diff --git a/e2e/keywords/common.resource b/e2e/keywords/common.resource
index 642aaae18f..17e2902876 100644
--- a/e2e/keywords/common.resource
+++ b/e2e/keywords/common.resource
@@ -19,6 +19,7 @@ Library             ../libs/keywords/setting_keywords.py
 Library             ../libs/keywords/backupstore_keywords.py
 Library             ../libs/keywords/backup_keywords.py
 Library             ../libs/keywords/sharemanager_keywords.py
+Library             ../libs/keywords/k8s_keywords.py
 
 *** Keywords ***
 Set test environment
@@ -35,6 +36,7 @@ Set test environment
     END
 
 Cleanup test resources
+    uncordon_all_nodes
     cleanup_control_plane_network_latency
     reset_node_schedule
     cleanup_node_exec
diff --git a/e2e/keywords/host.resource b/e2e/keywords/host.resource
index c15ffe8f0d..0ad7047162 100644
--- a/e2e/keywords/host.resource
+++ b/e2e/keywords/host.resource
@@ -37,3 +37,8 @@ Restart cluster
 Power on off node
     Run keyword And Continue On Failure
     ...    power_on_node_by_name    ${powered_off_node}
+
+Power off node ${node_id}
+    ${powered_off_node} =     get_node_by_index    ${node_id}
+    power_off_node_by_name    ${powered_off_node}
+    Set Test Variable    ${powered_off_node}
\ No newline at end of file
diff --git a/e2e/keywords/k8s.resource b/e2e/keywords/k8s.resource
index 75ff214d4e..d553daef4e 100644
--- a/e2e/keywords/k8s.resource
+++ b/e2e/keywords/k8s.resource
@@ -5,6 +5,7 @@ Library             ../libs/keywords/k8s_keywords.py
 Library             ../libs/keywords/workload_keywords.py
 Library             ../libs/keywords/volume_keywords.py
 Library             ../libs/keywords/host_keywords.py
+Library             ../libs/keywords/node_keywords.py
 
 *** Variables ***
 
@@ -51,5 +52,62 @@ Force drain volume of ${workload_kind} ${workload_id} replica node
     Set Test Variable    ${drained_node}
     Set Test Variable    ${last_volume_node}
 
+Force drain node ${node_id}
+    ${node_name} =    get_node_by_index    ${node_id}
+    force_drain_node    ${node_name}
+
+Drain volume of ${workload_kind} ${workload_id} volume node
+    ${workload_name} =   generate_name_with_suffix    ${workload_kind}    ${workload_id}
+    ${volume_name} =    get_workload_volume_name    ${workload_name}
+    ${drained_node} =    get_volume_node    ${volume_name}
+    ${last_volume_node} =    get_volume_node    ${volume_name}
+    drain_node    ${drained_node}
+    wait_for_all_pods_evicted    ${drained_node}
+    Set Test Variable    ${drained_node}
+    Set Test Variable    ${last_volume_node}
+
 Uncordon the drained node
     uncordon_node    ${drained_node}
+
+Cordon ${workload_kind} ${workload_id} volume node
+    ${workload_name} =   generate_name_with_suffix    ${workload_kind}    ${workload_id}
+    ${volume_name} =    get_workload_volume_name    ${workload_name}
+    ${volume_node} =    get_volume_node    ${volume_name}
+    cordon_node    ${volume_node}
+    check_node_is_not_schedulable    ${volume_node}
+
+Force drain all nodes
+    FOR    ${node_id}    IN RANGE    0    3
+        ${node_name} =    get_node_by_index    ${node_id}
+        force_drain_node    ${node_name}
+        wait_for_all_pods_evicted    ${node_name}
+    END
+
+Check node ${node_id} cordoned
+    ${node_name} =    get_node_by_index    ${node_id}
+    check_node_cordoned    ${node_name}
+
+Force drain node ${node_id} and wait for ${duration} second    
+    ${drained_node} =    get_node_by_index    ${node_id}
+    ${instance_manager_name} =     get_instance_manager_on_node    ${drained_node}
+    ${drain_process} =    force_drain_node_and_wait    ${drained_node}    ${duration}
+    Set Test Variable    ${drain_process}
+    Set Test Variable    ${instance_manager_name}
+    Set Test Variable    ${drained_node}
+
+The drain process not completed
+    check_drain_process_not_completed    ${drain_process}
+
+The drain process completed
+    wait_for_all_pods_evicted    ${drained_node}
+    check_drain_process_completed    ${drain_process}
+
+Drain logs should contain
+    [Arguments]    ${log}
+    ${drain_logs} =     get_drain_process_error_log    ${drain_process}
+    Should Contain    ${drain_logs}    ${log}
+
+Check PDB not exist
+    [Arguments]    ${instance_manger}
+    check_instance_manager_pdb_not_exist    ${instance_manger}
+
diff --git a/e2e/keywords/longhorn.resource b/e2e/keywords/longhorn.resource
index 377a7e38a9..e774eba5ae 100644
--- a/e2e/keywords/longhorn.resource
+++ b/e2e/keywords/longhorn.resource
@@ -3,6 +3,7 @@ Documentation       Longhorn Keywords
 
