[FEATURE] v2 volume replica management

[innobead]

Is your feature request related to a problem? Please describe (👍 if you like this request)

Support managing replicas (add/remove/get/replace) of each RAID1 bdev and ensure non-disruptive IO during management if the volume is degraded.

Describe the solution you'd like

Have RPC commands for replica management. Each replica initialization still needs to manage by longhorn-spdk-engine.

Additional context

cc @longhorn/dev-data-plane

[DamiaSan]

There is a development under review to remove base bdev from RAID1:
https://review.spdk.io/gerrit/c/spdk/spdk/+/15267/35

[DamiaSan]

For the functionality to add a base bdev, we are waiting that the development will be put under code review: #5387
If this development will not be ready soon, we will proceed on our own

[innobead]

For the functionality to add a base bdev, we are waiting that the development will be put under code review: #5387 If this development will not be ready soon, we will proceed on our own

I transfer this issue to EPIC and include #5387 in this EPIC. I only see https://review.spdk.io/gerrit/c/spdk/spdk/+/15267, anything else? It seems pending for a while, I believe we should start the downstream first. This is mandatory for us in 1.5.0.

cc @shuo-wu @derekbit

[DamiaSan]

I agree, we will develop this functionality in downstream

[longhorn-io-github-bot]

Pre Ready-For-Testing Checklist

• Where is the reproduce steps/test steps documented?
  The documentation of the needed RAID1 functionality, like add and remove of a base bdev, can be found in the jsonrpc.md file
• Which areas/issues this PR might have potential impacts on?
  Area SPDK