mlflow/mlflow · error · MlflowException

INVALID_PARAMETER_VALUE

INVALID_PARAMETER_VALUE

Error message

Invalid pip_requirements for job function: {pip_requirements}, parsing error: {e!r}

What it means

`mlflow.server.jobs.job` validates the `pip_requirements` argument by parsing it with pip's requirement parser before creating a job. If any requirement string cannot be parsed (malformed specifier, bad URL, invalid extras, non-string entry), an MlflowException with INVALID_PARAMETER_VALUE is raised naming the offending list and the parse error.

Source

Thrown at mlflow/server/jobs/__init__.py:111

            relative file references such as "-r requirements.txt" are not supported.
        exclusive: (optional) If True, only one instance of this job with the same params
            can run at a time. If a list of parameter names is provided, only those
            parameters are considered when determining exclusivity. Default is False.
    """
    from mlflow.utils import PYTHON_VERSION
    from mlflow.utils.requirements_utils import _parse_requirements
    from mlflow.version import VERSION

    if not python_version and not pip_requirements:
        python_env = None
    else:
        python_version = python_version or PYTHON_VERSION
        try:
            pip_requirements = [
                req.req_str for req in _parse_requirements(pip_requirements, is_constraint=False)
            ]
        except Exception as e:
            raise MlflowException.invalid_parameter_value(
                f"Invalid pip_requirements for job function: {pip_requirements}, "
                f"parsing error: {e!r}"
            )
        if mlflow_home := os.environ.get("MLFLOW_HOME"):
            # Append MLflow dev version dependency (for testing)
            pip_requirements += [mlflow_home]
        else:
            pip_requirements += [f"mlflow=={VERSION}"]

        python_env = _PythonEnv(
            python=python_version,
            dependencies=pip_requirements,
        )

    def decorator(fn: Callable[P, R]) -> Callable[P, R]:
        fn._job_fn_metadata = JobFunctionMetadata(
            name=name,
            fn_fullname=f"{fn.__module__}.{fn.__name__}",

View on GitHub (pinned to 6a27f2decc)

Solutions

  1. Inspect the quoted pip_requirements in the message and fix the malformed entry
  2. Validate requirements locally with `pip install --dry-run -r` or pkg_resources before submitting
  3. Pass a list of well-formed PEP 508 requirement strings (e.g. ['mlflow>=2.0', 'pandas==2.1.0'])

Example fix

// before
@job(pip_requirements=["numpy>=", "pandas"])
// after
@job(pip_requirements=["numpy>=1.24", "pandas==2.1.0"])
Defensive patterns

Strategy: validation

Validate before calling

from packaging.requirements import Requirement

def validate_pip_requirements(reqs):
    for r in reqs:
        try:
            Requirement(r)
        except Exception as e:
            raise ValueError(f"Bad requirement {r!r}: {e}")

Prevention

When it happens

Trigger: Calling mlflow.server.jobs.job(...) (or the decorated job submission flow) with pip_requirements containing a malformed string like 'numpy>=', 'invalid package!', a local path with bad syntax, or non-string values.

Common situations: Hand-built requirement strings with typos or missing version comparators; passing a requirements.txt file path instead of parsed lines; interpolating variables that produce empty or invalid specifiers.

Related errors

AI-assisted analysis of mlflow/mlflow@6a27f2decc (2026-08-29). Data as JSON: /api/errors/2c9c0ea8d5384204. Report an issue: GitHub.