 Library             ../libs/keywords/instancemanager_keywords.py
 Library             ../libs/keywords/workload_keywords.py
+Library             ../libs/keywords/k8s_keywords.py
 
 *** Variables ***
 @{longhorn_workloads}
@@ -43,3 +44,12 @@ Check Longhorn workload pods ${condition} annotated with ${key}
     Run Keyword IF    '${condition}' == 'not'    Should Not Be True    ${is_annotated}
     ...    ELSE IF    '${condition}' == 'is'    Should Be True    ${is_annotated}
     ...    ELSE    Fail    Invalid condition ${condition}
+
+Check instance-manager pod is not running on drained node    
+    ${pod} =     get_instance_manager_on_node    ${drained_node}
+    Should Be Equal    ${pod}    ${None}
+
+Check instance-manager pod is running on node ${node_id}    
+    ${node_name} =    get_node_by_index    ${node_id}
+    ${pod} =     get_instance_manager_on_node    ${node_name}
+    Should Not Be Equal    ${pod}    ${None}
\ No newline at end of file
diff --git a/e2e/keywords/volume.resource b/e2e/keywords/volume.resource
index 085188f074..1887b3b1fc 100644
--- a/e2e/keywords/volume.resource
+++ b/e2e/keywords/volume.resource
@@ -206,6 +206,12 @@ Wait volume ${volume_id} replica on node ${node_id} stopped
     ${node_name} =    get_node_by_index    ${node_id}
     wait_for_replica_stopped    ${volume_name}    ${node_name}
 
+Check volume ${volume_id} replica on node ${node_id} exist
+    ${volume_name} =    generate_name_with_suffix    volume    ${volume_id}
+    ${node_name} =    get_node_by_index    ${node_id}
+    ${replica_name}    get_replica_name_on_node    ${volume_name}    ${node_name}
+    Should Not Be Equal    ${replica_name}    ${None}
+
 Check volume ${volume_id} data is intact
     ${volume_name} =    generate_name_with_suffix    volume    ${volume_id}
     check_data_checksum    ${volume_name}
diff --git a/e2e/libs/k8s/k8s.py b/e2e/libs/k8s/k8s.py
index ea854b8fa6..4c90d1d4e8 100644
--- a/e2e/libs/k8s/k8s.py
+++ b/e2e/libs/k8s/k8s.py
@@ -1,6 +1,7 @@
 import time
 import subprocess
 import asyncio
+import os
 from kubernetes import client
 from workload.pod import create_pod
 from workload.pod import delete_pod
@@ -9,6 +10,10 @@
 from utility.utility import subprocess_exec_cmd
 from utility.utility import logging
 from utility.utility import get_retry_count_and_interval
+from utility.utility import check_popen_process_not_completed
+from utility.utility import check_popen_process_completed
+from utility.utility import get_popen_process_error_log
+from robot.libraries.BuiltIn import BuiltIn
 
 async def restart_kubelet(node_name, downtime_in_sec=10):
     manifest = new_pod_manifest(
@@ -71,3 +76,45 @@ def wait_all_pods_evicted(node_name):
         time.sleep(retry_interval)
 
