Dockerised Kubernetes

Great for training and experiments!

You can now deploy the super-lightweight Kubernetes cluster, using the k3d project by Rancher.

The cluster is packaged for training classes that are easy to setup, thanks to Docker-in-Docker, and with the help of ttyd and inlets.

Start a (lightweight!) Kubernetes Cluster

docker-compose up cluster

The console of the master machine can be reached at:

http://localhost:50022/

From there you can run the usual kubectl commands, for example:

kubectl get po -A

A pre-configured nginx-based load balancer is setup on port 8081:

http://localhost:8081/

Also note that a pre-configured Docker Registry is started with the cluster, and can be referenced in tags like this registry.dev/webapp:1.0 in commands like:

docker push registry.dev/webapp:1.0

Note that registry images are peristed inside the folder ./registry/data/.

Download Docker images for preloading them

To save time and bandwidth, you can download the minimum set of images needed to startup the k3s cluster using the script:

./docker/images/save.sh

Also, if you want to preload other images in the Kubernetes nodes, save the Docker tar archives in the folder ./docker/images/preload/: look at the file ./docker/images/save.sh to see how easily you can do it.

Start publicly accessible student machines

First install and start an inlet server on a publicly accessible server:

curl -sLS https://get.inlets.dev | sh
inlets server --port=8080 --token 123456

Then copy the sample .env-example file to .env and fill it with the correct values for the env vars defined:

INLETS_TOKEN=123456
INLETS_REMOTE=inlets.server.ext:8080
INLETS_DOMAIN=publicserver.ext:8080

Setup a publicserver.ext host entry and a wildcard A record for *.publicserver.ext. Idf your DNS server/provider does not support wildcars hosts, configure a few DNS entries with the form ssh{1..20}.INLETS_DOMAIN (for example ssh0.publicserver.ext). You can also setup a www.publicserver.ext CNAME entry to publicserver.ext host.

Finally you can start the number of student machines you need, with a command like:

docker-compose up machine3

You can replace the number 3 in machine3 above with number from 1 to 20.

The student machines are then reachable at:

http://localhost:50122/ http://localhost:50222/ http://localhost:50322/ ... http://localhost:51922/ http://localhost:52022/

Each student will also be able to access its own machine shell at the urls:

http://ssh1.publicserver.ext:8080 http://ssh2.publicserver.ext:8080 http://ssh3.publicserver.ext:8080 [...] http://ssh19.publicserver.ext:8080 http://ssh20.publicserver.ext:8080

The default load balancer is configured to respond at:

http://publicserver.ext:8080 http://www.publicserver.ext:8080

Don't forget!

To stop the services run:

docker-compose down

Note that registry images are persisted inside the folder ./registry/data/.