This repo demonstrates deploying Metabase as a docker container in Google Kubernetes Engine (GKE).

Specifically, this Kubernetes (k8s) configuration comprises:

• k8-deployment.yaml
  Deployment of pod template with metabase docker image and "sidecar" container running GCP cloud sql proxy for database access.

• k8-service.yaml
  NodePort service to open automatic "high port" on each node that the loadbalancer can proxy

• k8-ingress.yaml
  Ingress resource that loadbalances traffic and terminates SSL/TLS. This configuration utilizes the "gce" loadbalancer class

• Download the GKE credentials:

gcloud container clusters get-credentials <cluster_name>

• Apply GKE credentials as kubectl context:

  • List available contexts:

  kubectl config get-contexts

  • Set context:

  kubectl config use-context <context>

• Reserve a GCP static ip:

gcloud compute addresses create metabase-ip --global

• Modify ./metabase/k8-deployment.yaml:

  • replace <INSTANCE_CONNETION_STRING> with your cloud_sql_proxy connection string
  • Update all environment vars prefixed MB_DB_... to match your cloud_sql_proxy configuration

• Create ./secrets/cloudsql-gcp-key.json with GCP service account key

• Create metabase user credentials as text files:

echo -n "metabase" > secrets/metabase-db-username.txt
echo -n "password" > secrets/metabase-db-password.txt

• Store the SSL/TLS cert and private key as secrets/tls.crt and secrets/tls.key

Provided shell script k8.sh can create and delete this Kuberentes configuration in the cluster.

• To create secrets, issue k8.sh secrets
• Deploy metabase to the cluster with k8.sh create
• Delete everything with k8.sh delete

• Official Metabase Container Instructions