6 replies

Drizzle & Planetscale data migration workflow

Tagged as "databases" because there's no Drizzle tag.

I'm running T3 + Drizzle + Planetscale. I've been using a planetscale branch ->

db:push

db:push

-> deploy request workflow for managing schema changes, which I like a lot. This doesn't handle modifications to existing data however, so I still need a way to run a set of SQL commands against an existing database exactly once.

Drizzle-kit claims to be able to do this with

drizzle-kit generate:mysql --custom

drizzle-kit generate:mysql --custom

, but this doesn't generate an empty SQL file for me like the docs say it will. On first run it creates all the tables in my schema, and subsequently it says there aren't any changes to make and it does nothing. (In other words it seems to ignore the

--custom

--custom

flag).

Do I have to implement this myself? Make a migrations table, track which migrations have been run, call the ones that haven't? Is there some other way? I can't be the only one with this issue.

Solution

Update: I figured out a workaround:

- Make a random schema change
- run

drizzle-kit generate

drizzle-kit generate

, and since drizzle thinks a schema change is needed it will actually generate a SQL file
- Discard said schema change
- Replace generated SQL with whatever UPDATE statements I need
- call drizzle's provided migrator function as appropriate
- success

Pretty sure ignoring the