Did the team overwrite database vars?
UPDATE: I didn't know what I was doing
Hey I think I'm going crazy. Multiple of my database vars are randomly empty across different projects and services as of two days ago. Even stranger, one of the projects had variables cleared out on one environment but not the other.
This didn't happen to all environments, but it looks like specifically it was the ones with the older DB names such as "PGHOST" or "PGPASSWORD". The newer names such as "POSTGRES_USER" seem to have been untouched.
Am I crazy? Was this a mistake with the DB migration logic? It didn't kill anything super important due to being really careful surrounding the important DB migrations, but now I'm paranoid.
12 Replies
Project ID:
N/An/a
where these variables that where referencing legacy databases?
I know there was an issue with a polyfill a while back on older services, this one is from a fresh metabase install from about 3 weeks ago, it shows the private vars but no public ones. I don't have a specific project to show for this issue, as I had to overwrite all the variables again to make services work.
I own the metabase template, and I don't think it comes with the database publicly exposed
Overall I don't think this is a big issue, it just took down a dev server and a few other little things. How do I publically expose that DB? I need to connect to it to get a user token off of it.
I just checked, yes this is correct, it does not come with the database exposed so there is not going to be public variable values
please see the overview https://railway.app/template/metabase
I clicked enable TCP and got the public port and proxy and I still can't seem to connect to it via Datagrip.
have you redeployed the database?
I did not
I will try that
Redeploying works. I think it should do that as part of the process. I will leave that in the feedback channel.
hold off on that please
needing to redeploy is a bug on its own, the team is aware of that
Oh okay