DataTorch Helm Charts

Scale DataTorch on Kubernetes a production-grade container orchestrator.

Introduction

DataTorch is a scalable annotation platform build for sharing and creating datasets.

This chart bootstraps a DataTorch deployment on Kubernetes cluster using Helm v3 package manager.

Note: You must have a license key and the image pull secret for this chart to work.

Prerequisites

• Helm 3.0+
• Kubernetes 1.13+ (1.15+ for ssl)

Installing Chart

1. Add datatorch repo helm repo add datatorch https://charts.datatorch.io
2. Use regcred script to create image pull secret
3. Update values.yaml with FQDN's, database connections and license
4. Deploy the instance with helm install (shown below)

To install the chart with release name my-release:

helm install my-release datatorch/datatorch

The command deploys DataTorch on kubernetes cluster in the default configuration.

Uninstalling Chart

To uninstall/delete the my-release deployment:

helm delete my-release

The command removes all the Kubernetes components associated with the chart and deletes the release.

SSL with NGINX and Cert-manager

Run the following commands to install cert-manager.

1. sh ssl/install-certmanager.sh
2. kubectl apply -f ssl/issuer-prod.yaml
3. kubectl apply -f ssl/issuer-staging.yaml
4. Add annotations and tls to ingress as shown in below.

Two issuers will be created (when testing used the letsencrypt-staging). Change the fields in each issuer to point to your email address. Once completed, you can apply the files to your cluster.

kubectl apply -f ./ssl/issuer-staging.yaml
kubectl apply -f ./ssl/issuer-prod.yaml

Add the following annotations to ingress

ingress:
  host: HOST_DOMAIN
  annotations:
    # ... other annotations
    cert-manager.io/issuer: "letsencrypt-prod"
    cert-manager.io/acme-challenge-type: http01
  tls:
    - secretName: TLS_SECRET_NAME
      hosts:
        - HOST_DOMAIN

pgbouncer

For Azure using managed postgres instance add pgbouncer to speed up connection.

The backend section has a extraContainers and extraVolumes section for this.