Added a flag, mark_masters_schedulable to mark master nodes scheduleable

ocs_ci/deployment/baremetal.py

@@ -1251,12 +1251,16 @@ def destroy_cluster(self, log_level="DEBUG"):
 
 
 @retry(exceptions.CommandFailed, tries=10, delay=30, backoff=1)
-def clean_disk(worker, namespace=constants.DEFAULT_NAMESPACE):
+def disks_available_to_cleanup(worker, namespace=constants.DEFAULT_NAMESPACE):
     """
-    Perform disk cleanup
+    disks available for cleanup
 
     Args:
         worker (object): worker node object
+        namespace (str): namespace where the oc_debug command will be executed
+
+    Returns:
+        disk_names_available_for_cleanup (list): The disk names available for cleanup on a node
 
     """
     ocp_obj = ocp.OCP()
@@ -1265,23 +1269,43 @@ def clean_disk(worker, namespace=constants.DEFAULT_NAMESPACE):
         node=worker.name, cmd_list=[cmd], namespace=namespace
     )
     disk_to_ignore_cleanup_raw = json.loads(str(out))
-    disk_to_ignore_cleanup_json = disk_to_ignore_cleanup_raw["blockdevices"]
-    selected_disks_to_ignore_cleanup = []
-    for disk_to_ignore_cleanup in disk_to_ignore_cleanup_json:
-        if disk_to_ignore_cleanup["mountpoint"] == "/boot":
-            logger.info(
-                f"Ignorning disk {disk_to_ignore_cleanup['pkname']} for cleanup because it's a root disk "
-            )
-            selected_disks_to_ignore_cleanup.append(
-                str(disk_to_ignore_cleanup["pkname"])
-            )
-        elif disk_to_ignore_cleanup["type"] == "rom":
-            logger.info(
-                f"Ignorning disk {disk_to_ignore_cleanup['kname']} for cleanup because it's a rom disk "
-            )
-            selected_disks_to_ignore_cleanup.append(
-                str(disk_to_ignore_cleanup["kname"])
-            )
+    disks_available = disk_to_ignore_cleanup_raw["blockdevices"]
+    boot_disks = set()
+    disks_available_for_cleanup = []
+    for disk in disks_available:
+        # First pass: identify boot disks and filter out ROM disks
+        if disk["type"] == "rom":
+            continue
+        if "nbd" in disk["kname"]:
+            continue
+        if disk["type"] == "part" and disk["mountpoint"] == "/boot":
+            boot_disks.add(disk["pkname"])
+        if disk["type"] == "disk":
+            disks_available_for_cleanup.append(disk)
+
+    # Second pass: filter out boot disks
+    disks_available_for_cleanup = [
+        disk for disk in disks_available_for_cleanup if disk["kname"] not in boot_disks
+    ]
+    disks_names_available_for_cleanup = [
+        disk["kname"] for disk in disks_available_for_cleanup
+    ]
+
+    return disks_names_available_for_cleanup
+
+
+@retry(exceptions.CommandFailed, tries=10, delay=30, backoff=1)
+def clean_disk(worker, namespace=constants.DEFAULT_NAMESPACE):
+    """
+    Perform disk cleanup
+
+    Args:
+        worker (object): worker node object
+        namespace (str): namespace where the oc_debug command will be executed
+
+    """
+    ocp_obj = ocp.OCP()
+    disks_available_on_worker_nodes_for_cleanup = disks_available_to_cleanup(worker)
 
     out = ocp_obj.exec_oc_debug_cmd(
         node=worker.name,
@@ -1292,7 +1316,7 @@ def clean_disk(worker, namespace=constants.DEFAULT_NAMESPACE):
     lsblk_devices = lsblk_output["blockdevices"]
 
     for lsblk_device in lsblk_devices:
-        if lsblk_device["name"] in selected_disks_to_ignore_cleanup:
+        if lsblk_device["name"] not in disks_available_on_worker_nodes_for_cleanup:
             logger.info(f'the disk cleanup is ignored for, {lsblk_device["name"]}')
             pass
         else:

