10 months ago
I'm using a Dockerfile as my deployment method.
Usually platforms (Heroku, Fly, Render, etc) allow you to run something like railway console which spins up a container in the same environment to allow you to run short-lived tasks like data migrations.
I tried using railway ssh after linking to the correct service and I receive:
➜ proj git:(master) ✗ railway ssh
Expected welcome message, received: ServerMessage { type: "error", payload: ServerPayload { data: Empty, message: "No deployment found to connect to", code: None } }Any help would be greatly appreciated.
7 Replies
10 months ago
Do you wanna do data migrations or do you just used it as an example?
10 months ago
Data migrations should be done in a pre-deploy command
10 months ago
Also, to be able to use SSH in a service there's some caveats, the built image must include a shell bash and it must be currently running
I think that's contextual. Sometimes migrations are better run as a post-deploy step (for example, if they depend on some state of the application).
In any case, we have a data migration that needs to be run as a once-off. Is SSH the only way?
10 months ago
SSH or railway run to ran locally but using railway environment variables