Skip to content

Latest commit

 

History

History
169 lines (138 loc) · 6.38 KB

README.md

File metadata and controls

169 lines (138 loc) · 6.38 KB
Raw

Custom Project Deployed with DigitalOcean Kubernetes (DOKS)

This repo is a template for customized, OpenRemote-based projects, which can be deployed with DigitalOcean k8s resource types using the included terraform configs.

The build pipeline starts with gradle and uses docker images, similar to what is described here: https://github.com/openremote/custom-project/blob/main/docker-compose.yml#L13

Volume management happens through DigitialOcean apis via terraform commands, rather than via docker volumes. For this reason, the deployment volume (used for customization) is deployed as a kubernetes initContainer which copies the customization folders/files from your custom deployment docker image into the mounted persistent volume 'deployment-data'.

You may notice some of the AWS-specific resources from the original examples have no obvious analogs in the configurations, it's because Kubernetes has internally managed features which replaced these:

  • CloudFormation
  • VPC & Security Groups
  • Route 53
  • SNS

DevOps Setup

Create your environment config, and populate it with your project-specific values

cp .ci_ci/digital_ocean/live/dev/cluster/terragrunt.hcl.example \
   .ci_ci/digital_ocean/live/dev/cluster/terragrunt.hcl

To configure & deploy a new environment (i.e. production), copy dev/cluster to {envname}/cluster and populate the terragrunt.hcl for that env.

Install doctl, terragrunt

OSX:

brew install doctl
brew install terragrunt

Ubuntu:

sudo snap install doctl
brew install terragrunt

Others: https://docs.digitalocean.com/reference/doctl/how-to/install/ https://terragrunt.gruntwork.io/docs/getting-started/install/

Install terraform

OSX:

brew tap hashicorp/tap && brew install hashicorp/tap/terraform

Others: https://developer.hashicorp.com/terraform/tutorials/aws-get-started/install-cli

Setup pass for local secrets management:

brew install pass # OSX
gpg --list-keys # check for plausible personal gpg key to encrypt secrets
# gpg --generate-key # use this if you don't already have one
gpg --list-keys # Copy the long string from the `pub` entry that it lists
pass init XXXXX # paste the string to initialize

For installation of pass on other OS's, check https://www.passwordstore.org/#download

Generate a personal access token for DigitalOcean

Go to the dashboard:

  • API > Tokens > Generate new token
  • give it a name
  • store it securely with pass insert openremote/do_token

Generate a spaces access credentials for DigitalOcean

Go to the daashboard:

  • API > Tokens > Spaces Keys
  • click "Generate New Key"
  • Give it a name, include environment
  • store securely with pass insert openremote/spaces_access_id, pass insert openremote/spaces_secret_key

Login to DigitalOcean cli tools

doctl auth init -t $(pass openremote/do_token)

Docker image pipeline

(one time only, completed) create Sk8net docker image registry at digitalocean
doctl registry create openremote
DigitalOcean docker registry login
doctl registry login
Build & push the proxy image that was adjusted for the DigitalOcean & k8s architecture

Clone the special proxy repo and build it separately from this project:

pushd ../ && git clone [email protected]:FreeSK8/sk8net_proxy.git && pushd sk8net_proxy
export PROXY_VERSION=latest
docker build -t openremote/proxy:$PROXY_VERSION ./deployment/build/
docker tag openremote/proxy:$PROXY_VERSION registry.digitalocean.com/openremote/openremote/proxy:$PROXY_VERSION
docker push registry.digitalocean.com/openremote/openremote/proxy:$PROXY_VERSION
popd && popd
Build & push a your project customizations via Deployment docker image
export DEPLOYMENT_VERSION=$(git rev-parse --short HEAD)
docker build -t openremote/custom-deployment:$DEPLOYMENT_VERSION ./deployment/build/
docker tag openremote/custom-deployment:$DEPLOYMENT_VERSION registry.digitalocean.com/openremote/openremote/custom-deployment:$DEPLOYMENT_VERSION
docker push registry.digitalocean.com/openremote/openremote/custom-deployment:$DEPLOYMENT_VERSION

Deploy changes by updating this hash in terragrunt.hcl under custom_deployment_hash

Build & push a custom Manager docker image (optional, only if you need it)

You will also need to update the reference in kubernetes_stateful_set.web for the manager image, point it to your private repo & tag.

export MANAGER_VERSION="${commit_hash}"
docker build -t openremote/manager:$MANAGER_VERSION ./openremote/manager/build/install/manager/
docker tag openremote/manager:$MANAGER_VERSION registry.digitalocean.com/openremote/openremote/manager:$MANAGER_VERSION
docker push registry.digitalocean.com/openremote/openremote/manager:$MANAGER_VERSION

Deployment

We use terragrunt to manage k8s deployments using environment based configs. Terraform state is stored in digital ocean spaces (aka S3) under the bucket openremote-terraform-states.

Create/update kubernetes fabric in dev env

cd .ci_cd/digital_ocean/live/dev/cluster
export TF_VAR_do_token=$(pass openremote/do_token)
export AWS_ACCESS_KEY_ID=$(pass openremote/spaces_access_id)
export AWS_SECRET_ACCESS_KEY=$(pass openremote/spaces_secret_key)
terragrunt apply -target=digitalocean_kubernetes_cluster.primary 

doctl kubernetes cluster kubeconfig save shared-dev

# human do this: Go into the digital ocean dashboard, container registry, click edit and
# enable integration for the newly created k8s cluster

terragrunt apply

# if a new loadbalancer was created (first time you deploy this env), you need to point a
# DNS record at it now. You can find the IP of the LB in the DigitalOcean dashboard.
Helpful things...

If a cluster is nuked and volumes are left orphaned, you can import them:

terragrunt import digitalocean_volume.deployment_data #(id available on inspection of the html table in DO volumes manager, lol)
terragrunt import digitalocean_volume.manager_data #(id)
terragrunt import digitalocean_volume.postgresql_data #(id)
terragrunt import digitalocean_volume.proxy_data #(id)

Remove state for already-deprovisioned resources, or if you want to detach existing resource from tf management

terragrunt state list

terragrunt state rm kubernetes_persistent_volume_claim.proxy_data # THIS IS AN EXAMPLE, target your desired resource