## TL;DR
Move the registry to a database backed store: the MLflow model registry does not work on a plain file store, so run the tracking server with a database backend store, migrate the existing runs, and point the registry operations at that server. This is an architecture limit, not a permissions error. File stores track experiments and runs but cannot provide registry semantics such as versioning and stage transitions.

```text
MLflow model registry failed error "INVALID_PARAMETER_VALUE: database backed store required for registry"
```

## Use this when
- The check or run fails with this symptom: an MLflow model registry fails with INVALID_PARAMETER_VALUE database backed store required for registry.
- You are working on an MLflow setup using a local file store that now needs the model registry and the failure repeats on the same data or configuration.
- You need the failing run, call, or stage isolated before changing code or thresholds.

## Not for this skill when
- A database backed server that is unreachable, where the registry error masks a connection problem to fix first.
- The error text names a different tool, stage, or root cause. Match the verbatim error first and use that skill.

## Steps
1. Check the tracking URI the client uses and whether a server with a database backend serves it. Verify: the current store type is confirmed
2. Start the MLflow server with a database backend store URI and a default artifact root. Verify: the server reports the database store on startup
3. Point clients at the server tracking URI instead of the local file path. Verify: runs and experiments resolve through the server
4. Register a test model. Verify: a model version is created in the registry
5. Plan the migration of existing file store runs the team still needs. Verify: required runs are accessible under the server

## Variant phrasings
### mlflow registry database backed store required INVALID_PARAMETER_VALUE
The shorter phrasing focused on the named failure.
### mlflow model registry failed file store error
The symptom phrasing a searcher tries next.
### mlflow registry database store setup fix
The variant naming the failing stage or tool directly.

Compatibility: MLflow tracking servers with SQL backend stores, model registry requirements

## Why it happens
The registry needs transactional, queryable storage for model versions and aliases, which the file store never implemented. Setups that begin as a single user file store hit this wall the moment the team starts registering models, usually mid project.

## Edge cases / pitfalls
- SQLite is a valid backend for small teams and is still a database store; the step up from files does not require a managed database on day one.
- Two stores in parallel, files for runs and a server for the registry, split the history; consolidate on the server instead.

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst_tHFyXNRHRzUsx3-NSCLgwg