Serverless Munki
This repository contains cross platform code to deploy a production ready Munki service, complete with AutoPkg, that runs entirely from within a single GitHub repository and an AWS S3 bucket. No other infrastructure is required. More specifically it contains the following:
- Terraform code to setup a Munki repo in AWS S3.
- Actions workflows to handle AutoPkg runs and related tasks.
- Directories for maintaining Munki items and AutoPkg overrides.
How it works
After following the deployment steps outlined below to setup your own GitHub repo and S3 bucket, an Actions workflow will run daily which does the following:
- Runs any AutoPkg recipes located in your
RecipOverrides/
folder. - Imports any new items into the the
munki_repo/
folder. - Git commits changes (pkgs, pkgsinfo) for each item into a separate branch.
- Creates a PR for each new item.
- Posts results to Slack (if enabled).
- Syncs approved changes in
munki_repo/
to your S3 bucket where the items will be available to client devices.
Deployment
Initial GitHub Setup
Firstly, you will need to create a new GitHub repository with Actions enabled. You can then clone this repo and copy its contents into your own private repo by running the following Terminal commands:
git clone [email protected]:adahealth/serverless-munki.git
cd serverless-munki
make init
By default this will create a new directory named my-serverless-munki
inside the parent directory of our cloned repo and initialize it as it's own Git repository. Now we can install (if you haven't already) and configure Git LFS for your repo. In our example, we are installing Git LFS via Homebrew but feel free to install it how ever you like.
brew install git-lfs
make lfs
Then you can go ahead and push your new repo to the Actions enabled GitHub repository you created earlier.
cd ../my-serverless-munki
git remote add origin <your-github-repo-url>
git branch -M master
git push -u origin master
AWS / Terraform setup
Log in to your AWS account and create an AWS IAM user with the following permissions: AWSLambdaFullAccess
, IAMFullAccess
, AmazonS3FullAccess
, CloudFrontFullAccess
. Then create an access key for the user and set the access key ID and secret key as environment variables. This is so that Terraform can authenticate to the AWS provider. Also, if you don't have Terraform installed you should do that now.
brew install [email protected]
export AWS_ACCESS_KEY_ID="<your-access-key-id>"
export AWS_SECRET_ACCESS_KEY="<your-secret-key>"
While we're at it, we can also add both the AWS_ACCESS_KEY_ID
and AWS_SECRET_ACCESS_KEY
as GitHub Actions secrets in our remote repo. They will be used in our Actions workflows when syncing our Munki files to our S3 bucket.
Next, we need to set our Terraform variables for our AWS configuration. Open the /terraform/variables.tf
file and adjust the variables to match what you want the bucket to be called, and set the username and password your Munki clients will use to access the repo.
# prefix should be globally unique. Some characters seem to cause issues;
# Something like yourorg_munki might be a good prefix.
variable "prefix" {
default = "YOU_BETTER_CHANGE_ME"
}
# you'd need to change this only if you have an existing bucket named
# "munki-s3-bucket"
variable "munki_s3_bucket" {
default = "munki-s3-bucket"
}
# the price class for your CloudFront distribution
# one of PriceClass_All, PriceClass_200, PriceClass_100
variable "price_class" {
default = "PriceClass_100"
}
# the username your Munki clients will use for BasicAuthentication
variable "username" {
default = "YOU_BETTER_CHANGE_ME"
}
# the password your Munki clients will use for BasicAuthentication
variable "password" {
default = "YOU_BETTER_CHANGE_ME"
}
Now we can change in to the terraform/
directory and check our Terraform plan.
cd terraform
terraform init
terraform plan
If everything is as expected we can apply the configuration.
terraform apply
That's it for our Munki "server" repository. We can use terraform outputs to obtain info for your client configuration.
terraform output cloudfront_url
# This is your SoftwareRepoURL.
terraform output username
terraform output password
# These are the credentials that your clients will use to access the S3 bucket.
Update the /.github/workflows/sync-repo.yml
file to include your bucket ID on line 41.
run : |
aws s3 sync "$GITHUB_WORKSPACE"/munki_repo s3://<ADD-YOUR-BUCKET-ID-HERE> --exclude '.DS_Store' --exclude '.keep' --delete
The Munki wiki covers configuring your clients to use BasicAuthentication using the username and password you've chosen. Be sure also to set Munki's SoftwareRepoURL to "https://<your-cloudfront_url>"
Slack notifications
To configure Slack notifications, simply create an incoming webhook in your Slack tenant and add the webook URL as a GitHub Actions secret with the name SLACK_WEBHOOK
Usage
AutoPkg
Add your AutoPkg recipe overrides to the RecipeOverrides/
folder, commit them to your remote repo and add any necessary parent recipe repos to the .github/workflows/autopkg-run.yml
workflow file by appending a repo-add
command to the "Add AutoPkg repos" step.
- name: Add AutoPkg repos
run: |
autopkg repo-add recipes
autopkg repo-add <parent-recipe-repo1>
autopkg repo-add <parent-recipe-repo2>
autopkg repo-add <parent-recipe-repo3>
# etc
Every time the autopkg-run workflow is triggered the following steps will happen inside of a GitHub Actions runner VM:
- Repository is checked out containing AutoPkg overrides and Munki Repo.
- Munki and AutoPkg is installed and configured.
- Each recipe in the RecipeOverides directory is run.
- If AutoPkg imported any new items into Munki, commit the changes and create a PR.
- If enabled, post results to Slack.
By default this is scheduled to run at 6am everyday between Monday and Friday. You can change this by editing the schedule in .github/workflows/autopkg-run.yml
.
After reviewing and merging any PRs created via the autopkg-run
workflow, the sync-repo
workflow will be triggered. This will sync any changes in your munki repo to your AWS S3 bucket where they will be available for your clients.
Updating recipe trust info
We update recipe trust info by manually running the update-trust-info
workflow. Make sure the parent recipe repo is included in the "Add AutoPkg Repos" step in the .github/workflows/update-trust-info.yml
file before triggering the workflow run.
Munki
You can populate and administer your munki repo whatever way you are used to by checking out your GitHub repo locally and making your required changes inside the munki_repo
folder. When changes are pushed to the remote Master branch, they will be automatically synced to your S3 bucket via the sync-repo
workflow.
Clean Repo
The clean-repo
workflow will remove older, unused software items from the Munki repo. By default it is scheduled to run every Tuesday at 19:00. You can change this by editing .github/workflows/clean-repo.yml
.
Acknowledgements
Terraform Munki Repo module from Graham Gilbert
The autopkg_tools.py
script is a fork of Facebook's autopkg_tools.py
The GitHub Actions workflows and this project in general are based heavily on the GitHub Actions AutoPkg setup from Gusto