Skip to content

Latest commit

 

History

History
102 lines (65 loc) · 3.9 KB

Kubernetes-1-GettingStarted.md

File metadata and controls

102 lines (65 loc) · 3.9 KB
Raw

Kubernetes 101 - Getting started

0 - Installing and configuring kubectl

The primary source for the Hands-On session is the official kubernetes documentation.

For getting started first the commandline application kubectl must be installed. A detailed installation guideline can be found here. While not strictly necessary it is recommended to configure bash/zsh/powershell autocompletion.

Login to the Kubermatic WebUI here with your LifeScience AAI login. Download the KubeConfig file and move it to the location ~/.kube/config.

Test the configuration by running: kubectl get pods -n <namespace>. A working configuration is indicated by a message like this: No resources found in <namespace> namespace.

1 - Running simple pods via YAML

Pods are the most basic resource for running workloads in a kubernetes cluster. Every resource for kubernetes can be described as YAML file. For example a simple pod can look like this:

apiVersion: v1
kind: Pod
metadata:
  name: busybox
spec:
  containers:
  - name: busybox
    image: busybox
    command: ["sh", "-c", "echo Hello Kubernetes! && sleep 3600"]
  • Run this pod usind kubectl apply -f <YAML> -n <namespace>.
  • Check if the pod is running.
  • Use kubectl logs <podname> -n <namespace> to get the logs of the pod.
  • Login to the pod via an interactive session kubectl exec -it <podname> -n <namespace> -- sh

Tip: You can update your default context for another namespace with kubectl config set-context --current --namespace=<insert-namespace-name-here>. If this is set to your specific namespace the -n <namespace> parameter can be omitted.

1 - (Excursus) Imperative use of kubectl

Most kubectl commands can either be used declarative via a YAML file or optionally imperative (similar to docker).

For example you can run the busybox image like this:

kubectl run busybox --image=busybox --command -- sh -c "echo Hello Kubernetes! && sleep 3600"

This is especially useful with the --dry-run=client and -o=yaml parameters, which allows you to create a basic YAML template via kubectl without referencing the documentation.

Example:

kubectl run busybox --image=busybox --dry-run=client -o=yaml --command -- sh -c "echo Hello Kubernetes! && sleep 3600" > busybox.yaml

This procedure can also be used to create an interactive pod session on the fly:

kubectl run -i --tty busybox --image=busybox --restart=Never

For additional tips and tricks please visit the official kubernetes kubectl cheatsheet

2 - Jobs

Jobs (official docs)

Jobs are kubernetes resources that manage pods intended to run to (successful) completion (like calculations or tasks). This is similar to regular batch processing engines like SLURM or SGE.

A simple example for a job looks like this:

apiVersion: batch/v1
kind: Job
metadata:
  name: pi
spec:
  template:
    spec:
      containers:
      - name: pi
        image: perl
        command: ["perl",  "-Mbignum=bpi", "-wle", "print bpi(2000)"]
      restartPolicy: Never
  backoffLimit: 4

This job uses the perl container to calculate 2000 decimal places of Pi.

  • Excecute the job
  • Examine the created pods
  • You can get a more detailed status description with kubectl describe jobs/pi
  • Examine the logs after the pod is finished.

The finished pod and job will stay forever. You can change this behaviour if you specify a time to live (TTL) with ttlSecondsAfterFinished: 100 in the spec section of the job. This results in automatic deletion of the job and pod after 100 seconds.

  • Delete the old job
  • Spawn a new job with a TTL deadline

Tip: Jobs can also be created imperatively:

kubectl create job pi --image=perl -- perl -Mbignum=bpi -wle "print bpi(2000)"