![\"\"](\"https:\/\/www.dbi-services.com\/blog\/wp-content\/uploads\/sites\/2\/2024\/03\/kured-300x292.png\")
<\/figure>\n\n\n\nIn the world of DevOps, where continuous development and deployment are essential, ensuring the smooth operation of a Kubernetes cluster is crucial. One key aspect is managing Kubernetes node reboots efficiently and safely. This is where Kured (KUbernetes REboot Daemon) steps in. Kured is a Kubernetes daemonset that automates the process of node reboots based on indications from the underlying operating system’s package management system. In this blog, we will explore how Kured significantly improves DevOps project downtime by simplifying the management of node reboots.<\/p>\n\n\n\n
CNCF<\/a> sandbox project, Kured<\/a> is a set of Kubernetes objects. And like any good tool, you can install Kured using Helm<\/a> for an easy installation (Yes, also through standard manifests file if you like to play).<\/p>\n\n\n\n Our sandbox environment is composed of two AWS EC2 instances, running an RKE2-backed Kubernetes cluster. If you are not familiar with RKE2<\/a>, this is a fully compliant Kubernetes distribution, with a specific focus on security. RKE stands for Rancher Kubernetes Engine and is a CNCF certified Kubernetes distribution<\/a>. Interested into some RKE2 content? My colleague Kevin posted some time ago the procedure<\/a> to setup nodes autoscaling in RKE2. <\/p>\n\n\n Here, we are dealing with a quite simple cluster, with a control-plane node and, aside from that, a worker node. Adding Kured to your cluster is pretty straightforward, thanks to Helm.<\/p>\n\n\n The Helm repository is added first; a Pay attention to the following:<\/p>\n\n\n\n As said in the documentation, Kured is checking regularly for the presence of the file I created my cluster some while ago and would require for sure some system upgrades. Let’s install Kured and see how it behaves.<\/p>\n\n\n\n First, we need to get the value file and update it. <\/p>\n\n\n I’m going to update the file to set the Installing Kured is just a matter of \nubuntu@k8s-node-master:~$ kubectl get nodes\nNAME STATUS ROLES AGE VERSION\nk8s-node-master Ready control-plane,etcd,master 5h49m v1.26.12+rke2r1\nk8s-node-worker Ready <none> 5h27m v1.26.12+rke2r1\n<\/pre><\/div>\n\n\n
\nubuntu@k8s-node-master:~$ helm repo add kubereboot https:\/\/kubereboot.github.io\/charts\n\nubuntu@k8s-node-master:~$ helm search repo kubereboot\nNAME CHART VERSION APP VERSION DESCRIPTION \nkubereboot\/kured 5.4.2 1.15.0 A Helm chart for kured\n<\/pre><\/div>\n\n\n
helm search<\/code> shows the latest available chart and app version. But before installing it, you can quickly have a look at some of them. There are dozens of fields you can customize, and a standard helm show values kubereboot\/kured <\/code>should do the trick! I’m pasting here the URL values.yaml<\/a> file, from GitHub. But the output of the helm command should be similar, according to the version you’ve selected.<\/p>\n\n\n\nKey<\/strong><\/td> Short description<\/strong><\/td><\/tr> configuration.startTime and configuration.endTime<\/td> Indicates the time window for rebooting.<\/td><\/tr> configuration.rebootDays<\/td> Reboot will be only performed in the selected days.<\/td><\/tr> configuration.period<\/td> Defines how often Kured needs to check for reboot.<\/td><\/tr> configuration.forceReboot<\/td> What Kured needs to do if drain times out?<\/td><\/tr> configuration.drainGracePeriod
configuration.drainPodSelector
configuration.drainTimeout<\/td>Drain related parameters. By setting this, you defines a grace period (remember the kubectl delete pods --force --grace-period=0<\/code> \ud83d\ude00 ), the list of pods to drain, and a timeout after which the drain will be canceled.<\/td><\/tr>configuration.blockingPodSelector<\/td> If certain pods are present on the node, the reboot will be canceled.<\/td><\/tr> configuration.concurrency<\/td> Can be useful for very large clusters. By default 1, meaning only one node will be rebooted at a time.<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n \/var\/run\/reboot-required<\/code>. Created by the package manager, this file comes along with the file \/var\/run\/reboot-required.pkgs<\/code>, which lists all the packages that requested the reboot. We can tune in the Helm chart values the name and location, but for Debian\/Ubuntu-like Linux distributions, no modifications should be necessary.<\/p>\n\n\n\nLet’s start!<\/h2>\n\n\n\n
\nubuntu@k8s-node-master:~$ helm show values kubereboot\/kured > values.yaml\nubuntu@k8s-node-master:~$ ls -l\ntotal 8\n-rw-rw-r-- 1 ubuntu ubuntu 6412 Mar 4 09:56 values.yaml\n<\/pre><\/div>\n\n\n
configuration.period<\/code> key. For the purpose of my tests, I don’t want to wait for an hour \ud83d\ude09 so I will shorten it.<\/p>\n\n\n\nhelm install<\/code>.<\/p>\n\n\n