ocs_ci/deployment/provider_client/storage_client_deployment.py

@@ -13,7 +13,7 @@
     check_phase_of_rados_namespace,
 )
 from ocs_ci.deployment.helpers.lso_helpers import setup_local_storage
-from ocs_ci.ocs.node import label_nodes, get_all_nodes, get_node_objs
+from ocs_ci.ocs.node import label_nodes, get_all_nodes, get_node_objs, get_nodes
 from ocs_ci.ocs.utils import (
     setup_ceph_toolbox,
     enable_console_plugin,
@@ -24,7 +24,7 @@
 )
 from ocs_ci.utility import templating, kms as KMS, version
 from ocs_ci.deployment.deployment import Deployment, create_catalog_source
-from ocs_ci.deployment.baremetal import clean_disk
+from ocs_ci.deployment.baremetal import clean_disk, disks_available_to_cleanup
 from ocs_ci.ocs.resources.storage_cluster import verify_storage_cluster
 from ocs_ci.ocs.resources.storage_client import StorageClient
 from ocs_ci.ocs.bucket_utils import check_pv_backingstore_type
@@ -78,6 +78,7 @@ def initial_function(self):
             namespace=config.ENV_DATA["cluster_namespace"],
         )
 
+        self.platform = config.ENV_DATA.get("platform").lower()
         self.deployment = Deployment()
         self.storage_clients = StorageClient()
 
@@ -95,13 +96,10 @@ def provider_and_native_client_installation(
         6. Disable ROOK_CSI_ENABLE_CEPHFS and ROOK_CSI_ENABLE_RBD
         7. Create storage profile
         """
-
-        # Allow ODF to be deployed on all nodes
         nodes = get_all_nodes()
         node_objs = get_node_objs(nodes)
-
-        log.info("labeling storage nodes")
-        label_nodes(nodes=node_objs, label=constants.OPERATOR_NODE_LABEL)
+        worker_node_objs = get_nodes(node_type=constants.WORKER_MACHINE)
+        no_of_worker_nodes = len(worker_node_objs)
 
         # Allow hosting cluster domain to be usable by hosted clusters
         path = "/spec/routeAdmission"
@@ -124,6 +122,31 @@ def provider_and_native_client_installation(
         wait_for_machineconfigpool_status(node_type="all")
         log.info("All the nodes are upgraded")
 
+        # Mark master nodes schedulable if mark_masters_schedulable: True
+        if config.ENV_DATA.get("mark_masters_schedulable", False):
+            path = "/spec/mastersSchedulable"
+            params = f"""[{{"op": "replace", "path": "{path}", "value": true}}]"""
+            assert self.scheduler_obj.patch(
+                params=params, format_type="json"
+            ), "Failed to run patch command to update control nodes as scheduleable"
+            # Allow ODF to be deployed on all nodes
+            log.info("labeling all nodes as storage nodes")
+            label_nodes(nodes=node_objs, label=constants.OPERATOR_NODE_LABEL)
+            worker_node_objs = get_nodes(node_type=constants.WORKER_MACHINE)
+            no_of_worker_nodes = len(worker_node_objs)
+        else:
+            log.info("labeling worker nodes as storage nodes")
+            label_nodes(nodes=worker_node_objs, label=constants.OPERATOR_NODE_LABEL)
+
+        disks_available_on_worker_nodes_for_cleanup = disks_available_to_cleanup(
+            worker_node_objs[0]
+        )
+        number_of_disks_available = len(disks_available_on_worker_nodes_for_cleanup)
+        log.info(
+            f"disks avilable for cleanup, {disks_available_on_worker_nodes_for_cleanup}"
+            f"number of disks avilable for cleanup, {number_of_disks_available}"
+        )
+
         # Install LSO, create LocalVolumeDiscovery and LocalVolumeSet
         is_local_storage_available = self.sc_obj.is_exist(
             resource_name=self.storageclass,
@@ -197,6 +220,15 @@ def provider_and_native_client_installation(
                 storage_cluster_data = self.add_encryption_details_to_cluster_data(
                     storage_cluster_data
                 )
+                storage_cluster_data["spec"]["storageDeviceSets"][0][
+                    "replica"
+                ] = no_of_worker_nodes
+
+                if self.platform in constants.HCI_PROVIDER_CLIENT_PLATFORMS:
+                    storage_cluster_data["spec"]["storageDeviceSets"][0][
+                        "count"
+                    ] = number_of_disks_available
+
                 templating.dump_data_to_temp_yaml(
                     storage_cluster_data, constants.OCS_STORAGE_CLUSTER_YAML
                 )
@@ -210,6 +242,14 @@ def provider_and_native_client_installation(
                 storage_cluster_data = self.add_encryption_details_to_cluster_data(
                     storage_cluster_data
                 )
+                storage_cluster_data["spec"]["storageDeviceSets"][0][
+                    "replica"
+                ] = no_of_worker_nodes
+
+                if self.platform in constants.HCI_PROVIDER_CLIENT_PLATFORMS:
+                    storage_cluster_data["spec"]["storageDeviceSets"][0][
+                        "count"
+                    ] = number_of_disks_available
                 templating.dump_data_to_temp_yaml(
                     storage_cluster_data, constants.OCS_STORAGE_CLUSTER_UPDATED_YAML
                 )

