Decommission a node in FlashArray
This guide describes a recommended workflow for decommissioning a Portworx node in your cluster.
Migrate application pods using Portworx volumes that are running on this node
If you plan to remove Portworx from a node, applications running on that node using Portworx need to be migrated. If Portworx is not running, existing application containers will end up with read-only volumes and new ones will fail to start.
Perform the following steps to migrate select pods.
-
Cordon the node using the following command:
kubectl cordon <node> -
Reschedule application pods using Portworx volumes on different nodes:
kubectl delete pod <pod-name> -n <application-namespace>Since application pods are expected to be managed by a controller like
DeploymentorStatefulSet, Kubernetes will spin up a new replacement pod on another node.
Remove Portworx from the cluster
Follow this section "Removing offline Nodes" or "Removing a functional node from a cluster" to decommission the Portworx node from the cluster.
Removing a functional node from a cluster
A functional Portworx node may need to be removed from the cluster. In this section, we'll demonstrate the removal of a node by running commands on the node itself as well as the removal of a node from another node.
The following output shows the state of the cluster and the different node IPs and node IDs:
pxctl status
Status: PX is operational
Node ID: xxxxxxxx-xxxx-xxxx-xxxx-0027f6bbcbd1
IP: X.X.X.0
Local Storage Pool: 1 pool
POOL IO_PRIORITY SIZE USED STATUS ZONE REGION
0 LOW 64 GiB 11 GiB Online c us-east-1
Local Storage Devices: 1 device
Device Path Media Type Size Last-Scan
0:1 /dev/xvdf STORAGE_MEDIUM_SSD 64 GiB 25 Feb 17 21:13 UTC
total - 64 GiB
Cluster Summary
Cluster ID: xxxxxxxx-xxxx-xxxx-xxxx-d4a612b74cc3
IP ID Used Capacity Status
172.31.40.38 xxxxxxxx-xxxx-xxxx-xxxx-2bc112f5f131 11 GiB 64 GiB Online
172.31.37.211 xxxxxxxx-xxxx-xxxx-xxxx-a85e0514ae8b 11 GiB 64 GiB Online
172.31.35.130 xxxxxxxx-xxxx-xxxx-xxxx-893373631483 11 GiB 64 GiB Online
172.31.45.106 xxxxxxxx-xxxx-xxxx-xxxx-2eeddcd64d51 11 GiB 64 GiB Online
172.31.45.56 xxxxxxxx-xxxx-xxxx-xxxx-ec8e1420e645 11 GiB 64 GiB Online
172.31.46.119 xxxxxxxx-xxxx-xxxx-xxxx-0027f6bbcbd1 11 GiB 64 GiB Online (This node)
172.31.39.201 xxxxxxxx-xxxx-xxxx-xxxx-936b1b58aa24 11 GiB 64 GiB Online
172.31.33.151 xxxxxxxx-xxxx-xxxx-xxxx-41e70a72eafd 11 GiB 64 GiB Online
172.31.33.252 xxxxxxxx-xxxx-xxxx-xxxx-428e727eb6b8 11 GiB 64 GiB Online
Global Storage Pool
Total Used : 99 GiB
Total Capacity : 576 GiB
Suspend active cloudsnap operations
-
Identify any active cloudsnap operations being run on the node that you intend to decommission:
pxctl cloudsnap statusThe
STATEof active operations shows asBackup-Active:NAME SOURCEVOLUME STATE NODE TIME-ELAPSED COMPLETED
xxxxxxxx-xxxx-xxxx-xxxx-278535e49860 885345022234521857 Backup-Done 10.13.90.125 39.746191264s Tue, 22 Mar 2022 22:53:37 UTC
xxxxxxxx-xxxx-xxxx-xxxx-3c3ccf47f276 186701534582547510 Backup-Done 10.13.90.122 1.677455484s Tue, 22 Mar 2022 23:59:49 UTC
xxxxxxxx-xxxx-xxxx-xxxx-73176c2d03e2 885345022234521857 Backup-Done 10.13.90.125 27.550329395s Wed, 23 Mar 2022 00:00:15 UTC
xxxxxxxx-xxxx-xxxx-xxxx-2307865c1b93 649554470078043771 Backup-Active 10.13.90.125 5m12.61653365sFrom this output, identify the volumes with active backups. For example, if node 10.13.90.125 is being decommissioned, then the volume with active backup is
649554470078043771. -
Identify the namespace of the volume the cloudsnap operation is occuring on. The namespace is displayed under the
