Kubernetes
Kubernetes is an orchestration tool that allows management of containers across multiple nodes. This demo uses only 1 node, but you could have 10s-1000s.
Kubernetes might be more developer-centric than some of the other technologies we’ve looked at. The idea is to run your code without having to know anything about the underlying system its running on.
Overview
- Pod: Smallest deployable unit. Consists of 1 or more containers. Kind of like “localhost”.
- Deployment: Multiple pods.
- Service: Expose a pod or deployment to network.
- Volume: Attach storage.
- Namespace: Permissions-based grouping of objects.
- Job: Run a container to completion.
And more not covered today:
- ConfigMap: Store strings or files for pods to use.
- Secret: Cncrypted configmap.
- Ingress: Expose HTTP+S routes to the network. Like a HTTP-specific Service.
- DaemonSet, ReplicaSet, LoadBalancer, etc.: More ways to do things.
Kube install check
Once ansible has finished ( Ansible ), you can check that everything is up:
kubectl get all --all-namespaces
Should look something like this (node everything in “Running” status):
NAMESPACE NAME READY STATUS RESTARTS AGE
kube-system pod/coredns-5c98db65d4-v6kr6 1/1 Running 0 25m
kube-system pod/coredns-5c98db65d4-v8bwz 1/1 Running 0 25m
kube-system pod/etcd-js-17-151.jetstream-cloud.org 1/1 Running 0 24m
kube-system pod/kube-apiserver-js-17-151.jetstream-cloud.org 1/1 Running 0 24m
kube-system pod/kube-controller-manager-js-17-151.jetstream-cloud.org 1/1 Running 0 24m
kube-system pod/kube-flannel-ds-amd64-dsg87 1/1 Running 0 6m1s
kube-system pod/kube-proxy-gf4st 1/1 Running 0 25m
kube-system pod/kube-scheduler-js-17-151.jetstream-cloud.org 1/1 Running 0 24m
NAMESPACE NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
default service/kubernetes ClusterIP 10.96.0.1 <none> 443/TCP 25m
kube-system service/kube-dns ClusterIP 10.96.0.10 <none> 53/UDP,53/TCP,9153/TCP 25m
NAMESPACE NAME DESIRED CURRENT READY UP-TO-DATE AVAILABLE NODE SELECTOR AGE
kube-system daemonset.apps/kube-flannel-ds-amd64 1 1 1 1 1 beta.kubernetes.io/arch=amd64 6m1s
kube-system daemonset.apps/kube-flannel-ds-arm 0 0 0 0 0 beta.kubernetes.io/arch=arm 13m
kube-system daemonset.apps/kube-flannel-ds-arm64 0 0 0 0 0 beta.kubernetes.io/arch=arm64 13m
kube-system daemonset.apps/kube-flannel-ds-ppc64le 0 0 0 0 0 beta.kubernetes.io/arch=ppc64le 13m
kube-system daemonset.apps/kube-flannel-ds-s390x 0 0 0 0 0 beta.kubernetes.io/arch=s390x 13m
kube-system daemonset.apps/kube-proxy 1 1 1 1 1 beta.kubernetes.io/os=linux 25m
NAMESPACE NAME READY UP-TO-DATE AVAILABLE AGE
kube-system deployment.apps/coredns 2/2 2 2 25m
NAMESPACE NAME DESIRED CURRENT READY AGE
kube-system replicaset.apps/coredns-5c98db65d4 2 2 2 25m
Check out how many nodes we have registered:
kubectl get nodes
NAME STATUS ROLES AGE VERSION
js-104-106.jetstream-cloud.org Ready master 6h12m v1.15.1
Run interactive container
You can do most anything in Kubernetes that you can do in Docker. E.g. run a simple interactive container:
kubectl run busybox -i --tty --image=busybox --restart=Never --rm -- sh
/ #
/ # busybox | head -1
BusyBox v1.31.0 (2019-07-16 01:13:11 UTC) multi-call binary.
Etherpad example
You can put as many Kube objects as you want in a YAML file and “apply” it all as one big chunk.
