Prerequisites For Building a Driver
When contributing a new Flocker storage driver, you will need to consider the following prerequisites:
- Your driver needs support for storing metadata for each volume on the storage backend.
- The driver needs a way to programmatically map a compute instance ID to the input format expected by your storage backend for the attach operation.
For example, if you have a 2 node compute+storage cluster on AWS, and your storage solution refers to the compute nodes as
aws1
and aws2
, your driver running on aws1
would need to be able to find out its compute instance name as aws1
, not i-1cf275d9
(which is the EC2 naming convention).
- The driver needs a way to request default storage features like compression, data deduplication, IOPs, and SSD/HDD while creating a volume.
- Optional - Please consider adding driver logs for debugging.
How could we improve this page?