Labelssection in the output from the following command. Replace<source_volume>with theSOURCEVOLUMEvalue for the volume that is in aBackup-Activestate from the previous output:pxctl volume inspect <source_volume>Volume : 649554470078043771
Name : pvc-xxxxxxxx-xxxx-xxxx-xxxx-d4827680f6de
Size : 500 GiB
Format : ext4
HA : 3
IO Priority : LOW
Creation time : Mar 22 20:37:52 UTC 2022
Shared : v4 (service)
Status : up
State : Attached: xxxxxxxx-xxxx-xxxx-xxxx-10f4e076cac8 (10.13.90.119)
Last Attached : Mar 22 20:37:58 UTC 2022
Device Path : /dev/pxd/pxd649554470078043771
Labels : mount_options=nodiscard=true,namespace=vdbench,nodiscard=true,pvc=vdbench-pvc-sharedv4,repl=3,sharedv4=true,sharedv4_svc_type=ClusterIP
... -
Suspend backup operations for the volume and wait for current backup to complete:
storkctl suspend volumesnapshotschedule vdbench-pvc-sharedv4-schedule -n vdbench -
Verify the suspension. The
SUSPENDfield will show astrue:storkctl get volumesnapshotschedule -n vdbenchNAME PVC POLICYNAME PRE-EXEC-RULE POST-EXEC-RULE RECLAIM-POLICY SUSPEND LAST-SUCCESS-TIME
vdbench-pvc-sharedv4-schedule vdbench-pvc-sharedv4 testpolicy Delete true 22 Mar 22 17:10 PDT
Repeat these steps until all active snaps complete and all backup operations are suspended on the node that you want to decommission.
Prevention of data loss
If any node hosts a volume with replication factor of 1, then Portworx disallows decommissioning of such nodes because there is data loss.
One possible workaround to decommission such a node is to increase the replication of single replica volumes by running volume ha-update.
-
List all the volumes hosts on the decommisioning node:
pxctl volume list --node xxxxxxxx-xxxx-xxxx-xxxx-2eeddcd64d51ID NAME SIZE HA SHARED ENCRYPTED PROXY-VOLUME IO_PRIORITY STATUS SNAP-ENABLED
633738568577538909 pvc-xxxxxxxx-xxxx-xxxx-xxxx-d4827680f6de 2 GiB 3 no no no LOW up - attached on 172.31.45.106 no
161898313715947409 pvc-xxxxxxxx-xxxx-xxxx-xxxx-68f0d970e10c 2 GiB 1 no no no LOW up - attached on 172.31.45.106 no -
Increase the replication factor:
pxctl volume ha-update --repl 2 161898313715947409
Once the volume is completely replicated onto another node, continue with the node decommissioning. This time, the volume already has another replica on another node, so decommissioning the node will reduce the replication factor of the volume and remove the node.
Placing the node in maintenance mode
After identifying the node to be removed (see "Identify the node that you want to remove from the cluster" above), place the node in maintenance mode.
Log in to the node to be decommissioned:
pxctl service maintenance --enter
This is a disruptive operation, PX will restart in maintenance mode.
Are you sure you want to proceed ? (Y/N): y
Entered maintenance mode.
Run the cluster delete command
Example 1: Running the cluster delete command from a different node
ssh to 172.31.46.119 and run the following command:
pxctl cluster delete xxxxxxxx-xxxx-xxxx-xxxx-2eeddcd64d51
Node xxxxxxxx-xxxx-xxxx-xxxx-2eeddcd64d51 successfully deleted.
Example 2: Running the cluster delete command from the same node
ssh to 172.31.33.252 and type:
pxctl cluster delete xxxxxxxx-xxxx-xxxx-xxxx-428e727eb6b8
Node xxxxxxxx-xxxx-xxxx-xxxx-428e727eb6b8 successfully deleted.
Clean up Portworx metadata on the node
To learn how to remove or clean up Portworx metadata on the decommmisioned node, see clean up Portworx metadata on the node.
Removing offline nodes
This document describes how to remove an offline node from a cluster. If you are specifically interested in decommissioning nodes with no storage that have been offline for an extended period, see Automatic decommission of storageless nodes.
