mlflow/mlflow · error · MlflowException

RESOURCE_DOES_NOT_EXIST

RESOURCE_DOES_NOT_EXIST

Error message

Could not find experiment with name {experiment_name}

What it means

Raised by _get_permission_from_experiment_name when the tracking store has no experiment matching the supplied experiment_name. The auth middleware must resolve the experiment to evaluate permissions, so a missing name yields RESOURCE_DOES_NOT_EXIST.

Source

Thrown at mlflow/server/auth/__init__.py:911

    if MLFLOW_ENABLE_WORKSPACES.get():
        if workspace_name := workspace_context.get_request_workspace():
            user = store.get_user(username)
            perm = store.get_role_permission_for_resource(user.id, "workspace", "*", workspace_name)
            if perm is not None:
                return perm
            # Honor the default-workspace auto-grant when configured.
            if _user_inherits_default_workspace_grant(workspace_name):
                return get_permission(auth_config.default_permission)
        return NO_PERMISSIONS

    return get_permission(auth_config.default_permission)


def _get_permission_from_experiment_name() -> Permission:
    experiment_name = _get_request_param("experiment_name")
    store_exp = _get_tracking_store().get_experiment_by_name(experiment_name)
    if store_exp is None:
        raise MlflowException(
            f"Could not find experiment with name {experiment_name}",
            error_code=RESOURCE_DOES_NOT_EXIST,
        )
    username = authenticate_request().username

    return _get_role_permission_or_default(
        _role_permission_for(
            username=username,
            resource_type="experiment",
            resource_key=store_exp.experiment_id,
            workspace_lookup_id=store_exp.experiment_id,
            workspace_fetcher=_get_tracking_store().get_experiment,
            workspace_label="experiment",
        ),
    )


def _get_permission_from_run_id() -> Permission:

View on GitHub (pinned to 6a27f2decc)

Solutions

Call client.get_experiment_by_name(name) first and handle None before hitting the guarded endpoint
Use the experiment_id instead of the name if you have it
Verify the tracking URI points at the store containing the experiment
Check the exact name via the UI or search_experiments; names are matched literally

Example fix

// before
exp = client.get_experiment_by_name('My-Experiment')  # auth resolves name
// after
exp = client.get_experiment_by_name('My Experiment')
if exp is None: raise SystemExit(f'no such experiment: My Experiment')

Defensive patterns

Strategy: try-catch

Validate before calling

exp = client.get_experiment_by_name(experiment_name)
if exp is None:
    raise SystemExit(f'Experiment {experiment_name!r} not found in {client.tracking_uri}')

Try / catch

try:
    do_authenticated_call(experiment_name=name)
except MlflowException as e:
    if e.error_code == 'RESOURCE_DOES_NOT_EXIST' and 'experiment' in str(e):
        exps = client.search_experiments(filter_string=f"name = '{name}'")
        if not exps:
            raise SystemExit(f'no experiment named {name}')
    else:
        raise

Prevention

Resolve names to experiment_id once and reuse the ID
Verify MLFLOW_TRACKING_URI points at the store that holds the experiment
Remember deleted experiments no longer resolve by name

When it happens

Trigger: Requesting an experiment permission check (validate_can_read_experiment_by_name path) with an experiment_name that was never created, was deleted, or is named differently in the configured tracking store (e.g. different backend or workspace).

Common situations: Typos in experiment names in notebooks/scripts; pointing a client at a different tracking store (staging vs prod) where the name doesn't exist; deleted experiments (soft-deleted ones no longer resolve); case-sensitivity mismatches.

Understand the failure class

Background: 'Could not be found', 'does not exist', 'not found in database': the resource-not-found family when an ID, slug, key, or URI lookup comes back empty — this error's family across 20 libraries.

Related errors

AI-assisted analysis of mlflow/mlflow@6a27f2decc (2026-08-29). Data as JSON: /api/errors/62e1bdb797989141. Report an issue: GitHub.