Need a story for database migrations

[GabeIsman]

I just ran into an issue where I wanted to make some fields Nullable. I updated the models to have the appropriate columns, but couldn't figure out how to migrate the DB schema. Looking into it a little bit it seems like alembic is probably the preferred tool for this. I tried to get it set up but stumbled a bit configuring it on some issues that I think would be easy to resolve if I were a little more familiar with the python module system. Would be great if someone with a little more familiarity could take this on and save me the headache!

In the meantime I'll just be blowing away my database and re-running initializedb to 'migrate' my schema.

[thequbit]

@gabeisman to be honest, I always just blow away the local sqlite db when I change the schema. I then update the tests / test data generation to reflect the schema change. I'm sure @ryansb has some thoughts on this but my take is if it isn't production, just blow it away.

[GabeIsman]

@thequbit That's totally fine (if less than ideal, sometimes test data takes a long time to set up), but we still need a story for production.

[thequbit]

Definitely will need a story for production. I'm not super familiar with the flow for bridge schema's for PostgreSQL, but it's certainly a well documented thing. For the time being, I think the extras field that will take json will keep us moving forward in development ( see #88 ).

As for setting up test data, I'll create a ticket for that so we can just run a script and have it load it in.

[ryansb]

There's an app for that (tm)!

In OpenStack we use alembic to generate migrations between sqla schema versions.

It pretty well resolves the migration story, but I don't feel like that's something we need to do "right now" as we don't have a production environment. I'd rather put the work into a loader for test data for now, and then build migrations when we have a "real" deployment. #JustInTimeDevOps