     assert evicted, 'failed to evict pods'
+
+def check_node_cordoned(node_name):
+    api = client.CoreV1Api()
+    node = api.read_node(node_name)
+    assert node.spec.unschedulable is True, f"node {node_name} is not cordoned."
+
+def force_drain_node_and_wait(node_name, duration):
+    _, retry_interval = get_retry_count_and_interval()
+    exec_cmd = ["kubectl", "drain", node_name, "--force", "--ignore-daemonsets", "--delete-emptydir-data"]
+    drain_process = subprocess.Popen(exec_cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+
+    for i in range(int(duration)):
+        logging(f"Performing {exec_cmd} and wait...  counts={i}")
+        if drain_process.poll() is not None:
+            raise AssertionError(f"Drain node {node_name} completed, but it was expected not to complete!")
+        time.sleep(retry_interval)
+
+    return drain_process
+
+def check_drain_process_not_completed(drain_process):
+    check_popen_process_not_completed(drain_process)
+
+def check_drain_process_completed(drain_process):
+    check_popen_process_completed(drain_process)
+
+def get_drain_process_error_log(drain_process):
+    return get_popen_process_error_log(drain_process)
+
+def get_instance_manager_on_node(node_name):
+    data_engine = BuiltIn().get_variable_value("${DATA_ENGINE}")
+    pods = get_all_pods_on_node(node_name)
+    for pod in pods:
+        labels = pod.metadata.labels
+        if labels.get("longhorn.io/data-engine") == data_engine and \
+           labels.get("longhorn.io/component") == "instance-manager":
+            return pod.metadata.name
+    return None
+
+def check_instance_manager_pdb_not_exist(instance_manager):
+    exec_cmd = ["kubectl", "get", "pdb", "-n", "longhorn-system"]
+    res = subprocess_exec_cmd(exec_cmd)
+    assert instance_manager not in res.decode('utf-8')
diff --git a/e2e/libs/keywords/host_keywords.py b/e2e/libs/keywords/host_keywords.py
index bd4f6c7ef8..b2a4fe64c2 100644
--- a/e2e/libs/keywords/host_keywords.py
+++ b/e2e/libs/keywords/host_keywords.py
@@ -52,3 +52,6 @@ def power_off_volume_node(self, volume_name):
 
     def power_on_node_by_name(self, node_name):
         self.host.power_on_node(node_name)
+
+    def power_off_node_by_name(self, node_name):
+        self.host.power_off_node(node_name)
diff --git a/e2e/libs/keywords/k8s_keywords.py b/e2e/libs/keywords/k8s_keywords.py
index 625696a8d3..85834dfe23 100644
--- a/e2e/libs/keywords/k8s_keywords.py
+++ b/e2e/libs/keywords/k8s_keywords.py
@@ -5,8 +5,16 @@
 from k8s.k8s import drain_node, force_drain_node
 from k8s.k8s import cordon_node, uncordon_node
 from k8s.k8s import wait_all_pods_evicted
+from k8s.k8s import get_all_pods_on_node
+from k8s.k8s import check_node_cordoned
+from k8s.k8s import force_drain_node_and_wait
+from k8s.k8s import check_drain_process_not_completed
+from k8s.k8s import get_instance_manager_on_node
+from k8s.k8s import get_drain_process_error_log
+from k8s.k8s import check_instance_manager_pdb_not_exist
+from k8s.k8s import check_drain_process_completed
 from utility.utility import logging
-
+from node import Node
 
 class k8s_keywords:
 
@@ -50,5 +58,38 @@ def force_drain_node(self, node_name):
     def uncordon_node(self, node_name):
         uncordon_node(node_name)
 
+    def cordon_node(self, node_name):
+        cordon_node(node_name)
+
     def wait_for_all_pods_evicted(self, node_name):
         wait_all_pods_evicted(node_name)
+
+    def uncordon_all_nodes(self):
+        nodes = Node.list_node_names_by_role("worker")
+
+        for node_name in nodes:
+            uncordon_node(node_name)
+
+    def get_all_pods_on_node(self, node_name):
+        return get_all_pods_on_node(node_name)
+
+    def check_node_cordoned(self, node_name):
+        check_node_cordoned(node_name)
+
+    def force_drain_node_and_wait(self, node_name, duration):
+        return force_drain_node_and_wait(node_name, duration)
+
+    def check_drain_process_not_completed(self, drain_process):
+        return check_drain_process_not_completed(drain_process)
+
+    def check_drain_process_completed(self, drain_process):
+        return check_drain_process_completed(drain_process)
+
+    def get_instance_manager_on_node(self, node_name):
+        return get_instance_manager_on_node(node_name)
+
+    def get_drain_process_error_log(self, drain_process):
+        return get_drain_process_error_log(drain_process)
+
+    def check_instance_manager_pdb_not_exist(self, instance_manager):
+        return check_instance_manager_pdb_not_exist(instance_manager)
\ No newline at end of file
diff --git a/e2e/libs/keywords/node_keywords.py b/e2e/libs/keywords/node_keywords.py
index 74ef9055bf..bd53dcfdf6 100644
--- a/e2e/libs/keywords/node_keywords.py
+++ b/e2e/libs/keywords/node_keywords.py
@@ -40,5 +40,9 @@ def enable_node_scheduling(self, node_name):
 
