Deploy TimescaleDB 2.0 (multi node) to Kubernetes using Helm

Praveen Perera

January 31, 2021

1 min read

cheatsheet, dev-ops, helm, kubernetes, timescale, tutorial, walk-through

Prerequisites

Before proceeding you will need the follow:

  1. A running kubernetes clusters.
  2. Helm CLI installed
  3. KubeCTL installed

If you don’t have a kubernetes cluster, see my getting started with kubernetes with rancher and DigitalOcean managed kubernetes series

Deploying TimescaleDB 2.0 (Multi Node)

  1. Clone the timescale/timescaledb-kubernetes repo 
git clone [email protected]:timescale/timescaledb-kubernetes.git

  1. Delete everything except charts/timescaledb-multinode

  2. Rename this folder charts

  3. In this folder open the values.yaml file.

  4. Under image change value for repository to: timescale/timescaledb-ha

  5. Change the value for tag to: pg12.5-ts2.0.0-p0 or find the current latest one at: https://hub.docker.com/r/timescale/timescaledb

  6. In the project root folder create a file called .pgpass

  7. Open .pgpass and enter the following (use randomly generated passwords and create as many accounts as you need) 

*:*:*:postgres:DATAPASS-XXXXXXXXXXXXX
*:*:*:admin:ACCESSPASS-YYYYYYYYYYYYYY

  1. Open charts/templates/svc-timescaledb-access.yaml

  2. Change type: LoadBalancer to type: ClusterIP

  3. Go into terminal open project root folder and run 

 helm install timescale-db charts --namespace=default --set credentials.accessNode.superuser="<ACCESSPASS-YYYYYYYYYYYYYY>" --set credentials.dataNode.superuser="<DATAPASS-XXXXXXXXXXXXX>"

Accessing database locally

  1. Get the password for access node 
PGPASSWORD_POSTGRES=$(kubectl get secret --namespace default timescale-db-timescaledb-access -o jsonpath="{.data.password-superuser}" | base64 --decode)
  1. Get the name of the access node pod 
ACCESS_POD=$(kubectl get pod -o name --namespace default -l release=timescale-db,timescaleNodeType=access)
  1. Start a port forward from the access node 
kubectl port-forward $ACCESS_POD 7000:5432 -n=default
  1. Now you can access the database that’s in your cluster as if it were running locally. Use whatever port you’re forwarding to (7000 in this case). The host is localhost, the user is postgres and the password is $PGPASSWORD_POSTGRES from step one