you test your planned deployment on a non-production environment first. If you have any questions, contact Red Hat support.
1.1. Custom Back Ends
For the purposes of this document, a custom back end is defined as a storage server/appliance or configuration that has yet to be integrated fully into the Red Hat OpenStack Platform Director. Some supported Block Storage back ends are already integrated into the Director; this means that pre-configured Director files are already provided out-of-the-box. An integrated back end can be configured and deployed to the Overcloud through these files. Examples of integrated back ends include Red Hat Ceph and single-back end configurations of Dell EqualLogic, Dell Storage Center, and NetApp appliances.
Further, some storage appliances already integrated into Director only support a single-instance back end. For example, the pre-configured Director files for Dell EqualLogic only allow for the deployment of a single back end. Deploying multiple back end instances of this appliance requires a custom configuration, as demonstrated in this document.
While you can manually configure the Block Storage service by directly editing its node’s /etc/cinder/cinder.conf, any settings will be overwritten by the Director in future, properly-orchestrated Overcloud updates. As such, the recommended method for deploying a Block Storage back end is through the Director. If a back end configuration is already fully integrated, you can simply edit and invoke its packaged environment files.
With custom back ends, however, you need to write your own environment file. This document includes an annotated sample that you can edit for your own deployment, namely /home/stack/templates/custom-env.yaml. This sample file is suitable for configuring the Block Storage service to use two NetApp back ends.
Other than prior knowledge on manually configuring Block Storage and the back end you want to deploy, this document also assumes that:
- If you are using third-party back end appliances, then they must already be properly configured as storage repositories.
- The Overcloud has already been deployed through Director, per instructions in Director Installation and Usage.
- You have the username and password of an account with elevated privileges. You can use the same account that was created to deploy the Overcloud; in Creating a Director Installation User, we create and use the stack user for this purpose.
- You have already mapped out the resulting configuration you want for the Block Storage back end in /etc/cinder/cinder.conf. With this, all that remains is the orchestration of your planned configuration through the Director.
2. Process Description
The Block Storage service’s settings are stored in /etc/cinder/cinder.conf; these settings include back end definitions. Most third-party back ends usable with (or even supported by) the Block Storage service provide setup instructions that involve editing /etc/cinder/cinder.conf settings. As mentioned in Section 1, “Introduction”, doing so will configure the Block Storage service; however, those settings will get overwritten in future Overcloud updates.
Regardless, any documentation relating to manual configuration through /etc/cinder/cinder.conf is still useful for Overcloud deployments. The Director, after all, applies the same configuration to /etc/cinder/cinder.conf, albeit through heat. As such, planning the back end configuration requires that you:
- Thoroughly plan the Block Storage back end configuration you want, and
- Map out the resulting /etc/cinder/cinder.conf file for this configuration.
Once you map out the resulting /etc/cinder/cinder.conf file, create the environment file that will orchestrate the back end settings. environment file describes this step in greater detail, using the sample file /home/stack/templates/custom-env.yaml. Having the environment file handy will help ensure that the back end settings persist through future Overcloud updates.
3. Create the Environment File
The environment file contains the settings for each back end you want to define. It also contains other settings relevant to the deployment of a custom back end. For more information about environment files, see Environment Files (in the Director Installation and Usage guide).
The following environment file defines two NetApp back ends, namely netapp1 and netapp2:
parameters: # 1 CinderEnableIscsiBackend: false CinderEnableRbdBackend: false CinderEnableNfsBackend: false NovaEnableRbdBackend: false GlanceBackend: file # 2 parameter_defaults: controllerExtraConfig: # 3 cinder::config::cinder_config: netapp1/volume_driver: # 4 value: cinder.volume.drivers.netapp.common.NetAppDriver netapp1/netapp_storage_family: value: ontap_7mode netapp1/netapp_storage_protocol: value: iscsi netapp1/netapp_server_hostname: value: 10.35.64.11 netapp1/netapp_server_port: value: 80 netapp1/netapp_login: value: root netapp1/netapp_password: value: 123456 netapp1/volume_backend_name: value: netapp_1 netapp2/volume_driver: # 5 value: cinder.volume.drivers.netapp.common.NetAppDriver # 6 netapp2/netapp_storage_family: value: ontap_7mode netapp2/netapp_storage_protocol: value: iscsi netapp2/netapp_server_hostname: value: 10.35.64.11 netapp2/netapp_server_port: value: 80 netapp2/netapp_login: value: root netapp2/netapp_password: value: 123456 netapp2/volume_backend_name: value: netapp_2 cinder_user_enabled_backends: ['netapp1','netapp2'] # 7
- The following parameters are set to false, and thereby disable other back end types we don’t need:
- CinderEnableIscsiBackend: other iSCSI back ends.