     def reset_node_schedule(self):
         nodes = self.node.list_node_names_by_role("worker")
+
         for node_name in nodes:
             self.enable_node_scheduling(node_name)
+
+    def check_node_is_not_schedulable(self, node_name):
+        self.node.check_node_schedulable(node_name, schedulable="False")
diff --git a/e2e/libs/node/node.py b/e2e/libs/node/node.py
index 565fe0c971..dc1491986c 100644
--- a/e2e/libs/node/node.py
+++ b/e2e/libs/node/node.py
@@ -10,6 +10,8 @@
 from utility.utility import get_retry_count_and_interval
 from utility.utility import logging
 
+from k8s.k8s import uncordon_node
+
 class Node:
 
     DEFAULT_DISK_PATH = "/var/lib/longhorn/"
@@ -159,3 +161,11 @@ def set_default_disk_scheduling(self, node_name, allowScheduling):
             if disk.path == self.DEFAULT_DISK_PATH:
                 disk.allowScheduling = allowScheduling
         self.update_disks(node_name, node.disks)
+
+    def check_node_schedulable(self, node_name, schedulable):
+        node = get_longhorn_client().by_id_node(node_name)
+        for _ in range(self.retry_count):
+            if node["conditions"]["Schedulable"]["status"] == schedulable:
+                break
+            time.sleep(self.retry_interval)
+        assert node["conditions"]["Schedulable"]["status"] == schedulable
diff --git a/e2e/libs/utility/utility.py b/e2e/libs/utility/utility.py
index 8f3521bb6b..c1e1e469c7 100644
--- a/e2e/libs/utility/utility.py
+++ b/e2e/libs/utility/utility.py
@@ -285,3 +285,25 @@ def get_name_suffix(*args):
         if arg:
             suffix += f"-{arg}"
     return suffix
+
+
+def check_popen_process_not_completed(process):
+    assert process.poll() is None, f"process {process} terminated which is not expected."
+
+
+def check_popen_process_completed(process):
+    retry_count, retry_interval = get_retry_count_and_interval()
+    for i in range(retry_count):
+        if process.poll() is not None:
+            return
+        time.sleep(retry_interval)
+    assert process.poll() is not None, f"process {process} not terminated which is not expected."
+
+
+def get_popen_process_error_log(drain_process):
+    if drain_process.poll() is None:
+        drain_process.terminate()
+
+    stdout, stderr = drain_process.communicate()
+    logging(f"{stderr.decode('utf-8')}")
+    return stderr.decode('utf-8')
diff --git a/e2e/requirements.txt b/e2e/requirements.txt
index 69b127c977..4a345f1cd3 100644
--- a/e2e/requirements.txt
+++ b/e2e/requirements.txt
@@ -6,4 +6,4 @@ kubernetes==27.2.0
 requests==2.32.3
 boto3==1.35.19
 pyyaml==6.0.2
-minio==5.0.10
+minio==5.0.10
\ No newline at end of file
diff --git a/e2e/tests/negative/node_drain.robot b/e2e/tests/negative/node_drain.robot
index e220ceb845..df993f0b01 100644
--- a/e2e/tests/negative/node_drain.robot
+++ b/e2e/tests/negative/node_drain.robot
@@ -10,6 +10,10 @@ Resource    ../keywords/k8s.resource
 Resource    ../keywords/deployment.resource
 Resource    ../keywords/workload.resource
 Resource    ../keywords/setting.resource
+Resource    ../keywords/longhorn.resource
+Resource    ../keywords/volume.resource
+Resource    ../keywords/host.resource
+Resource    ../keywords/node.resource
 
