Skip to content

GitOps

0 Goal

In this tutorial, we will try to use DevStream to achieve the following:

  1. create a new repository for a Python web application written with Flask;
  2. setup basic CI pipelines for the repo we created with GitHub Actions;
  3. install Argo CD for GitOps in an existing Kubernetes cluster;
  4. create an Argo CD application that deploys the web application generated in step 1.

Note:

in step 3, Argo CD is installed in an existing Kubernetes cluster. Setting up infrastructure like a Kubernetes cluster is something DevStream chooses not to do.

If you want to follow this tutorial and give it a try yourself but don't know how to get a Kubernetes cluster up and running locally, maybe the following blogs (also from DevStream) would help:

1 TL;DR: See It in Action

If you prefer to see this GitOps doc in action, check out the video demo below:

It's from an older version of DevStream with slightly different configuration files, but you get the gist. We will update the video with the latest version of DevStream soon; bear with us for a little while.

For Chinese readers, watch this one:

However, if you are like us, who prefer to do things hands-on and get their hands dirty, read on, follow the steps, and have a go yourself!

2 Overview

DevStream will use the following plugins to achieve the goal described in Section 0:

  1. repo-scaffolding
  2. github-actions
  3. helm-installer
  4. argocdapp

However, you do not have to worry about these plugins because DevStream will manage them automatically for you.

3 Getting Started: Download DevStream (dtm)

Create a temporary working directory for this tutorial:

Bash
mkdir test
cd test/

Then, under the newly created directory, execute the following command:

Bash
sh -c "$(curl -fsSL https://download.devstream.io/download.sh)"

This script checks your system and downloads the corresponding dtm binary. Then the binary will be granted execution permission.

If you do an ls, you can see the binary dtm has been downloaded:

Bash
tiexin@mbp ~/work/devstream-io/test $ ls
dtm

And as a test, try to execute it, and you will get similar output to the following:

Bash
tiexin@mbp ~/work/devstream-io/test $ ./dtm
DevStream is an open-source DevOps toolchain manager

######                 #####
#     # ###### #    # #     # ##### #####  ######   ##   #    #
#     # #      #    # #         #   #    # #       #  #  ##  ##
#     # #####  #    #  #####    #   #    # #####  #    # # ## #
#     # #      #    #       #   #   #####  #      ###### #    #
#     # #       #  #  #     #   #   #   #  #      #    # #    #
######  ######   ##    #####    #   #    # ###### #    # #    #

Usage:
  dtm [command]

Available Commands:
  apply       Create or update DevOps tools according to DevStream configuration file
  completion  Generate the autocompletion script for dtm for the specified shell
  delete      Delete DevOps tools according to DevStream configuration file
  destroy     Destroy DevOps tools deployment according to DevStream configuration file & state file
  develop     Develop is used for develop a new plugin
  help        Help about any command
  init        Download needed plugins according to the config file
  list        This command only supports listing plugins now
  show        Show is used to print plugins' configuration templates or status
  upgrade     Upgrade dtm to the latest release version
  verify      Verify DevOps tools according to DevStream config file and state
  version     Print the version number of DevStream

Flags:
      --debug   debug level log
  -h, --help    help for dtm

Use "dtm [command] --help" for more information about a command.

Optional: you can move dtm to a directory which is in your $PATH. For example: mv dtm /usr/local/bin/. This will allow you to run dtm directly without having to prefix it with the dot and slash (./dtm).

For more methods on how to install DevStream, see install dtm.

4 Config File

Run command below to get a valid config.yaml file:

Bash
./dtm show config -t gitops > config.yaml

Then modify the vars section in the config.yaml file accordingly. Please update the values for githubUser and dockerUser to your real users.

In the example above, I set these vars like the following:

Variable Example Note
githubUser IronCore864 case-sensitive, use your GitHub username strictly here
dockerUser ironcore864 case-sensitive, use your DockerHub username strictly here

5 Environment Variables

The following environment variables are required for this to work:

Bash
export GITHUB_TOKEN="YOUR_GITHUB_TOKEN_HERE"
export IMAGE_REPO_PASSWORD="YOUR_DOCKERHUB_TOKEN_HERE"

Note:

if you don't know how to create these two tokens, check out:

6 Init

Run:

Bash
./dtm init -f config.yaml

This downloads the required plugins, according to the config file, automatically.

You'll get some outputs similar to the following:

Bash
2022-12-05 17:46:01 ℹ [INFO]  Using dir </Users/tiexin/.devstream/plugins> to store plugins.
2022-12-05 17:46:01 ℹ [INFO]  -------------------- [  repo-scaffolding-darwin-arm64_0.10.2  ] --------------------
... (omitted)
... (omitted)
2022-12-05 17:46:51 ✔ [SUCCESS]  Initialize finished.

