> ## Content Index > Fetch the complete content index at: https://linuxhandbook.com/llms.txt > Use this file to discover other available public pages before exploring further. # How to Drain a Node in Kubernetes - URL: https://linuxhandbook.com/kubectl-drain-node/ - Published: 2020-10-14T10:15:03.000Z - Updated: 2022-10-13T15:44:47.000Z - Description: Learn how to properly drain nodes in Kubernetes to prepare for maintainance. - Author: Rakesh Jain - Tags: Using Kubernetes In this [Kubernetes tutorial](https://linuxhandbook.com/tag/kubernetes/), you will learn to drain a node using kubectl drain command to prepare for maintenance. It is as simple as entering this command: ``` kubectl drain node_name ``` You can [get the nodes details](https://linuxhandbook.com/read-nodes-kubernetes/) using `kubectl get nodes` command. But there is more to draining nodes in [Kubernetes](https://kubernetes.io/?ref=linuxhandbook.com) so let's take a detailed look at it. ## Why do you need to drain nodes? Kubernetes is designed to be fault tolerant of worker node failures. There might be different reasons a worker node becomes unusable such as because of a hardware problem, a cloud provider problem, or if there are network issues between worker and master node, the Kubernetes master handles it effectively. But that doesn’t mean it will always be the case. And this is when you need to drain the nodes and remove all the pods. The draining is the process for safely evicting all the pods from a node. This way, the containers running on the pod terminate gracefully. ## How to properly drain nodes in Kubernetes Let's start with the practical demonstration. ### Step 1: Mark the node as unschedulable (cordon) To perform maintenance on a node, you should unschedule and then drain a node. First have a look at the currently running nodes: ```shell root@kmaster-rj:~# kubectl get nodes NAME STATUS ROLES AGE VERSION kmaster-rj Ready master 44d v1.18.8 kworker-rj1 Ready 44d v1.18.8 kworker-rj2 Ready 44d v1.18.8 root@kmaster-rj:~# ``` Look at the pods running on different nodes: ```shell root@kmaster-rj:~# kubectl get pods -o wide NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES my-dep-557548758d-gprnr 1/1 Running 1 4d23h 172.16.213.48 kworker-rj1 my-dep-557548758d-d2pmd 1/1 Running 1 4d15h 172.16.213.57 kworker-rj2 pod-delete-demo 1/1 Running 1 2d 172.16.213.56 kworker-rj1 root@kmaster-rj:~# ``` Now mark the node as unschedulable by running the following command: ```shell root@kmaster-rj:~# kubectl cordon kworker-rj2 node/kworker-rj2 cordoned root@kmaster-rj:~# ``` List the nodes again: ```shell root@kmaster-rj:~# kubectl get nodes NAME STATUS ROLES AGE VERSION kmaster-rj Ready master 44d v1.18.8 kworker-rj1 Ready 44d v1.18.8 kworker-rj2 Ready,SchedulingDisabled 44d v1.18.8 root@kmaster-rj:~# ``` You can notice that the node kworker-rj2 is now labeled as SchedulingDisabled. Till this step it doesn't evict the pods running on that node. Verify the pod status: ```shell root@kmaster-rj:~# kubectl get pods -o wide NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES my-dep-557548758d-gprnr 1/1 Running 1 4d23h 172.16.213.48 kworker-rj1 my-dep-557548758d-d2pmd 1/1 Running 1 4d15h 172.16.213.57 kworker-rj2 pod-delete-demo 1/1 Running 1 2d 172.16.213.56 kworker-rj1 root@kmaster-rj:~# ``` You can see that pod "my-dep-557548758d-d2pmd" still running on kworker-rj2 node. [ ![Linux Foundation Kubernetes Certification Discount](https://linuxhandbook.com/content/images/2022/10/lf-foundation-kubernetes.webp) ](https://linuxhandbook.com/training/kubernetes-fundamentals) ### Step 2: Drain the node to prepare for maintenance Now drain the node in preparation for maintenance to remove pods that are running on the node by running the following command: ```shell root@kmaster-rj:~# kubectl drain kworker-rj2 --grace-period=300 --ignore-daemonsets=true node/kworker-rj2 already cordoned WARNING: ignoring DaemonSet-managed Pods: kube-system/calico-node-fl8dl, kube-system/kube-proxy-95vdf evicting pod default/my-dep-557548758d-d2pmd pod/my-dep-557548758d-d2pmd evicted node/kworker-rj2 evicted root@kmaster-rj:~# ``` **NOTE:** kubectl drain cannot delete Pods not managed by ReplicationController, ReplicaSet, Job, DaemonSet or StatefulSet. You need to use --force to override that and by doing that the individual pods will be deleted permanently. Now look at the pods: ```shell root@kmaster-rj:~# kubectl get pods -o wide NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES my-dep-557548758d-gprnr 1/1 Running 1 4d23h 172.16.213.48 kworker-rj1 my-dep-557548758d-dsanh 1/1 Running 0 27s 172.16.213.38 kworker-rj1 pod-delete-demo 1/1 Running 1 2d 172.16.213.56 kworker-rj1 root@kmaster-rj:~# ``` The pod which was running on kworker-rj2 node evicted from there and started as a new pod on kworker-rj1 node. Nodes status remains the same: ```shell root@kmaster-rj:~# kubectl get nodes NAME STATUS ROLES AGE VERSION kmaster-rj Ready master 44d v1.18.8 kworker-rj1 Ready 44d v1.18.8 kworker-rj2 Ready,SchedulingDisabled 44d v1.18.8 root@kmaster-rj:~# ``` ### Step 3: Uncordon the node after maintenance completes You need to run following command afterwards to tell Kubernetes that it can resume scheduling new pods onto the node. ```bash root@kmaster-rj:~# kubectl uncordon kworker-rj2 node/kworker-rj2 uncordoned ``` Verify the node status: ```bash root@kmaster-rj:~# kubectl get nodes NAME STATUS ROLES AGE VERSION kmaster-rj Ready master 44d v1.18.8 kworker-rj1 Ready 44d v1.18.8 kworker-rj2 Ready 44d v1.18.8 ``` Node kworker-rj2 becomes ready again to handle new workloads. I hope you like this quick tip about draining nodes in Kubernetes.