 Test Setup    Set test environment
 Test Teardown    Cleanup test resources
@@ -89,3 +93,142 @@ Force Drain Replica Node While Replica Rebuilding
         And Wait for deployment 1 pods stable
         And Check deployment 1 data in file data.txt is intact
     END
+
+Drain node with force
+    [Documentation]    Drain node with force
+    ...    1. Deploy a cluster contains 3 worker nodes N1, N2, N3.
+    ...    2. Deploy Longhorn.
+    ...    3. Create a 1-replica deployment with a 3-replica Longhorn volume. The volume is attached to N1.
+    ...    4. Write some data to the volume and get the md5sum.
+    ...    5. Force drain and remove N2, which contains one replica only.
+    ...        kubectl drain <Node name> --delete-emptydir-data=true --force=true --grace-period=-1 --ignore-daemonsets=true --timeout=<Desired timeout in secs>
+    ...    6. Wait for the volume Degraded.
+    ...    7. Force drain and remove N1, which is the node the volume is attached to.
+    ...        kubectl drain <Node name> --delete-emptydir-data=true --force=true --grace-period=-1 --ignore-daemonsets=true --timeout=<Desired timeout in secs>
+    ...
+    ...    8. Wait for the volume detaching then being recovered. Will get attached to the workload/node.
+    ...    9. Verify the instance manager pods are gone and not recreated after the drain.
+    ...    10. Validate the volume content. The data is intact.
+    Given Create storageclass longhorn-test with    dataEngine=${DATA_ENGINE}
+    And Create persistentvolumeclaim 0 using RWO volume with longhorn-test storageclass
+    And Create deployment 0 with persistentvolumeclaim 0
+    And Wait for volume of deployment 0 healthy
+    And Write 2048 MB data to file data.txt in deployment 0
+
+    And Force drain volume of deployment 0 replica node
+    Then Wait for volume of deployment 0 attached to the original node and degraded
+
+    And Force drain volume of deployment 0 volume node
+    And Wait for volume of deployment 0 attached to another node and degraded
+    And Check instance-manager pod is not running on drained node    dataEngine=${DATA_ENGINE}
+    Then Check deployment 0 data in file data.txt is intact
+
+Drain node without force
+    [Documentation]    Drain node without force
+    ...    1. Cordon the node. Longhorn will automatically disable the node scheduling when a Kubernetes node is cordoned.
+    ...    2. Evict all the replicas from the node.
+    ...    3. Run the following command to drain the node with force flag set to false.
+    ...        kubectl drain <Node name> --delete-emptydir-data --force=false --grace-period=-1 --ignore-daemonsets=true --timeout=<Desired timeout in secs>
+    ...    4. Observe that the workloads move to another node. The volumes should first detach and attach to workloads once they move to another node.
+    ...    5. One by one all the pods should get evicted.
+    ...    6. Verify the instance manager pods are gone and not recreated after the drain.
+    Given Create storageclass longhorn-test with    dataEngine=${DATA_ENGINE}
+    And Create persistentvolumeclaim 0 using RWO volume with longhorn-test storageclass
+    And Create deployment 0 with persistentvolumeclaim 0
+    And Wait for volume of deployment 0 healthy
+    And Write 2048 MB data to file data.txt in deployment 0
+
+    When Cordon deployment 0 volume node
+    And Delete replica of deployment 0 volume on volume node
+    And Drain volume of deployment 0 volume node
+    And Wait for volume of deployment 0 attached to another node and degraded
+    And Check instance-manager pod is not running on drained node
+    Then Check deployment 0 data in file data.txt is intact
+
+Test kubectl drain nodes for PVC/PV/LHV is created through Longhorn API
+    [Documentation]    Test kubectl drain nodes for PVC/PV/LHV is created through Longhorn API
+    ...    Given 1 PVC/PV/LHV created through Longhorn API And LHV is not yet attached/replicated.
+    ...    When kubectl drain nodes.
+    ...        NODE=centos-worker-0
+    ...        kubectl cordon ${NODE}
+    ...        kubectl drain --force --ignore-daemonsets --delete-emptydir-data --grace-period=10 ${NODE}
+    ...    Then all node should successfully drain.
+    When Create volume 0 with   dataEngine=${DATA_ENGINE}
+    And Wait for volume 0 detached
+    And Create persistentvolume for volume 0
+    And Create persistentvolumeclaim for volume 0
+    And Force drain all nodes
+
+Stopped replicas on deleted nodes should not be counted as healthy replicas when draining nodes
+    [Documentation]    Stopped replicas on deleted nodes should not be counted as healthy replicas when draining nodes
+    ...    When draining a node, the node will be set as unscheduled and all pods should be evicted.
