postgres-statefulset

This is an example of using Kubernetes StatefulSets to get a Postgres instance running with replication enabled. This also uses the standard Postgres container. Blog article here

The work here is based off the official documentation here https://wiki.postgresql.org/wiki/Streaming_Replication

Configuration

1. Edit config/secret.yml with the Postgres database password and the replication password
2. Run kubectl apply -f config/secret.yml and then cd config && ./create_configmap.sh

Note, replication password is used to connect to the master and stream updates to the replica. It just needs to be a random password.

Running

Running this example is easy!

Start Master servers

Run kubectl apply -f statefulset-master.yml and wait for Master to be running

Start Master service (IMPORTANT or replica cannot find the master)

Run kubectl apply -f service.yml

Start Replica server

Run kubectl apply -f statefulset-replica.yml and wait for Replica to be running

If you run kubectl logs -f postgres-replica-0, you can see in the logs that it starts replication:

2019-01-08 05:07:01.035 UTC [24] LOG:  started streaming WAL from primary at 0/6000000 on timeline 1

That's it, you have a full Postgres master + replicating server that's ready to use in production.

Multiple Replicas

You can also set replicas to more than 1 if you want N replicas.

Motivation

I was using Helm to manage Postgres, but they switched to bitnami/postgresql instead of the standard postgres. Also, upon upgrading, I could not use my existing helm setup to get replication working.

Lately I've realized it's just easier to write my own definitions instead of messing around with Helm charts.