- CinderEnableRbdBackend: Red Hat Ceph.
- CinderEnableNfsBackend: NFS.
- NovaEnableRbdBackend: ephemeral Red Hat Ceph storage.
- The GlanceBackend parameter sets what the Image service should use to store images. The following values are supported:
- file: store images on /var/lib/glance/images on each Controller node.
- swift: use the Object Storage service for image storage.
- cinder: use the Block Storage service for image storage.
- controllerExtraConfig defines custom settings that will be applied to all Controller nodes. The cinder::config::cinder_config class means the settings should be applied to the Block Storage (cinder) service. This, in turn, means that the back end settings will ultimately end in the /etc/cinder/cinder.conf file of each Controller node.
- The netapp1/volume_driver and netapp2/volume_driver settings follow the section/setting syntax. With the Block Storage service, each back end is defined in its own section in /etc/cinder/cinder.conf. Each setting that uses the netapp1 prefix will be defined in a new [netapp1] back end section.
- Likewise, netapp2 settings are defined in a separate [netapp2] section.
- The value prefix defines the value of the preceding setting.
- The cinder_user_enabled_backends class sets and enables custom back ends. As the name implies, this class should only be used for user-enabled back ends; specifically, those defined in the cinder::config::cinder_config class.
Do not use cinder_user_enabled_backends to list back ends you can enable natively through Director. These include Red Hat Ceph, NFS, and single back ends for supported NetApp or Dell appliances. For example, if you are also enabling a Red Hat Ceph back end, do not list it in cinder_user_enabled_backends; rather, enable it using CinderEnableRbdBackend: true.
For more information on defining a Red Hat Ceph back end for OpenStack Block Storage, see Red Hat Ceph Storage for the Overcloud.
Section 4, “Deploy the Configured Back Ends” describes how to use the environment file /home/stack/templates/custom-env.yaml to orchestrate the custom back end’s deployment. To see the resulting /etc/cinder/cinder.conf settings from /home/stack/templates/custom-env.yaml, see Section A.2, “Resulting Configuration from Sample Environment File”.
4. Deploy the Configured Back Ends
Once you have created the custom-env.yaml file in /home/stack/templates/, log in as the stack user. Then, deploy the custom back end configuration by running:
$ openstack overcloud deploy --templates -e /home/stack/templates/custom-env.yaml
Once the Director completes the orchestration, test the back end. See Section 5, “Test the Configured Back End” for instructions.
5. Test the Configured Back End
After deploying the back ends to the Overcloud, test whether you can successfully create volumes on them. Doing so will require loading the necessary environment variables first. These variables are defined in /home/stack/overcloudrc by default.
To load these variables, run the following command as the stack user:
$ source /home/stack/overcloudrc
For more information, see Accessing the Basic Overcloud.
Next, create a volume type for each back end. Log in to the Controller node of the Overcloud as the stack user and run the following:
$ cinder type-create backend1 $ cinder type-create backend2
These commands will create the volume types backend1 and backend2, one for each back end defined through the cinder::config::cinder_config class of xref:envfile.
Finally, map each volume type to the volume_backend_name of a back end enabled through the cinder_user_enabled_backends class of xref:envfile. The following commands will map the volume type backend1 to netapp1 and backend2 to netapp2:
$ cinder type-key backend1 set volume_backend_name=netapp1 $ cinder type-key backend2 set volume_backend_name=netapp2
At this point, you should now be ready to test each back end. To do start, create a 1GB volume named netapp_volume_1 on the netapp1 back end by invoking the backend1 volume type:
$ cinder create --volume-type backend1 --display_name netappvolume_1 1
Likewise, create a similar volume on the netapp2 back end by invoking the backend2 volume type:
$ cinder create --volume-type backend2 --display_name netappvolume_2 1