I had to migrate my stash-box installation from the initial server.
Before I shut down the old server, I migrated the database to a docker container set of postgres and stash-box using the compose.yml from the docs and pre-made images.
I tested logins, content, images, and API keys and everything worked.
I got a new hosting service and had to do a bare-metal installation (similar to the original server) and when I migrated over the database, everything worked (logins, images, etc) but the API keys are broken now. Resetting a user’s API key resolves the “invalid or expired API key” error, but with nearly 1000 users, I would prefer to avoid having everyone reset their API keys.
Any ideas on why the API keys broke (or maybe they expired?)
The stash-box Metadata Provider hostnames and labels were not the same on the interim test system and the original host. (interim was localhost) so I think I ruled out that.
I installed three psql extension pg_trgm, pgcrypto, and bktree;
The original source and target hosts had postgresql 16.1 and libvips 8.7.1 compiled from source on both systems.
The original and interim systems used the default port, the target system uses a different port due to the host’s requirements.
Also the jwt_secret_key and session_store_key were not the same on the original or interim systems, so I didn’t consider that they are also different on the new system as being an issue either.
I wrote some instructions that more clearly explain how to do a bare metal installation and will share them after I’ve cleaned them up and sanitized them.
Any thoughts on what I can do before letting users back onto the new system where they’ll have to reset their API keys? It’s not a dealbreaker, but I’m wondering if there’s something larger broken. Logs don’t have errors, SQL didn’t have errors either on the database import.