For example, take a look at mpackard-etherpad.yml.
vim mpackard-etherpad.yml
If you’re on a shared system, change all the “mpackard” to your name. Note there is Deployment and a Service section. The Deployment indicates what container(s) to run, and the Service creates a network route to the containers.
cat mpackard-etherpad.yml
Deploy it:
kubectl apply -f mpackard-etherpad.yml
See if it’s running:
kubectl get pod,deployment,service
NAME READY STATUS RESTARTS AGE
pod/etherpad-55d84f9ffc-xjdlc 0/1 ContainerCreating 0 14s
NAME READY UP-TO-DATE AVAILABLE AGE
deployment.extensions/etherpad 0/1 1 0 15s
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service/etherpad NodePort 10.109.117.107 <none> 80:31763/TCP 14s
service/kubernetes ClusterIP 10.96.0.1 <none> 443/TCP 10m
Note the external port 31763 of the etherpad service.
Point your browser at your vm and that port number, e.g. http://129.114.104.121:31763
Now take it down:
kubectl delete -f etherpad.yml
deployment.apps "etherpad" deleted
service "etherpad" deleted
Edit the yml file and change “replicas” to 2, then save and re-apply.
kubectl apply -f mpackard-etherpad.yml
kubectl get pods
NAME READY STATUS RESTARTS AGE
pod/mpackard-etherpad-5bd7855697-4vgcb 1/1 Running 0 5s
pod/mpackard-etherpad-5bd7855697-wcfw4 1/1 Running 0 144m
Hit the URL again, create an etherpad, and hit refresh a couple times. What happens?
Try killing one of the pods:
kubectl delete pod/mpackard-etherpad-5bd7855697-4vgcb
kubectl get pods
…it comes back!
Discussion Questions
- Why do you think the etherpad fails 50% of the time?
- What would be the advantage of adding more nodes?
Storage
Kube has many ways to handle persistent storage for your pods.
- persisten volume claims
- more traditional file mounting (nfs, etc)
Volume example
Hopefully you have a working local NFS server from the Ansible step.
ls /kubedata/nginx
enterprise.jpg index.html
This is going to get mounted inside our container so that nginx web server can serve it.
Take a look at mpackard-nginx.html. Change the “mpackard” to your username. Note the addition of the Volume and Volumemounts section.
Apply it:
kubectl apply -f mpackard-nginx.html
Check for running:
kubectl get pods,deployments,services
NAME READY STATUS RESTARTS AGE
pod/mpackard-nginx-688877c6f9-b99d7 1/1 Running 0 16m
pod/mpackard-nginx-688877c6f9-ksdp4 1/1 Running 0 16m
NAME READY UP-TO-DATE AVAILABLE AGE
deployment.extensions/mpackard-nginx 2/2 2 2 16m
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service/mpackard-nginx NodePort 10.97.84.239 <none> 80:31523/TCP 16m
Now point your browser at your IP & port (31523 here). Hit refresh a few times. Even though there are 2 nginx pods, they are serving the same data.
Discussion Questions
- What issues would you have with read-write data (e.g. if you were hosting a database)?
- How would you chain some apps together to make a workflow?
- What other features would you like to make this work for your environment?
Job example
Run a container to completion.
kubectl apply -f https://k8s.io/examples/controllers/job.yaml
Wait a second… did you just run that without looking at it first? :)
Anyway, check the output:
kubectl logs job.batch/pi
Discussion Questions
- How are Pods and Jobs different?
- What might be another way to run “jobs” using a queueing system like RabbitMQ or Redis?
- Why shouldn’t I run random untrusted stuff off the internet?
- What sorts of things could you accomplish by stringing a bunch of jobs/pods/services together?
Fin
Kubernetes Reset (destroys all current data)
Caution, destroys data
If you really need to reset Kubernetes and start over with new install:
kubeadm reset -f
sudo systemctl stop kubelet
sudo rm -rf /etc/kubernetes /var/lib/kubelet /var/lib/etcd
Then re-run the install playbook.
| Previous: Ansible | Top: Course Overview |