To show the process of applying resource on resourcebinding status.

[Patrick0308]

What would you like to be added:
To show the process of applying resource on resourcebinding status.

New Status Field:

currentResourceVersion: the current resource version need to be applied.
appliedResourceVersion：the full applied success resource version.
aggregatedStatus[].appliedResourceVersion: the applied resource version on a cluster.

FullyApplied Condition Reason

FullyApplied: Applying current resource version successfully. Status is true.
FullyAppliedFailed: Applying failed. Status is false.
Applying: In process of applying current resource version. Status is false

Example

if in process of applying resource:

  currentResourceVersion: "123"
  appliedResourceVersion: "122"
  phase: Applying
  aggregatedStatus:
  - applied: false
    phase: Applying
    clusterName: cluster1
    appliedResourceVersion: "122"
    health: Unknown
  - applied: false
    phase: Applying
    clusterName: cluster2
    appliedResourceVersion: "122"
    health: Unknown
  conditions:
  - lastTransitionTime: "2024-06-03T06:42:29Z"
    message: Binding has been scheduled successfully.
    reason: Success
    status: "True"
    type: Scheduled
  - lastTransitionTime: "2024-06-03T06:42:29Z"
    message: Applying resource on cluster1, cluster2
    reason: Applying
    status: "False"
    type: FullyApplied
  schedulerObservedGeneration: 7

if a cluster applied failed:

  currentResourceVersion: "123"
  appliedResourceVersion: "122"
  phase: ApplyFailed
  aggregatedStatus:
  - applied: true
    phase: Applied
    clusterName: cluster1
    appliedResourceVersion: "123"
    health: Unknown
  - appliedMessage: "error message"
    phase: ApplyFailed
    clusterName: cluster2
    appliedResourceVersion: "122"
    health: Unknown
  conditions:
  - lastTransitionTime: "2024-06-03T06:42:29Z"
    message: Binding has been scheduled successfully.
    reason: Success
    status: "True"
    type: Scheduled
  - lastTransitionTime: "2024-06-03T06:42:29Z"
    message: error message
    reason: FullyAppliedFailed
    status: "False"
    type: FullyApplied
  schedulerObservedGeneration: 7

if all clusters applied success:

  currentResourceVersion: "123"
  appliedResourceVersion: "123"
  phase: Applied
  aggregatedStatus:
  - applied: true
    phase: Applied
    clusterName: cluster1
    appliedResourceVersion: "123"
    health: Unknown
  - applied: true
    phase: Applied
    clusterName: cluster2
    appliedResourceVersion: "123"
    health: Unknown
  conditions:
  - lastTransitionTime: "2024-06-03T06:42:29Z"
    message: Binding has been scheduled successfully.
    reason: Success
    status: "True"
    type: Scheduled
  - lastTransitionTime: "2024-06-03T06:42:29Z"
    message: All works have been successfully applied
    reason: FullyAppliedSuccess
    status: "True"
    type: FullyApplied
  schedulerObservedGeneration: 7

Why is this needed:

Although the client updated resource, and checked FullyApplied condition status which is true, the client not sure that new resource was applied to each cluster successfully. Because maybe that's in process of applying.

[grosser]

I like how thorough this is, but I also think a very minimal version could be good enough (the extra features like seeing it per cluster and current version could be nice for debugging in detail though)

• we don't really need currentResourceVersion since that is in metadata.resourceVersion
• we don't really need aggregatedStatus since we could set the version once all clusters are reconciled

[Patrick0308]

we don't really need currentResourceVersion since that is in metadata.resourceVersion.

To get FullyCondition status need current resource version information. However I just found resourcebinding has a current ResourceVersion already which is spec.resource.resourceVersion. So there don't need status.currentResourceVersion

we don't really need aggregatedStatus since we could set the version once all clusters are reconciled.

I wanna show clusters applied progress. I will add more details.

[grosser]

if you need per cluster then this design makes sense 👍