def external_post_deploy_validation(self):
"""
This function validates successful deployment of OCS
in external mode, some of the steps overlaps with
converged mode
"""
cephcluster = CephClusterExternal()
cephcluster.cluster_health_check(timeout=300)
def __init__(self):
"""
Initializer for Sanity class - Init CephCluster() in order to
set the cluster status before starting the tests
"""
self.pvc_objs = list()
self.pod_objs = list()
self.ceph_cluster = CephClusterExternal()
def setup(
self,
request,
scenario,
num_of_nodes,
num_of_fail_nodes,
disrupt_provisioner,
project_factory,
multi_pvc_factory,
dc_pod_factory,
):
"""
Identify the nodes and start DeploymentConfig based app pods using
PVC with ReadWriteOnce (RWO) access mode on selected nodes
Args:
scenario (str): Scenario of app pods running on OCS or dedicated nodes
(eg., 'colocated', 'dedicated')
num_of_nodes (int): number of nodes required for running test
num_of_fail_nodes (int): number of nodes to make unresponsive during test
disrupt_provisioner (bool): True to disrupt the leader provisioner
pods if not running on selected nodes, else False
project_factory: A fixture to create new project
multi_pvc_factory: A fixture create a set of new PVCs
dc_pod_factory: A fixture to create deploymentconfig pods
Returns:
tuple: containing the params used in test cases
"""
ocs_nodes, non_ocs_nodes = self.identify_and_add_nodes(
scenario, num_of_nodes)
test_nodes = ocs_nodes if (scenario == "colocated") else non_ocs_nodes
logger.info(f"Using nodes {test_nodes} for running test")
def finalizer():
helpers.remove_label_from_worker_node(node_list=test_nodes,
label_key="nodetype")
# Check ceph health
ceph_health_check(tries=40)
request.addfinalizer(finalizer)
project = project_factory()
if helpers.storagecluster_independent_check():
ceph_cluster = CephClusterExternal()
else:
ceph_cluster = CephCluster()
# Wait for mon pods to reach expected count
# Bug 1778273 - [RFE]: Configure 5 MONs for OCS cluster with 5 or more nodes
# This wait is required for some of the previous OCS versions (< 4.5)
current_mon_count = int(
ceph_cluster.CEPHCLUSTER.get_resource(resource_name="",
column="MONCOUNT"))
assert ceph_cluster.POD.wait_for_resource(
condition=constants.STATUS_RUNNING,
selector=constants.MON_APP_LABEL,
resource_count=current_mon_count,
timeout=900,
)
ceph_cluster.mons = []
ceph_cluster.scan_cluster()
# Select nodes for running app pods and inducing network failure later
app_pod_nodes = self.select_nodes_for_app_pods(scenario, ceph_cluster,
ocs_nodes,
non_ocs_nodes,
num_of_fail_nodes)
# Create multiple RBD and CephFS backed PVCs with RWO accessmode
num_of_pvcs = self.num_of_app_pods_per_node * num_of_fail_nodes
rbd_pvcs = multi_pvc_factory(
interface=constants.CEPHBLOCKPOOL,
project=project,
size=self.pvc_size,
access_modes=[constants.ACCESS_MODE_RWO],
num_of_pvc=num_of_pvcs,
)
cephfs_pvcs = multi_pvc_factory(
interface=constants.CEPHFILESYSTEM,
project=project,
size=self.pvc_size,
access_modes=[constants.ACCESS_MODE_RWO],
num_of_pvc=num_of_pvcs,
)
# Create deploymentconfig based pods
dc_pods = []
# Start app-pods on selected node(s)
for node_name in app_pod_nodes:
logger.info(f"Starting app pods on the node {node_name}")
helpers.label_worker_node(node_list=[node_name],
label_key="nodetype",
label_value="app-pod")
for num in range(self.num_of_app_pods_per_node):
dc_pods.append(
dc_pod_factory(
interface=constants.CEPHBLOCKPOOL,
pvc=rbd_pvcs.pop(0),
node_selector={"nodetype": "app-pod"},
))
assert pod.verify_node_name(
dc_pods[-1], node_name
), f"Pod {dc_pods[-1].name} is not running on labeled node {node_name}"
dc_pods.append(
dc_pod_factory(
interface=constants.CEPHFILESYSTEM,
pvc=cephfs_pvcs.pop(0),
node_selector={"nodetype": "app-pod"},
))
assert pod.verify_node_name(
dc_pods[-1], node_name
), f"Pod {dc_pods[-1].name} is not running on labeled node {node_name}"
helpers.remove_label_from_worker_node(node_list=[node_name],
label_key="nodetype")
# Label other test nodes to be able to run app pods later
helpers.label_worker_node(node_list=test_nodes,
label_key="nodetype",
label_value="app-pod")
# Get ceph mon,osd pods running on selected node if colocated scenario
# and extra OCS nodes are present
# Recovery steps for MON and OSDS not required from OCS 4.4 onwards
# Refer to BZ 1830015 and BZ 1835908
ceph_pods = []
if float(config.ENV_DATA["ocs_version"]) < 4.4 and (
scenario == "colocated" and len(test_nodes) > 3):
pods_to_check = ceph_cluster.osds
# Skip mon pods if mon_count is 5 as there may not be enough nodes
# for all mons to run after multiple node failures
if ceph_cluster.mon_count == 3:
pods_to_check.extend(ceph_cluster.mons)
for pod_obj in pods_to_check:
if pod.get_pod_node(pod_obj).name in app_pod_nodes[0]:
ceph_pods.append(pod_obj)
logger.info(
f"Colocated Mon, OSD pods: {[pod_obj.name for pod_obj in ceph_pods]}"
)
disruptor = []
if disrupt_provisioner:
disruptor = self.disrupt_plugin_provisioner_pods(app_pod_nodes)
return ceph_cluster, dc_pods, ceph_pods, app_pod_nodes, test_nodes, disruptor