ocs_ci/ocs/resources/storage_client.py

@@ -16,6 +16,8 @@
 from ocs_ci.helpers.managed_services import (
     get_all_storageclassclaims,
 )
+from ocs_ci.ocs.resources.ocs import get_ocs_csv
+from ocs_ci.ocs.resources.storage_cluster import verify_storage_cluster
 from ocs_ci.utility.utils import TimeoutSampler
 
 log = logging.getLogger(__name__)
@@ -503,11 +505,25 @@ def verify_native_storageclient(self):
         storageclaims, associated storageclasses and storagerequests are created successfully.
 
         """
+        ocs_csv = get_ocs_csv()
+        client_csv_version = ocs_csv.data["spec"]["version"]
+        ocs_version = version.get_ocs_version_from_csv(only_major_minor=True)
+        log.info(
+            f"Check if OCS version: {ocs_version} matches with CSV: {client_csv_version}"
+        )
+        assert (
+            f"{ocs_version}" in client_csv_version
+        ), f"OCS version: {ocs_version} mismatch with CSV version {client_csv_version}"
         if self.ocs_version >= version.VERSION_4_16:
             namespace = config.ENV_DATA["cluster_namespace"]
         else:
             namespace = constants.OPENSHIFT_STORAGE_CLIENT_NAMESPACE
 
+        # Check ocs-storagecluster is in 'Ready' status
+        log.info("Verify storagecluster on Ready state")
+        verify_storage_cluster()
+
+        # Fetch storage-client name
         storageclient_name = self.get_storageclient_name(namespace)
 
         # Verify storageclient is in Connected status

ocs_ci/templates/ocs-deployment/provider-mode/ocs_storagecluster.yaml

@@ -35,7 +35,7 @@ spec:
   providerAPIServerServiceType: ClusterIP
   storageDeviceSets:
   - config: {}
-    count: 2
+    count: 4
     dataPVCTemplate:
       metadata: {}
       spec:
@@ -52,5 +52,5 @@ spec:
     name: local-storage-deviceset
     placement: {}
     preparePlacement: {}
-    replica: 6
+    replica: 3
     resources: {}

ocs_ci/templates/ocs-deployment/provider-mode/ocs_storagecluster_updated.yaml

@@ -33,7 +33,7 @@ spec:
   providerAPIServerServiceType: NodePort
   storageDeviceSets:
   - config: {}
-    count: 2
+    count: 4
     dataPVCTemplate:
       metadata: {}
       spec:
@@ -50,5 +50,5 @@ spec:
     name: local-storage-deviceset
     placement: {}
     preparePlacement: {}
-    replica: 6
+    replica: 3
     resources: {}