When to consider Automation
If you deploy new Management or Workload clusters on a daily or weekly basis, you should probably consider how to use PowerShell, PowerCLI or Terraform to enforce consistency and eliminate human errors in deployment and destruction of such clusters.
These techniques are discused in the Awesome SolidFire repo.
Tools: Ansible, Powershell, Terraform
Considering that ez-rancher employs Terraform and there are both SolidFire (“ElementSW”) and vSphere Providers for Terraform, for this I’d prefer Terraform followed by PowerShell and only then Ansible.
But you can choose any approach that works for you. If you have any of these already in place, that would obviously be a factor.
You can find examples of SolidFire storage provisioning with Terraform on YouTube.
Awesome SolidFire has additional details on SolidFire automation.
Automation
Almost everything can be automated, but a lot of it isn’t specific to NetApp HCI or SolidFire. We’ll cover only the parts that need our consideration. One-off actions (such as the creation of a Rancher Resource Pool) may be covered at a later time.
- Deploy
- Create dedicated storage account and deploy NetApp Trident to User Cluster
- Create VMware networks, datastores
- Different automation for vSphere clusters with VDS vs. VSS
- Rancher Management Cluster has different network requirements from User (Workload) Cluster
- Update
- Modify storage QoS policy (“retype”) and back-end QoS settings in NetApp Trident
- Expand volume size of Trident or VMs
- Create additional datastores or networks
- Destroy
- Undo the creation of resources from Deploy, including Trident storage account if used by a User Cluster
The latter two categories are covered in Automation and Monitoring.