-
Identify the cluster that needs to be managed:
pxctl statusStatus: PX is operational
Node ID: xxxxxxxx-xxxx-xxxx-xxxx-3e2b01cd0bc3
IP: X.X.X.197
Local Storage Pool: 2 pools
Pool IO_Priority Size Used Status Zone Region
0 LOW 200 GiB 1.0 GiB Online default default
1 LOW 120 GiB 1.0 GiB Online default default
Local Storage Devices: 2 devices
Device Path Media Type SizLast-Scan
0:1 /dev/mapper/volume-27dbb728 STORAGE_MEDIUM_SSD 200 GiB 08 Jan 17 05:39 UTC
1:1 /dev/mapper/volume-0a31ef46 STORAGE_MEDIUM_SSD 120 GiB 08 Jan 17 05:39 UTC
total - 320 GiB
Cluster Summary
Cluster ID: xxxxxxxx-xxxx-xxxx-xxxx-0242ac110002
Node IP: X.X.X.197 - Capacity: 2.0 GiB/320 GiB Online (This node)
Node IP: 10.99.117.129 - Capacity: 1.2 GiB/100 GiB Online
Node IP: 10.99.119.1 - Capacity: 1.2 GiB/100 GiB Online
Global Storage Pool
Total Used : 4.3 GiB
Total Capacity : 520 GiB -
List the nodes in the cluster:
pxctl cluster listCluster ID: xxxxxxxx-xxxx-xxxx-xxxx-0242ac110002
Status: OK
Nodes in the cluster:
ID DATA IP CPU MEM TOTAL MEM FREE CONTAINERS VERSION STATUS
xxxxxxxx-xxxx-xxxx-xxxx-3e2b01cd0bc3 X.X.X.197 1.629073 8.4 GB 7.9 GB N/A 1.1.2-c27cf42 Online
xxxxxxxx-xxxx-xxxx-xxxx-dd5084dce208 10.XX.117.129 0.125156 8.4 GB 8.0 GB N/A 1.1.3-b33d4fa Online
xxxxxxxx-xxxx-xxxx-xxxx-d3478c485a61 10.99.119.1 0.25 8.4 GB 8.0 GB N/A 1.1.3-b33d4fa Online -
List the volumes in the cluster:
pxctl volume listID NAME SIZE HA SHARED ENCRYPTED PRIORITSTATUS
845707146523643463 testvol 1 GiB 1 no no LOW up - attached on X.X.X.197In this case, there is one volume in the cluster and it is attached to node with IP X.X.X.97
-
Identify the node that you want to remove from the cluster. Skip to next step if the node you want to remove is already offline.
To manually set a node offline, apply px/enabled=remove label to node:
kubectl label nodes < node > px/enabled=remove --overwrite -
In the following example, node X.X.X.197 has been marked as offline:
pxctl cluster listCluster ID: xxxxxxxx-xxxx-xxxx-xxxx-0242ac110002
Status: OK
Nodes in the cluster:
ID DATA IP CPU MEM TOTAL MEM FREE CONTAINERS VERSION STATUS
xxxxxxxx-xxxx-xxxx-xxxx-dd5084dce208 10.99.117.129 5.506884 8.4 GB 8.0 GB N/A 1.1.3-b33d4fa Online
xxxxxxxx-xxxx-xxxx-xxxx-d3478c485a61 10.99.119.1 0.25 8.4 GB 8.0 GB N/A 1.1.3-b33d4fa Online
xxxxxxxx-xxxx-xxxx-xxxx-3e2b01cd0bc3 X.X.X.197 - - N/A 1.1.2-c27cf42 Offline -
Attach and detach the volume in one of the surviving nodes:
pxctl host attach 845707146523643463Volume successfully attached at: /dev/pxd/pxd845707146523643463pxctl host detach 845707146523643463Volume successfully detached -
Delete the local volume that belonged to the offline node:
pxctl volume delete 845707146523643463Volume 845707146523643463 successfully deleted. -
Delete the node that is offline:
pxctl cluster delete xxxxxxxx-xxxx-xxxx-xxxx-3e2b01cd0bc3Node xxxxxxxx-xxxx-xxxx-xxxx-3e2b01cd0bc3 successfully deleted. -
List the nodes in the cluster to make sure that the node is removed:
pxctl cluster listCluster ID: xxxxxxxx-xxxx-xxxx-xxxx-0242ac110002
Status: OK
Nodes in the cluster:
ID DATA IP CPU MEM TOTAL MEM FREE CONTAINERS VERSION STATUS
xxxxxxxx-xxxx-xxxx-xxxx-dd5084dce208 10.99.117.129 4.511278 8.4 GB 8.0 GB N/A 1.1.3-b33d4fa Online
xxxxxxxx-xxxx-xxxx-xxxx-d3478c485a61 10.99.119.1 0.500626 8.4 GB 8.0 GB N/A 1.1.3-b33d4fa Online -
Show the cluster status:
pxctl statusStatus: PX is operational
Node ID: xxxxxxxx-xxxx-xxxx-xxxx-dd5084dce208
IP: X.X.X.199
Local Storage Pool: 1 pool
Pool IO_Priority Size Used Status Zone Region
0 LOW 100 GiB 1.2 GiB Online default default
Local Storage Devices: 1 device
Device Path Media Type Size Last-Scan
0:1 /dev/mapper/volume-9f6be49c STORAGE_MEDIUM_SSD 100 GiB08 Jan 17 06:34 UTC
total - 100 GiB
Cluster Summary
Cluster ID: xxxxxxxx-xxxx-xxxx-xxxx-0242ac110002
Node IP: 10.99.117.129 - Capacity: 1.2 GiB/100 GiB Online (This node)
