Easy Way to Schedule Pod Restart in Kubernetes

Photo by Ian Taylor on Unsplash

Introduction

In this post I will show how to easily restart pods in Kubernetes using CronJobs. We will use CronJob, not to run our pods, but to schedule a Kubernetes API command that will restart our deployment once day with kubectl rollout restart , if something goes wrong, the old pods will not be down or removed. Rollout will be create new ReplicaSets, and will wait for them to be up, before killing off old pods, and rerouting the traffic. Service will continue uninterrupted. To fully do this we will have to setup RBAC, so that the Kubernetes client running from inside the cluster has permissions to do needed calls to the Kubernetes API.

Let’s Start

Create Service Account

kind: ServiceAccount
apiVersion: v1
metadata:
  name: restart-deploy
  namespace: testns

Create Role

apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:
  name: restart-deployment
  namespace: testns
rules:
  - apiGroups: ["apps", "extensions"]
    resources: ["deployments"]
    resourceNames: ["test-pod"]
    verbs: ["get", "patch", "list", "watch"]

Create RoleBinding

apiVersion: rbac.authorization.k8s.io/v1
kind: RoleBinding
metadata:
  name: restart-deployment
  namespace: prod
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: Role
  name: restart-deployment
subjects:
  - kind: ServiceAccount
    name: restart-deployment
    namespace: testns

Create CronJob

apiVersion: batch/v1
kind: CronJob
metadata:
  name: restart-deployment
  namespace: testns
spec:
  successfulJobsHistoryLimit: 1
  failedJobsHistoryLimit: 2
  concurrencyPolicy: Forbid
  schedule: '0 * * * *'
  jobTemplate:
    spec:
      backoffLimit: 2 
      activeDeadlineSeconds: 600 
      template:
        spec:
          serviceAccountName: restart-deployment
          restartPolicy: Never
          containers:
            - name: kubectl
              image: bitnami/kubectl 
              command:
                - 'kubectl'
                - 'rollout'
                - 'restart'
                - 'deployment/test'

If we want the cronjob to wait for the deployment to roll out, we must change change the command of cronjob:

apiVersion: batch/v1
kind: CronJob
metadata:
  name: restart-deployment
  namespace: testns
spec:
  successfulJobsHistoryLimit: 1
  failedJobsHistoryLimit: 2
  concurrencyPolicy: Forbid
  schedule: '0 * * * *'
  jobTemplate:
    spec:
      backoffLimit: 2 
      activeDeadlineSeconds: 600 
      template:
        spec:
          serviceAccountName: restart-deployment
          restartPolicy: Never
          containers:
            - name: kubectl
              image: bitnami/kubectl
              command:
                - bash
                - -c
                - >-
                  kubectl rollout restart deployment/test &&
                  kubectl rollout status deployment/test

You can download full file from this link.