googlecloudplatform / kubernetes-workshops

Kubernetes Workshops

This repository contains both complete workshops, as well as segmented workshop modules that can be combined to create Kubernetes workshops of various lengths and focus.

Modules contain a README.md that walks through the module. If code, configuration, or scrips are needed, it is included and tested.

This is not an official Google product.

Full Workshops

workshop	version	description
Kubernetes 101	v1.2.0	Covers the basics of using Kubernetes to manage applications at scale. In this workshop, you'll take an app, build it into a docker container, then use Kubernetes to deploy, scale, and update it. This workshop comes in multiple versions: A video course, a codelab with an accompanying talk, or a set of workshop material with slides.

Individual Modules

Name	Slides	Level	Time Estimate	Completion Status
Cluster Bring Up	Link	Beginner	1 hour	Ready
Quickstart	Link	Beginner	1 hour	Ready
Core Kubernetes Concepts	Link	Beginner	4 hours	Ready
Storing State	Link	Intermediate	2 hours	Ready
Advanced Concepts	Link	Intermediate	2 hours	Ready
Dockerize an App	Link	Intermediate	2 hours	Defer
Networking	Link	Intermediate	2 hours	Defer
Troubleshooting	Link	Intermediate	2 hours	Delay
Putting it all together	Link	Advanced	2 hours	Delay

Status: Not Started --> In Progress --> Draft --> Ready

Overview of Modules

This is incomplete. These are just brainstorming / rough notes.

Quickstart

• Quick
• Not complex, uses kubectl run, kubectl expose.
• Demonstrates the ease of using Kubernetes without learning all the concepts and config files up front.

Cluster Bring Up

• Uses the open source kubernetes release with cluster/kube-up.sh for cloud bringup
• Option for local docker

Core Kubernetes Concepts

• Introduce one concept at a time, and then use that concept
  • Order: pod, service, rc, deployment
• go over a declarative pod representation of quickstart app
  • contains 1 pod
• logs, exec, port forwarding
• introduce service
• overview pods, labels, selectors, and services
• change pod to RC
• discuss RC
• scale pod up
• introduce deployments
• move everything under a deployment
• update to new versions of our app, quick rolling update
  • lightweight here - more detail in "Advanced" module

Storing State

• Deploy an app with MySQL
• multiple iterations where to store the data, how it goes away
  • start with host volume, end at persistent disk
• More of a lecture module, slides discuss state in greater length

Dockerize an App

• Start with an app
• Write the Dockerfile
• build
• push to registry

Advanced Concepts

• Lecture

  • App/container patterns http://blog.kubernetes.io/2015/06/the-distributed-system-toolkit-patterns.html
  • mapping non-containerized apps

• Hands on:

  • A/B deployment
  • Canary patterns
  • Rolling Deployments
  • Autoscaling

Networking

• More of a lecture module, slides discuss networking in greater length

• Types of external services VIP/nodeport, run service with each and see how we get into the cluster

• discuss subnets, explore on running nodes

• How K8s networking works

• Setting up an external load balancer - Nginx

• Ideas on how to plug into your environment

Troubleshooting

• Logging & monitoring
• Troubleshooting / Debugging

Putting it all together

• deploy a production ready app

• Use all the above

• Build up a significant realistic app

• ( not so much lecture, just deploy all this stuff: )

• web frontends, caching, backend jobs, datastore, load testing

• Logging & monitoring

• Troubleshooting

• Autoscaling

Contributing changes

• See CONTRIBUTING.md

Licensing

• See LICENSE

kubernetes-workshops's Issues

Should this repo be marked as Archived or updated per the latest K8s Version?

Hello @jeffmendoza

I have a query, if this repo be marked as Archived or the codebase(documents and manifests) be updated per the latest K8s Version?

Thanks,
S

Figure out command line quoting

what style to use?

supported clusters

will have bring up for local and cloud

most docs should be split into two, one for local, one for cloud

Should link to docs

When introducing concepts, link out to docs

Core concepts - should we use RC or RS?

deployments create RS, should we just start with those, instead of RC?

Complete Quickstart Module

Fix tests

bundles/kubernetes-101/workshop/app/auth/main.go:13:2: cannot find package "github.com/askcarter/io16/app/handlers" in any of:
    /home/travis/gopath/src/github.com/GoogleCloudPlatform/kubernetes-workshops/bundles/kubernetes-101/workshop/app/vendor/github.com/askcarter/io16/app/handlers (vendor tree)
    /home/travis/.gimme/versions/go1.7.1.linux.amd64/src/github.com/askcarter/io16/app/handlers (from $GOROOT)
    /home/travis/gopath/src/github.com/askcarter/io16/app/handlers (from $GOPATH)
bundles/kubernetes-101/workshop/app/auth/main.go:14:2: cannot find package "github.com/askcarter/io16/app/health" in any of:
    /home/travis/gopath/src/github.com/GoogleCloudPlatform/kubernetes-workshops/bundles/kubernetes-101/workshop/app/vendor/github.com/askcarter/io16/app/health (vendor tree)
    /home/travis/.gimme/versions/go1.7.1.linux.amd64/src/github.com/askcarter/io16/app/health (from $GOROOT)
    /home/travis/gopath/src/github.com/askcarter/io16/app/health (from $GOPATH)
bundles/kubernetes-101/workshop/app/auth/main.go:15:2: cannot find package "github.com/askcarter/io16/app/user" in any of:
    /home/travis/gopath/src/github.com/GoogleCloudPlatform/kubernetes-workshops/bundles/kubernetes-101/workshop/app/vendor/github.com/askcarter/io16/app/user (vendor tree)
    /home/travis/.gimme/versions/go1.7.1.linux.amd64/src/github.com/askcarter/io16/app/user (from $GOROOT)
    /home/travis/gopath/src/github.com/askcarter/io16/app/user (from $GOPATH)

Which Quickstart, or both?

1. 'kubectl run' a container. Poke it with exec/logs etc.
2. 'kubectl create' a huge deployment with one file, show the power of kubernetes.

replicationcontroller instead deployment

Following this guide: https://github.com/GoogleCloudPlatform/kubernetes-workshops/tree/master/quickstart

If i run:
kubectl run lobsters --image=gcr.io/google-samples/lobsters:1.0

it returns:
replicationcontroller "lobsters" created

and not:
deployment "lobsters" created

Error in creating skydns with clusterIP on the local

I was trying to use the local docker to launch the skydns and I ran into the issue.
Error from server: error when creating "./skydns.yaml": Service "kube-dns" is invalid: spec.clusterIP: Invalid value: "10.0.0.10": provided IP is not in the valid range

https://github.com/GoogleCloudPlatform/kubernetes-workshops/blob/master/bring-up/skydns.yaml#L12

and when i removed the clusterIP I was able to create the service.

Is the IP required?

fix the labs to work directly from gcloud console

we did the obvious, and followed the labs instructions to

curl -k https://<EXTERNAL_IP>:31000

And then ... yup, waste a tonne of time trying to work out what we did wrong before buggings Ian, who eventually went ... OH, console...

@ianlewis @kelseyhightower