Best Practices For Running Migrations In Self-Hosted Kubernetes Environments

[addefisher]

What is the recommended best practice for running database migrations via the manage.py script when running the Prowler app in a Kubernetes cluster?

Based on https://github.com/prowler-cloud/prowler/blob/master/api/Dockerfile and https://github.com/prowler-cloud/prowler/blob/master/api/docker-entrypoint.sh, it appears that migrations are run as part of container startup. However, I've had to restart the prowler-api container in order to get migrations to apply successfully.

I'm not sure if I should be

• Never attempting to run migrations explicitly, and just relying on docker-entrypoint.sh to handle this.
• Running migrations as a Kubernetes Job resource from a dedicated container.
• Running migrations from outside the Kubernetes cluster (for example from GitHub actions)

[HugoPBrito]

Hi @addefisher,

All three options you mentioned seem valid.

Nevertheless, we recommend delegating the responsibility to the entrypoint. Migrations should only be applied when we release a new version that includes new migrations.

I hope this answers your question.

[addefisher]

Thanks @HugoPBrito! I ended up using option 1. I just needed to add init containers to all the containers which depend on postgres so that they wait to start until the database is available.