A demo constructed for the Toronto internship program orientation.
In the first demo, we will be building and running a simple go server locally.
In the second demo, we will be deploying it as an application to an OpenShift cluster via oc
With sections from previous year demos by:
Angel Misevski Jie Kang Josh Pinkney
For the Container demo, we will be doing it via podman. If you do not have this locally (which podman) you can get it via dnf install podman.
For the Cluster demo, we will be using oc (the OpenShift Origin Client), but a lot of the commands are also the same via kubectl. You should have oc locally already (check with which oc). If you do not, we recommend downloading the latest version via OpenShift mirrors.
Using oc and kubectl from the commandline involves a whole lot of typing. It's really useful to have autocompletion for command names, arguments, resource types, and object names. To enable this (if it's not already in your .bashrc), you can run
source <(kubectl completion bash); source <(oc completion bash)To get more information, use oc completion --help. (Note: if you're using zsh, that's supported too! Just replace bash with zsh above)
Finally, you should clone this repo locally (e.g. git clone https://github.com/yuqi-zhang/intern-demos.git)
Inside the container repo we will have the necessary bits to build a web server. It is a simple frontend (index.tmpl) and gol
7EC5
ang backend app.go with the necessary go.mod and go.sum to build. You can build this locally (requires you to install golang) via ./build-local.sh and then running the server via ./server.
Now if you go to http://localhost:8080/ you can see the simple server running.
Let us instead run that in a container. With podman installed, run:
podman build -t localhost/go-server .
To create the image. Then, podman run -it -p 8080:8080 localhost/go-server to start it.
Now it's running the same thing, but via a container.
You can also push the image to a registry, e.g.: podman push localhost/go-server:latest quay.io/jerzhang/demo-go-server
Once you are done, you can see containers with podman ps -a and podman rm the container.
The openshift/manifests directory contains the YAML files we'll be deploying to our cluster.
You can apply a template to the cluster using the command
oc apply -f templates/<file>.yamlYou can also use
oc apply -f templates/to apply all files in the templates folder at once.
First, let's login to our cluster. Go to the cluster URL provided. Ignore the security warnings, and choose the login with htpasswd option. Your username is your Red Hat id and your password is openshift.
Once inside, you can click on your name on the top right, and select Copy login command, which will bring you to another sign-on page. Login, hit display token, and you should see something like:
oc login --token=sha256~${TOKEN} --server=${URL}Copy that, and run that in a local terminal, where you plan on applying the config files. We will be doing a lot of commands via the command line, but you can actually also do the same via the web console.
Once you've successfully authenticated, create a project in your own name: e.g. oc new-project jerzhang
And you can see projects with oc project.
From now on, whenever you run a command, it will be under this project (namespace).
Now, lets create our deployment:
oc apply -f manifests/deployment.yamlThis creates the web server we just did in the containers demo in our cluster as a deployment. The image we are using is quay.io/jerzhang/demo-go-server:latest which is what we just pushed.
we can check the status of our deployment using oc get and oc describe
oc get deployment demo-deployment
# output:
NAME READY UP-TO-DATE AVAILABLE AGE
demo-deployment 0/1 1 0 2sWhile we wait for everything to get started, we can also:
- Check out the pods created by the deployment using
oc get po -l 'app=demo-app'. Here, we're using a label selector to only get pods with labelapp=demo-app. Otherwise, we'd get all pods in the current namespace. - View the yaml we used for our deployment, with its current status using
oc get deploy demo-deployment -o yaml(you can use-o jsonto output JSON as well!)
We've now got a deployment that contains our server running in Kubernetes! How do we access it from our browser?
Next, we need to create a service to route traffic to the set of pods maintained by our deployment:
oc apply -f manifests/service.yamland then we need to create a route to expose that service to the internet:
oc apply -f manifests/route.yamlNote: routes are an OpenShift specific object. For base Kubernetes, you will have to set up ingress instead.
Once you've created the route, you can get the URL you'll use to access the deployment from oc:
oc get routes
# output
NAME HOST/PORT PATH SERVICES PORT TERMINATION WILDCARD
demo-route demo-route-<current-namespace>.<url-for-cluster> demo-service http NoneAccessing this URL should show you the server we ran locally.
Our deployment currently has a replica of 1, meaning there is exactly 1 pod running for it.
What happens if something breaks?
oc delete pod/demo-deployment-<hash>It will recover very fast (kubernetes is trying to reconcile the deployment). But that doesn't seem very good in terms of availability.
Let's try to scale the deployment up to three replicas using
oc scale deploy demo-deployment --replicas=3Wait for them to be ready
oc get deployment demo-deployment
# output:
NAME READY UP-TO-DATE AVAILABLE AGE
demo-deployment 1/3 3 1 6m30sdelete a pod:
oc delete pod/demo-deployment-<hash>No downtime.
To remove everything we've deployed thus far, you can use a label selector to delete all the objects we created in the demo:
oc delete all -l 'app=demo-app'