Database migration

By default, Coder deploys a built-in database in the installation's Kubernetes namespace. We recommend using this database only for evaluation purposes.

At the end of your evaluation period, you may need to migrate the data from the built-in database to an out-of-cluster PostgreSQL database for production use. This article will walk you through the process of doing so.

Requirements

• You must be a cluster admin for your Kubernetes cluster.

• Azure database users: if you're using Azure Database for PostgreSQL, note that Coder only works with Single Server; the Flexible Server (Preview) and Hyperscale (Citus) options do not support the TimescaleDB extension required.

Migration Steps

1. Access the database pod and dump the database into a file:

   kubectl exec -it statefulset/timescale -n coder -- pg_dump -U coder -d coder > backup.sql
   

   Copy

2. Optional: If your database is large, you can truncate Coder's telemetry, metrics, and audit log data to reduce the file size:

   TRUNCATE metric_events;
   TRUNCATE environment_stats;
   TRUNCATE audit_logs;
   

   Copy

3. Access your PostgreSQL instance and create user coder and database coder

4. Import the data you exported in the first step into your external database:

   psql -U coder < backup.sql
   

   Copy

5. Connect your Coder instance to the database:

   helm upgrade --reuse-values -n coder coder coder/coder \
       --set postgres.default.enable=false \
       --set postgres.host=<HOST_ADDRESS> \
       --set postgres.port=<PORT_NUMBER> \
       --set postgres.user=<DATABASE_USER> \
       --set postgres.database=<DATABASE_NAME> \
       --set postgres.passwordSecret=<secret-name> \
       --set postgres.sslMode=require
   

   Copy

6. Optional: If you'd like to delete the Timescale persistent volume, run:

   kubectl delete pvc timescale-data-timescale-0 -n coder
   

   Copy

At this point, you should be able to log in to your Coder deployment successfully.