Node IP: 10.99.119.1 - Capacity: 1.2 GiB/100 GiB Online
Global Storage Pool
Total Used : 2.3 GiB
Total Capacity : 200 GiB
Automatic decommission of storageless nodes
Storageless nodes that are initialized and added to the cluster may not be needed once they complete their tasks (such as in a scheduler workflow). If they are taken offline or destroyed, the cluster will still retain the nodes and mark them as offline.
If eventually a majority of such nodes exist, the cluster won't have a quorum of nodes that are online. The solution is to run cluster delete commands and remove such nodes. This gets more laborious with more such nodes or an increased frequency of such nodes added and taken down.
To help with this, Portworx waits until a grace period of 48 hours has passed. After this period, offline nodes with no storage will be removed from the cluster. There is no CLI command needed to turn on or trigger this feature.
Remove Portworx installation from the node
Apply the px/enabled=remove label and it will remove the existing Portworx systemd service. It will also apply the px/enabled=false label to stop Portworx from running in future.
For example, below command will remove existing Portworx installation from minion2 and also ensure that Portworx pod doesn’t run there in future.
kubectl label nodes < node > px/enabled=remove --overwrite
Decommission from Kubernetes: If the plan is to decommission this node altogether from the Kubernetes cluster, no further steps are needed.
Ensure application pods using Portworx don’t run on this node
If you need to continue using the node without Portworx, you will need to ensure your application pods using Portworx volumes don’t get scheduled here.
You can ensure this by adding the schedulerName: stork field to your application specs (Deployment, Statefulset, etc). Stork is a scheduler extension that will schedule pods using Portworx PVCs only on nodes that have Portworx running. Refer to the Using scheduler convergence article for more information.
Another way to achieve this is to use inter-pod affinity
-
You need to define a pod affinity rule in your applications that ensure that application pods get scheduled only on nodes where the Portworx pod is running.
-
Consider the following nginx example:
apiVersion: apps/v1
kind: Deployment
metadata:
name: nginx-deployment
spec:
selector:
matchLabels:
app: nginx
replicas: 1
template:
metadata:
labels:
app: nginx
spec:
affinity:
# Inter-pod affinity rule restricting nginx pods to run only on nodes where Portworx pods are running (Portworx pods have a label
# name=portworx which is used in the rule)
podAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- labelSelector:
matchExpressions:
- key: name
operator: In
values:
- "portworx"
topologyKey: kubernetes.io/hostname
namespaces:
- "kube-system"
hostNetwork: true
containers:
- name: nginx
image: nginx
ports:
- containerPort: 80
volumeMounts:
- name: nginx-persistent-storage
mountPath: /usr/share/nginx/html
volumes:
- name: nginx-persistent-storage
persistentVolumeClaim:
claimName: px-nginx-pvcIt can also be configured on cluster level by adding
spec.stork.args.webhook-controllerSet totruein StorageCluster to make Stork the default scheduler for workloads using Portworx volumes:apiVersion: core.libopenstorage.org/v1
kind: StorageCluster
metadata:
name: portworx
namespace: portworx
spec:
stork:
enabled: true
args:
webhook-controller: true
Uncordon the node
You can now uncordon the node using: kubectl uncordon <node>
If you want to permanently decommission the node, you can skip the following step.
(Optional) Rejoin node to the cluster
If you want Portworx to start again on this node and join as a new node, follow the node rejoin steps.
Disconnect or delete drives from your FlashArray
If you are using Pure FlashArray as a cloud drive provider, follow the additional instructions to decommission FlashArray nodes.