Set cluster_kubeconfig based on the type of resource available in provider cluster only

ocs_ci/ocs/constants.py

@@ -357,6 +357,43 @@
  PROVIDER_CLIENT_DEPLOYMENT_DIR, "native_storage_client.yaml"
 )
 
+PROVIDER_CLUSTER_RESOURCE_KINDS = [
+ "cephblockpoolradosnamespaces",
+ "cephblockpoolradosnamespace",
+ "cephblockpools",
+ "cephblockpool",
+ "cephclients",
+ "cephclient",
+ "cephclusters",
+ "cephcluster",
+ "cephfilesystems",
+ "cephfilesystem",
+ "cephfilesystemsubvolumegroups",
+ "cephfilesystemsubvolumegroup",
+ "cephobjectstores",
+ "cephobjectstore",
+ "cephobjectstoreusers",
+ "cephobjectstoreuser",
+ "localvolumediscoveries",
+ "localvolumediscovery",
+ "localvolumediscoveryresults",
+ "localvolumediscoveryresult",
+ "localvolumes",
+ "localvolume",
+ "localvolumesets",
+ "localvolumeset",
+ "storageclusters",
+ "storagecluster",
+ "storageconsumers",
+ "storageconsumer",
+ "storageprofiles",
+ "storageprofile",
+ "storagerequests",
+ "storagerequest",
+ "storagesystems",
+ "storagesystem",
+]
+
 OCS_CLIENT_OPERATOR_CONTROLLER_MANAGER_PREFIX = "ocs-client-operator-controller-manager"
 OCS_CLIENT_OPERATOR_CONSOLE = "ocs-client-operator-console"
 STORAGE_CLIENT_NAME = "storage-client"

ocs_ci/ocs/ocp.py

@@ -81,10 +81,33 @@ def __init__(
  self._data = {}
  self.selector = selector
  self.field_selector = field_selector
- self.cluster_kubeconfig = cluster_kubeconfig
+ # In provider mode multicluster run, certain kind of resources are available in the provider cluster only.
+ # Setting cluster_kubeconfig of provider cluster in such cases will enable running "oc" commands seamlessly even
+ # when dealing with two instances of this class simultaneously despite the cluster context. This is achievable
+ # because all the methods use "exec_oc_cmd" method to run "oc" commmands. Primary cluster context being a
+ # client cluster, the test cases need not switch context to provider cluster before initializing a resource of
+ # the kinds listed in constants.PROVIDER_CLUSTER_RESOURCE_KINDS
+ if (
+ (not cluster_kubeconfig)
+ and config.multicluster
+ and config.ENV_DATA.get("odf_provider_mode_deployment", False)
+ and kind.lower() in constants.PROVIDER_CLUSTER_RESOURCE_KINDS
+ ):
+ provider_cluster_index = config.get_provider_index()
+ provider_kubeconfig_path = os.path.join(
+ config.clusters[provider_cluster_index].ENV_DATA["cluster_path"],
+ config.clusters[provider_cluster_index].RUN.get("kubeconfig_location"),
+ )
+ self.cluster_kubeconfig = provider_kubeconfig_path
+ # TODO : self.cluster_context = provider_cluster_index, remove cluster_kubeconfig check in if condition
+ else:
+ self.cluster_kubeconfig = cluster_kubeconfig
  self.threading_lock = threading_lock
  self.silent = silent
  self.skip_tls_verify = skip_tls_verify
+ # TODO: Set cluster_context based on the conditions of setting cluster_kubeconfig. Currently, setting
+ # cluster_context expects the current context to be the cluster where the resource is present.
+ # This cannot deal with simultaneous usage of two instances in two different clusters.
  self.cluster_context = config.cluster_ctx.MULTICLUSTER.get("multicluster_index")
 
  @property