Storage client deployment automation (#9144)

This pr contains the code changes for deployment of a provider cluster with ODF and native storage-client

Signed-off-by: Amrita Mahapatra <[email protected]>

conf/README.md

@@ -318,6 +318,9 @@ higher priority).
       * `hosted_odf_registry` - registry for hosted ODF
       * `hosted_odf_version` - version of ODF to be deployed on hosted clusters
 * `wait_timeout_for_healthy_osd_in_minutes` - timeout waiting for healthy OSDs before continuing upgrade (see https://bugzilla.redhat.com/show_bug.cgi?id=2276694 for more details)
+* `odf_provider_mode_deployment` - True if you would like to enable provider mode deployment.
+* `client_subcription_image` - ODF subscription image details for the storageclients.
+* `channel_to_client_subscription` - Channel value for the odf subscription image for storageclients.
 
 #### UPGRADE
 

conf/deployment/vsphere/ai_1az_rhcos_vsan_3m_3w.yaml

@@ -11,5 +11,5 @@ ENV_DATA:
   worker_num_cpus: '16'
   master_memory: '16384'
   compute_memory: '65536'
-  extra_disks: 2
+  extra_disks: 4
   fio_storageutilization_min_mbps: 10.0

ocs_ci/deployment/baremetal.py

@@ -1251,7 +1251,7 @@ def destroy_cluster(self, log_level="DEBUG"):
 
 
 @retry(exceptions.CommandFailed, tries=10, delay=30, backoff=1)
-def clean_disk(worker):
+def clean_disk(worker, namespace=constants.DEFAULT_NAMESPACE):
     """
     Perform disk cleanup
 
@@ -1262,41 +1262,51 @@ def clean_disk(worker):
     ocp_obj = ocp.OCP()
     cmd = """lsblk --all --noheadings --output "KNAME,PKNAME,TYPE,MOUNTPOINT" --json"""
     out = ocp_obj.exec_oc_debug_cmd(
-        node=worker.name, cmd_list=[cmd], namespace=constants.BM_DEBUG_NODE_NS
+        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_disk_to_ignore_cleanup = disk_to_ignore_cleanup["pkname"]
-            # Adding break when root disk is found
-            break
+            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"])
+            )
+
     out = ocp_obj.exec_oc_debug_cmd(
         node=worker.name,
         cmd_list=["lsblk -nd -e252,7 --output NAME --json"],
-        namespace=constants.BM_DEBUG_NODE_NS,
+        namespace=namespace,
     )
     lsblk_output = json.loads(str(out))
     lsblk_devices = lsblk_output["blockdevices"]
 
     for lsblk_device in lsblk_devices:
-        if lsblk_device["name"] == str(selected_disk_to_ignore_cleanup):
+        if lsblk_device["name"] in selected_disks_to_ignore_cleanup:
+            logger.info(f'the disk cleanup is ignored for, {lsblk_device["name"]}')
             pass
         else:
             logger.info(f"Cleaning up {lsblk_device['name']}")
             out = ocp_obj.exec_oc_debug_cmd(
                 node=worker.name,
                 cmd_list=[f"wipefs -a -f /dev/{lsblk_device['name']}"],
-                namespace=constants.BM_DEBUG_NODE_NS,
+                namespace=namespace,
             )
             logger.info(out)
             out = ocp_obj.exec_oc_debug_cmd(
                 node=worker.name,
                 cmd_list=[f"sgdisk --zap-all /dev/{lsblk_device['name']}"],
-                namespace=constants.BM_DEBUG_NODE_NS,
+                namespace=namespace,
             )
             logger.info(out)
 

ocs_ci/deployment/deployment.py

@@ -154,6 +154,7 @@
 from ocs_ci.utility.ibmcloud import run_ibmcloud_cmd
 from ocs_ci.deployment.cnv import CNVInstaller
 
+
 logger = logging.getLogger(__name__)
 
 
@@ -356,6 +357,11 @@ def do_deploy_ocs(self):
         Deploy OCS/ODF and run verification as well
 
         """
+        if config.ENV_DATA.get("odf_provider_mode_deployment", False):
+            logger.warning(
+                "Skipping normal ODF deployment because ODF deployment in Provider mode will be performed"
+            )
+            return
         if not config.ENV_DATA["skip_ocs_deployment"]:
             for i in range(config.nclusters):
                 if config.multicluster and (i in get_all_acm_indexes()):
@@ -579,6 +585,24 @@ def do_deploy_fusion(self):
             )
             time.sleep(30)
 
+    def do_deploy_odf_provider_mode(self):
+        """
+        Deploy ODF in provider mode and setup native client
+        """
+        # deploy provider-client deployment
+        from ocs_ci.deployment.provider_client.storage_client_deployment import (
+            ODFAndNativeStorageClientDeploymentOnProvider,
+        )
+
+        storage_client_deployment_obj = ODFAndNativeStorageClientDeploymentOnProvider()
+
+        # Provider-client deployment if odf_provider_mode_deployment: True
+        if (
+            config.ENV_DATA.get("odf_provider_mode_deployment", False)
+            and not config.ENV_DATA["skip_ocs_deployment"]
+        ):
+            storage_client_deployment_obj.provider_and_native_client_installation()
+
     def deploy_cluster(self, log_cli_level="DEBUG"):
         """
         We are handling both OCP and OCS deployment here based on flags
@@ -641,6 +665,7 @@ def deploy_cluster(self, log_cli_level="DEBUG"):
         self.do_deploy_oadp()
         self.do_deploy_rdr()
         self.do_deploy_fusion()
+        self.do_deploy_odf_provider_mode()
         if config.DEPLOYMENT.get("cnv_deployment"):
             CNVInstaller().deploy_cnv()
         if config.ENV_DATA.get("clusters"):