Add ServerMaintenance CRD

docs/concepts/servermaintenance.md

@@ -0,0 +1,59 @@
+# ServerMaintenance
+
+`ServerMaintenance` represents a maintenance operation for a physical server. It transitions a `Server` from an 
+operational state (e.g., Available/Reserved) into a Maintenance state. Each `ServerMaintenance` object tracks the 
+lifecycle of a specific maintenance task, ensuring servers are properly taken offline, updated, and restored.
+
+## Key Points
+
+- `ServerMaintenance` is namespaced and can represent various maintenance types (e.g., BIOSUpdate, Cleanup).
+- Only one `ServerMaintenance` can be active for a given `Server` at a time. Others remain pending.
+- When the active `ServerMaintenance` completes, the next pending maintenance starts.
+- If no more maintenance tasks are pending, the `Server` returns to its previous operational state and can be 
+  powered back on.
+- The `metal-operator` manages `ServerMaintenance` resources and updates the `Server` state accordingly.
+- A maintenance-related operator (e.g., `firmware-operator`) decides if a `Server` needs maintenance, creates the 
+  `ServerMaintenance` resource, and a corresponding `ServerBootConfiguration`, and references it as a 
+  MaintenanceBootConfiguration in the `Server` spec. It also handles powering servers on/off.
+
+## Workflow
+
+1. **Determining Maintenance:**  
+   The maintenance operator creates a `ServerMaintenance` resource for the chosen `Server`.
+
+2. **Transition to Maintenance:**
+   The `metal-operator` notices the new `ServerMaintenance`, transitions the `Server` into `Maintenance` status.
+
+3. **Boot Configuration:**
+   The maintenance operator creates a `ServerBootConfiguration` resource and references it in the `Server` spec as the 
+   MaintenanceBootConfiguration. This configuration is used to e.g. boot custom tooling to perform BIOS/firmware updates
+   or run cleanup tasks on the `Server`.
+
+4. **Performing Maintenance:**
+   The maintenance operator powers off the `Server`, performs the required maintenance task, then updates the 
+   `ServerMaintenance` to `Complete`.
+
+5. **Post-Maintenance:**
+   Once complete, if no more maintenance tasks are pending, the `metal-operator` restores the `Server` to its previous
+   state (e.g., Available/Reserved). The maintenance operator can then power the `Server` back on.
+
+## Example ServerMaintenance Resource
+
+```yaml
+apiVersion: metal.ironcore.dev/v1alpha1
+kind: ServerMaintenance
+metadata:
+  name: bios-update
+  namespace: ops
+spec:
+  type: BIOSUpdate
+  serverRef:
+    name: server-foo
+status:
+  state: Pending
+```
+
+Once conditions are met, the `metal-operator` transitions the `Server` to `InMaintenance`. The maintenance operator 
+powers the server off, applies the maintenance boot configuration, performs the maintenance, marks `ServerMaintenance`
+as Complete, and then powers the server on. If multiple `ServerMaintenance` objects exist, the next pending one starts
+next; otherwise, the `Server` returns to its previous operational state.

mkdocs.yml

@@ -56,6 +56,7 @@ nav:
     - Servers: concepts/servers.md
     - ServerBootConfigurations: concepts/serverbootconfigurations.md
     - ServerClaims: concepts/serverclaims.md
+    - ServerMaintenance: concepts/servermaintenance.md
 - Usage:
   - metalctl: usage/metalctl.md
 - Development Guide: