Kubernetes Deployment Guide with Helm

Kubernetes is a robust container orchestration system for easy application deployment and management. Helm takes that a step further with by packaging up required helm “charts” into one deployment command.



In order to deploy the SEED platform on a Kubernetes you will need “cluster” which will be configured by your cloud service of choice. Each installation will be slightly different depending on the service. Below are links to quick-start guides for provisioning a cluster and connecting. These instructions are specifically for AWS, but after the Kubernetes cluster is launched, the helm commands can be used in the same way.

  • Amazon Web Services (AWS)

  • Google Cloud Platform (GCP)

  • Azure (AKS)

AWS CLI Configuration

Download and configure the AWS CLI with instructions: https://docs.aws.amazon.com/cli/latest/userguide/install-cliv2.html

aws configure
AWS Access Key ID [None]: <insert key> (from account)
AWS Secret Access Key [None]: <insert secret key> (from account)
Default region name [None]: us-east-1
Default output format [None]: json


Download and install Kubectl:

  • Windows

  • Mac (with Homebrew) brew install kubectl

    ` brew install kubectl `

Kubectl is the main function in which you will be interfacing with your deployed application on your cluster. This CLI is what connects you to your cluster that you have just provisioned. If your cloud service did not have you configure kubectl in your cluster setup, you can download it here. Once kubectl is installed and configured to your cluster you can run some simple commands to ensure its working properly:

#View the cluster
kubectl cluster-info

#View pods, services and replicasets (will be empty until deploying an app)
kubectl get all

All of the common kubectl commands can be found in these docs


For those unfamiliar with CLIs, there are a number of GUI applications that are able to deploy on your stack with ease. One of which is Kubernetes native application called Dashboard UI or a third-party application called Octant brew install octant.


Helm organizes all of your Kubernetes deployment, service, and volume yml files into “charts” that can be deployed, managed, and published with simple commands. To install Helm:

EKS Control (AWS Specific)

EKSCtl is a command line tool to manage Elastic Kubernetes clusters on AWS. If not using AWS, then disregard this section.

  • Windows

  • Mac (with Homebrew) brew install eksctl

To launch a cluster on using EKSCts, run the following command in the terminal (assuming adequate permissions for the user). Also make sure to replace items in the <> brackets.

eksctl create cluster \
--name <cluster-name> \
--version 1.21 \
--region us-east-1 \
--node-type m5.large \
--nodes 1 \
--nodes-min 1 \
--nodes-max 1 \
--managed \
--tags environment=<env-type, e.g., dev, prod>


SEED stores its charts in the charts directory of the Github Repo. There are two main charts that are deployed when starting SEED on Kubernetes.

  • persistentvolumes - these are the volumes to store SEED media data and SEED Postgres data

  • seed - this stores all of the other deployment and service files for the application

Unlike persistentvolumes, the seed charts must be modified with user environment variables that will be forwarded to the docker container for deployment. Before deployment, the user MUST set these variables to their desired values.

This chart contains the deployment specification for the SEED web container. Replace all the values in <>.

# Environment variables for the web container
- env:
    # AWS Email service variables to send emails to new users - can be removed if not using this functionality.
    - name: AWS_ACCESS_KEY_ID
      value: <access_key_id>
      value: <secret_access_key>
      value: us-west-2
      value: email.us-west-2.amazonaws.com
    - name: SERVER_EMAIL
      value: [email protected]
    # Django Variables
      value: config.settings.docker
    - name: SECRET_KEY
      value: <replace-secret-key>
    - name: SEED_ADMIN_ORG
      value: default
      value: <super-secret-password>
    - name: SEED_ADMIN_USER
      value: <[email protected]>
    # Postgres variables
    - name: POSTGRES_DB
      value: seed
      value: <super-secret-password> # must match db-postgres-deployment.yaml and web-celery-deployment.yaml
    - name: POSTGRES_PORT
      value: "5432"
    - name: POSTGRES_USER
      value: seeduser
    # Bsyncr analysis variables
      value: "5000"
      value: bsyncr
    # Sentry monitoring - remove if not applicable
    - name: SENTRY_JS_DSN
      value: <enter-dsn>
    - name: SENTRY_RAVEN_DSN
      value: <enter-dsn>
    # Google self registration security - remove if not applicable
      value: <reCAPTCHA-site-key>
      value: <reCAPTCHA-key>
    image: seedplatform/seed:<insert deployment image version>
    #versions can be found here https://github.com/SEED-platform/seed/releases/tag/v2.9.3

This chart contains the deployment specification for the Celery container to connect to Postgres. Replace the Postgres password to match web-deployment.

  value: <super-secret-password>  # must match db-postgres-deployment.yaml and web-celery-deployment.yaml

This chart contains the deployment specification for the bsyncr analysis server. Request a NOAA token from this website.

- name: NOAA_TOKEN
  value: <token>


Once you are connected to your cluster and have your settings configured with the environment variables of you choice in the charts, you are ready to deploy the app. First, make sure that the correct context is selected which is needed if there is more than one cluster:

kubectl config get-contexts
kubectl config use-context <context-name>

Deploy the site using the helm commands in the root of the charts directory.

  • helm install --generate-name persistentvolumes

  • helm install --generate-name seed

You will be able to see SEED coming online with statuses like container creating, and running with:

  • kubectl get all

Once all of the pods are running you will be able to hit the external ingress through the URL listed in the web service information. It should look something like

service/web           LoadBalancer   <my-unique-url>   80:32291/TCP

Managing Existing Clusters

Upgrade/Redeploy the Helm Stack

To upgrade or dedeploy a helm chart, first find the helm release that you want to upgrade, then run the upgrade with the selected chart.

helm list
helm upgrade <cluster-name> ./seed

Managing the Kubernetes Cluster (AWS Specific)

Enable kubectl to talk to one of the created clusters by running the following command in the terminal after configuring the AWS credentials and cli.

aws eks --region <aws-region> update-kubeconfig --name <cluster-name>

Logging In

After a successful deployment in order to login you will need to create yourself as a user in the web container. To do this, we will exec into the container and run some Django commands. * View all deployments and services, kubectl get all * kubectl get pods * kubectl exec -it <pod-id> -- bash

Now that we are in the container, we can make a user. .. code-block:: bash

./manage.py create_default_user –username=admin@my.org –organization=seedorg –password=badpass

You can now use these credentials to log in to the SEED website.

Update web and web-celery

The command below will restart the pods and re-pull the docker images.

kubectl rollout restart deployment web && kubectl rollout restart deployment web-celery

Other Resources

Common kubectl actions can be found here