Rosa hcp node labeling (red-hat-storage#11024)

* label nodes for node stop and autoscaling

Signed-off-by: Daniel Osypenko <[email protected]>

conf/README.md

@@ -349,6 +349,7 @@ higher priority).
 * `continue_upgrade_after_checks_even_if_not_healthy` -  if set to true Rook will continue the OSD daemon upgrade process even if the PGs are not clean.
 * `upgrade_osd_requires_healthy_pgs` - If set to true OSD upgrade process won't start until PGs are healthy.
 * `workaround_mark_disks_as_ssd` - WORKAROUND: mark disks as SSD (not rotational - `0` in `/sys/block/*d*/queue/rotational`)
+* `node_labels` - Comma-separated labels to be applied to the nodes in the cluster, e.g. 'cluster.ocs.openshift.io/openshift-storage="",node-role.kubernetes.io/infra=""', default - empty string
 
 #### UPGRADE
 

conf/deployment/aws/rosa_hcp_1az_3w_m5.12x.yaml

@@ -26,3 +26,4 @@ ENV_DATA:
   ms_env_type: "staging"
   addon_name: "ocs-converged"
   persistent-monitoring: false
+  node_labels: cluster.ocs.openshift.io/openshift-storage=""

conf/deployment/aws/rosa_hcp_1az_6w_m5.2x.yaml

@@ -26,3 +26,4 @@ ENV_DATA:
   ms_env_type: "staging"
   addon_name: "ocs-converged"
   persistent-monitoring: false
+  node_labels: cluster.ocs.openshift.io/openshift-storage=""

ocs_ci/deployment/rosa.py

@@ -100,6 +100,11 @@ def deploy(self, log_level=""):
             machinepool_details.wait_replicas_ready(
                 target_replicas=config.ENV_DATA["worker_replicas"], timeout=1200
             )
+            if node_labels := config.ENV_DATA.get("node_labels"):
+                if machinepool_id := config.ENV_DATA.get("machine_pool"):
+                    rosa.label_nodes(
+                        self.cluster_name, machinepool_id, node_labels, rewrite=False
+                    )
 
         logger.info("generate kubeconfig and kubeadmin-password files")
         if config.ENV_DATA["ms_env_type"] == "staging":

ocs_ci/framework/conf/default_config.yaml

@@ -279,6 +279,9 @@ ENV_DATA:
   #RDR Green field
   rdr_osd_deployment_mode: "greenfield"
 
+  # Label nodes with specific labels, used for example fot ODF deployment on ROSA HCP
+  node_labels: ""
+
   # Assisted Installer related settings
 
 # This section is related to upgrade

ocs_ci/ocs/machinepool.py

@@ -137,6 +137,7 @@ class MachinePool:
     exist: bool = field(
         default=False
     )  # not a part of the data fetched from the cluster
+    labels: Dict[str, str] = field(default_factory=dict)
 
     def __post_init__(self):
         """Automatically populate fields by fetching machine pool details."""
@@ -173,8 +174,16 @@ def from_dict(cls, data: dict, cluster_name=None):
                 "id"
             ),  # this parameter is different in node_conf and data fetched from machinepool
             cluster_name=cluster_name,
+            labels=data.get("labels", {}),
         )
 
+    def refresh(self):
+        """Refresh the machine pool details."""
+        details = self.get_machinepool_details(self.cluster_name, self.machinepool_id)
+        if details:
+            self.__dict__.update(details.__dict__)
+            self.exist = True
+
     def get_machinepool_updated_replicas(self) -> Dict[str, int]:
         """
         Retrieve the number of replicas and current replicas for this machine pool.
@@ -463,10 +472,6 @@ def build_machinepool_cmd_base(cluster_name, node_conf, action):
             raise ValueError(
                 "When 'enable_autoscaling' is True, 'min_replicas' and 'max_replicas' are required."
             )
-    elif node_conf.get("replicas") is None:
-        raise ValueError(
-            "Parameter 'replicas' is required when autoscaling is disabled."
-        )
 
     cmd = f"rosa {action} machinepool --cluster {cluster_name} "
 

ocs_ci/utility/rosa.py

@@ -19,6 +19,7 @@
     ResourceWrongStatusException,
     TimeoutExpiredError,
 )
+from ocs_ci.ocs.machinepool import MachinePools, NodeConf
 from ocs_ci.utility import openshift_dedicated as ocm
 from ocs_ci.utility import utils
 
@@ -1112,3 +1113,41 @@ def get_associated_oidc_config_id(cluster_name):
         logger.warning(f"Failed to get OIDC config id: {proc.stderr.decode().strip()}")
         return ""
     return proc.stdout.decode().strip()
+
+
+def label_nodes(cluster_name, machinepool_id, labels, rewrite=False):
+    """
+    Label nodes of the given cluster.
+    ! Important
+    This method rewrites existing behavior of labeling nodes in the cluster, it appends the labels to the existing
+    labels, but not rewrite them. This prevents the issue of accidental overwriting the existing labels.
+
+    Args:
+        cluster_name (str): The cluster name
+        machinepool_id (str): The machinepool id
+        labels (str): The labels to apply
+        rewrite (bool): If True, rewrite the labels. False, otherwise.
+
+    Returns:
+        str: The output of the command
+    """
+    machine_pools = MachinePools(cluster_name)
+    machine_pool = machine_pools.filter(machinepool_id="workers", pick_first=True)
+    if not rewrite:
+        labels_dict = machine_pool.labels
+        logger.info(f"Existing labels: {labels_dict}")
+        # convert to comma separated string
+        if labels_dict:
+            labels = (
+                ",".join([f"{key}={value}" for key, value in labels_dict.items()])
+                + ","
+                + labels
+            )
+        else:
+            labels = labels
+    machine_pools.edit_machine_pool(
+        NodeConf(**{"machinepool_id": machinepool_id, "labels": labels}),
+        wait_ready=False,
+    )
+    machine_pool.refresh()
+    return machine_pool.labels