7 Apply

Run:

Bash
./dtm apply -f config.yaml -y

You will see similar outputs as the following:

Text Only
2022-12-05 17:49:49 ℹ [INFO]  Apply started.
2022-12-05 17:49:49 ℹ [INFO]  Using local backend. State file: devstream.state.
2022-12-05 17:49:49 ℹ [INFO]  Tool (repo-scaffolding/myapp) found in config but doesn't exist in the state, will be created.
2022-12-05 17:49:49 ℹ [INFO]  Tool (helm-installer/argocd) found in config but doesn't exist in the state, will be created.
2022-12-05 17:49:49 ℹ [INFO]  Tool (github-actions/flask) found in config but doesn't exist in the state, will be created.
2022-12-05 17:49:49 ℹ [INFO]  Tool (argocdapp/default) found in config but doesn't exist in the state, will be created.
2022-12-05 17:49:49 ℹ [INFO]  Start executing the plan.
2022-12-05 17:49:49 ℹ [INFO]  Changes count: 4.
... (omitted)
... (omitted)
2022-12-05 17:51:51 ℹ [INFO]  -------------------- [  Processing progress: 4/4.  ] --------------------
2022-12-05 17:51:51 ℹ [INFO]  Processing: (argocdapp/default) -> Create ...
2022-12-05 17:51:52 ℹ [INFO]  application.argoproj.io/helloworld created
2022-12-05 17:51:52 ✔ [SUCCESS]  Tool (argocdapp/default) Create done.
2022-12-05 17:51:52 ℹ [INFO]  -------------------- [  Processing done.  ] --------------------
2022-12-05 17:51:52 ✔ [SUCCESS]  All plugins applied successfully.
2022-12-05 17:51:52 ✔ [SUCCESS]  Apply finished.

8 Check the Results

Let's continue to look at the results of the apply command.

8.1 Repository

The repository is created automatically by DevStream with scaffolding code:

8.2 CI Pipelines with GitHub Actions

GitHub Actions pipelines are created and executed:

8.3 Argo CD Installation

Argo CD is installed in your Kubernetes cluster:

Bash
tiexin@mbp ~/work/devstream-io/test $ kubectl get namespaces
NAME                 STATUS   AGE
argocd               Active   5m42s
default              Active   6m28s
kube-node-lease      Active   6m29s
kube-public          Active   6m29s
kube-system          Active   6m29s
local-path-storage   Active   6m25s
tiexin@mbp ~/work/devstream-io/test $ kubectl get pods -n argocd
NAME                                               READY   STATUS    RESTARTS   AGE
argocd-application-controller-0                    1/1     Running   0          5m43s
argocd-applicationset-controller-66687659f-dsrtd   1/1     Running   0          5m43s
argocd-dex-server-6944757486-clshl                 1/1     Running   0          5m43s
argocd-notifications-controller-7944945879-b9878   1/1     Running   0          5m43s
argocd-redis-7887bbdbbb-xzppj                      1/1     Running   0          5m43s
argocd-repo-server-d4f5cc7cb-8gj24                 1/1     Running   0          5m43s
argocd-server-5bb75c4bd9-g948r                     1/1     Running   0          5m43s

8.4 Continuous Deployment with Argo CD

The CI pipelines build a Docker image and push it into Dockerhub, and an Argo CD application created by DevStream deploys the app already:

Bash
tiexin@mbp ~/work/devstream-io/test $ kubectl get deployment -n default
NAME         READY   UP-TO-DATE   AVAILABLE   AGE
helloworld   1/1     1            1           5m16s
tiexin@mbp ~/work/devstream-io/test $ kubectl get pods -n default
NAME                          READY   STATUS    RESTARTS   AGE
helloworld-69b5586b94-wjwd9   1/1     Running   0          5m18s
tiexin@mbp ~/work/devstream-io/test $ kubectl get services -n default
NAME         TYPE        CLUSTER-IP    EXTERNAL-IP   PORT(S)   AGE
helloworld   ClusterIP   10.96.73.97   <none>        80/TCP    5m27s
kubernetes   ClusterIP   10.96.0.1     <none>        443/TCP   8m2s

If you do a port-forwarding:

Bash
kubectl port-forward -n default svc/helloworld 8080:80

And accesses localhost:8080 in your browser, you can see the deployed app return a "Hello, World!" to you. Hooray!

9 Clean Up

Run:

Bash
./dtm delete -f config.yaml -y

And you will get similar outputs to the following:

Bash
2022-12-05 17:59:25 ℹ [INFO]  Delete started.
2022-12-05 17:59:26 ℹ [INFO]  Using local backend. State file: devstream.state.
2022-12-05 17:59:26 ℹ [INFO]  Tool (argocdapp/default) will be deleted.
2022-12-05 17:59:26 ℹ [INFO]  Tool (github-actions/flask) will be deleted.
2022-12-05 17:59:26 ℹ [INFO]  Tool (repo-scaffolding/myapp) will be deleted.
2022-12-05 17:59:26 ℹ [INFO]  Tool (helm-installer/argocd) will be deleted.
2022-12-05 17:59:26 ℹ [INFO]  Start executing the plan.
2022-12-05 17:59:26 ℹ [INFO]  Changes count: 4.
... (omitted)
... (omitted)
2022-12-05 17:59:35 ℹ [INFO]  -------------------- [  Processing done.  ] --------------------
2022-12-05 17:59:35 ✔ [SUCCESS]  All plugins deleted successfully.
2022-12-05 17:59:35 ✔ [SUCCESS]  Delete finished.

Then you can delete what we created:

Bash
cd ../
rm -rf test/
rm -rf ~/.devstream/