controller: add disk tags to BlockDeviceSpec

[Vicente-Cheng]

Problem:
We want to add tags when we add extra disks.
(We need to add tags after we provision the disk )

Solution:
Add Tags to BlockDeviceSpec, then the controller can add it when provisioning

Related Issue:
harvester/harvester#2717

Test plan:

• Add disk tags with provision, e.g. (blockdevices CRD)

...
spec:
  devPath: /dev/sda
  fileSystem:
    forceFormatted: true
    mountPoint: ""
    provisioned: true
  nodeName: harvester-node-0
  tags:
  - default
  - ssd
...

• Check the nodes.longhorn.io. The related disk should have the same tags.
• Add/Remove tags with provisioned blockdevices

Tasks

• blockdevices controller should handle tag anytime
• add node watcher to sync tags
• add CI for it

[Vicente-Cheng]

Need new Longhorn deployment script, depends on harvester/vagrant-rancherd#8

[tserong]

Just to make sure I understand, this is meant to take tags defined via either harvester or longhorn, and add them to the block device spec, and then to the block device status. If so, it looks like it should generally do the right thing, I just have a couple of comments.

Commit 25afa25 ("controller: add disk tags to BlockDeviceSpec") does two things - it updates go.mod etc. with new/updated dependencies, and it also makes those changes to the controller. I'd suggest splitting that into two separate commits, one for the dependency update, and one for the main code change.

[Vicente-Cheng]

Just to make sure I understand, this is meant to take tags defined via either harvester or longhorn, and add them to the block device spec, and then to the block device status. If so, it looks like it should generally do the right thing, I just have a couple of comments.

Blockdevices is the CR for the harvester, longhorn directly managed the disk with longhorn.io/v1beta2/nodes CR.
So here we only focus on the harvester Blockdevices CR but any change in the longhorn.io/v1beta2/nodes related to harvester Blockdevices CR will be synced.

Commit 25afa25 ("controller: add disk tags to BlockDeviceSpec") does two things - it updates go.mod etc. with new/updated dependencies, and it also makes those changes to the controller. I'd suggest splitting that into two separate commits, one for the dependency update, and one for the main code change.

Good suggestions, I split another commit for it. That will make the review easier. Thanks for the suggestion.

[tserong]

LGTM

[tserong]

Oops, no, wait a sec - I read the code but didn't check the tests, one of which has failed: https://github.com/harvester/node-disk-manager/actions/runs/7112001697/job/19361613496?pr=60

[Vicente-Cheng]

Oops, no, wait a sec - I read the code but didn't check the tests, one of which has failed: https://github.com/harvester/node-disk-manager/actions/runs/7112001697/job/19361613496?pr=60

This failed because longhorn.io/v1beta2/nodes updated too frequently (means we will provision failed and queue again), CI does not know it. I am working on it.

[Vicente-Cheng]

Depends on harvester/go-common#8
I moved some general functions to the above repo. Once it is merged, I will update the PR again.

BTW, CI is also fixed now.

[Vicente-Cheng]

Hi @tserong, Please help recheck this PR.
Thanks!

[tserong]

LGTM!

[bk201]

LGTM, thanks!

[Vicente-Cheng]

@Mergifyio backport v0.6.x

[mergify]

backport v0.6.x

✅ Backports have been created

• #75 controller: add disk tags to BlockDeviceSpec (backport #60) has been created for branch v0.6.x