Add tech preview of local storage in cinder

[janhorstmann]

• Create volume group on third supplied block device
• Checkout cinder dm-clone driver
• Add compute nodes to storage group
• Deploy cinder lvm backend and iscsi containers
• Add and enable local backend using dm-clone driver
• Add volume type for local backend

[janhorstmann]

This still needs to be tested

@berendt Fo now I have used the third supplied block device as local storage. From my understanding it is only used for ceph when explicitly requested. Should I add additional storage for the dm-clone driver in the testbed terraform or should I make its deployment optional and only deploy when the third device is not used for ceph :thinking:

[berendt]

@janhorstmann Please fix the Ansible Lint issues.

[janhorstmann]

This still needs to be tested

@berendt Fo now I have used the third supplied block device as local storage. From my understanding it is only used for ceph when explicitly requested. Should I add additional storage for the dm-clone driver in the testbed terraform or should I make its deployment optional and only deploy when the third device is not used for ceph 🤔

Test was okay. The issue with the third block device still exists.

[janhorstmann]

This still needs to be tested @berendt Fo now I have used the third supplied block device as local storage. From my understanding it is only used for ceph when explicitly requested. Should I add additional storage for the dm-clone driver in the testbed terraform or should I make its deployment optional and only deploy when the third device is not used for ceph 🤔

Test was okay. The issue with the third block device still exists.

As discussed, the third block device will be used for volume type local

[janhorstmann]

On my last testbed deployment the volume group for the local storage was not created. The playbook itself worked fine, but it was not created automatically. I will draft this again, until I have time to investigate.

[janhorstmann]

On my last testbed deployment the volume group for the local storage was not created. The playbook itself worked fine, but it was not created automatically. I will draft this again, until I have time to investigate.

I made a habit out of running ceph and infrastructure deployment in parallel since they were independent of each other. In this case it lead to a race between environments/custom/playbook-wipe-partitions.yml and environments/custom/playbook-cinder-driver-dm-clone.yml. I have moved the latter into scripts/deploy/300-openstack-services.sh now.