+    ...    By Longhorn’s default settings, the replica will only be evicted if there is another healthy replica on the running node.
+    ...    Related Issue:
+    ...    - https://github.com/longhorn/longhorn/issues/2237
+    ...
+    ...    Given Longhorn with 2 nodes cluster: Node_1, Node_2
+    ...    And Update the Node Drain Policy setting to block-if-contains-last-replica
+    ...    And Create a 5Gi detached volume with 2 replicas.
+    ...    And Stop Node_1 that contains one of the replicas.
+    ...    When Attempts to drain Node_2 that contains remaining replica.
+    ...        kubectl drain <Node_2 name> --delete-emptydir-data=true --force=true --grace-period=-1 --ignore-daemonsets=true
+    ...    Then The Node_2 becomes cordoned.
+    ...    And All pods on Node_2 are evicted except the replica instance manager pod.
+    ...        kubectl get pods --field-selector spec.nodeName=<Node_2 name> -o wide -n longhorn-system
+    ...    And The message like below keeps appearing.
+    ...        evicting pod longhorn-system/instance-manager-r-xxxxxxxx
+    ...        error when evicting pods/"instance-manager-r-xxxxxxxx" -n "longhorn-system" (will retry after 5s): Cannot evict pod as it would violate the pod's disruption budget.
+    ...    And The last healthy replica exists on the Node_2.
+    [Teardown]    Cleanup test resources include off nodes
+    Given Disable node 0 scheduling
+    And Set setting node-drain-policy to block-if-contains-last-replica
+    And Given Create volume 0 with    size=5Gi    numberOfReplicas=2    dataEngine=${DATA_ENGINE}
+    And Attach volume 0 to node 1
+    And Wait for volume 0 healthy
+    And Write data to volume 0
+    And Detach volume 0 from attached node
+    And Wait for volume 0 detached
+    And Power off node 1
+
+    When Force drain node 2 and wait for 90 second
+    And The drain process not completed
+    And Check instance-manager pod is running on node 2
+    And Drain logs should contain    log=error when evicting pods/\"${instance_manager_name}
+    And Check volume 0 replica on node 2 exist
+
+Setting Allow Node Drain with the Last Healthy Replica protects the last healthy replica with Pod Disruption Budget (PDB)
+    [Documentation]    Setting Allow Node Drain with the Last Healthy Replica protects the last healthy replica with Pod Disruption Budget (PDB) 
+    ...    Related Issue:
+    ...    - https://github.com/longhorn/longhorn/issues/2237
+    ...
+    ...    Given Longhorn with 2 nodes cluster: Node_1, Node_2
+    ...    And Update the Node Drain Policy setting to block-if-contains-last-replica and confirm with following command:
+    ...        kubectl get settings.longhorn.io/node-drain-policy -n longhorn-system
+    ...    And Create a 5Gi detached volume with 2 replicas.
+    ...    And Stop Node_1 that contains one of the replicas.
+    ...    And Drain Node_2 so that all pods on Node_2 are evicted, but the replica instance manager pod is still on Node_2 because it is protected by PDB.
+    ...        kubectl drain <Node_2 name> --delete-emptydir-data=true --force=true --grace-period=-1 --ignore-daemonsets=true
+    ...    And Update the Node Drain Policy setting to always-allow and confirm with following command:
+    ...        kubectl get settings.longhorn.io/node-drain-policy -n longhorn-system
+    ...    Then The pod longhorn-system/instance-manager-r-xxxxxxxx will be evicted successfully and the following command can be used to ensure that only daemonset pods such as engine-image, longhorn-csi-plugin and longhorn-manager daemonset pods are running on Node_2:
+    ...        kubectl get pods --field-selector spec.nodeName=<Node_2 name> -o wide -n longhorn-system
+    ...    And The PDB will be deleted and can be verified with the following command:
+    ...        kubectl get pdb <replica name, e.g., instance-manager-r-xxxxxxxx> -n longhorn-system
+    [Teardown]    Cleanup test resources include off nodes
+    Given Disable node 0 scheduling
+    And Set setting node-drain-policy to block-if-contains-last-replica
+    And Given Create volume 0 with    size=5Gi    numberOfReplicas=2        dataEngine=${DATA_ENGINE}
+    And Attach volume 0 to node 1
+    And Wait for volume 0 healthy
+    And Write data to volume 0
+    And Detach volume 0 from attached node
+    And Wait for volume 0 detached
+    And Power off node 1
+
+    When Force drain node 2 and wait for 90 second
+    And The drain process not completed
+    And Check instance-manager pod is running on node 2
+
+    When Set setting node-drain-policy to always-allow
+    And The drain process completed
+    And Check PDB not exist    instance_manger=${instance_manager_name}