mlflow/mlflow · error · Exception

You need to set either {_JOHNSNOWLABS_ENV_HEALTHCARE_SECRET}

Error message

You need to set either {_JOHNSNOWLABS_ENV_HEALTHCARE_SECRET} or {_JOHNSNOWLABS_ENV_VISUAL_SECRET} environment variable. Please contact John Snow Labs to get one.

What it means

get_default_pip_requirements needs to build a wheel URI for the John Snow Labs healthcare/visual NLP libraries, which requires an JSL secret. It raises when neither the healthcare secret nor the visual secret environment variable is set, telling the user to obtain one from John Snow Labs.

Source

Thrown at mlflow/johnsnowlabs/__init__.py:148

    # if json license is detected, we parse it and set the env vars
    loaded_license = json.loads(os.environ[_JOHNSNOWLABS_ENV_JSON_LICENSE_KEY])
    os.environ.update({k: str(v) for k, v in loaded_license.items() if v is not None})


def get_default_pip_requirements():
    """
    Returns:
        A list of default pip requirements for MLflow Models produced by this flavor.
        Calls to :func:`save_model()` and :func:`log_model()` produce a pip environment
        that, at minimum, contains these requirements.
    """
    from johnsnowlabs import settings

    if (
        _JOHNSNOWLABS_ENV_HEALTHCARE_SECRET not in os.environ
        and _JOHNSNOWLABS_ENV_VISUAL_SECRET not in os.environ
    ):
        raise Exception(
            f"You need to set either {_JOHNSNOWLABS_ENV_HEALTHCARE_SECRET} "
            f"or {_JOHNSNOWLABS_ENV_VISUAL_SECRET} environment variable. "
            f"Please contact John Snow Labs to get one."
        )

    _SPARK_NLP_JSL_WHEEL_URI = (
        "https://pypi.johnsnowlabs.com/{secret}/spark-nlp-jsl/spark_nlp_jsl-"
        + f"{settings.raw_version_medical}-py3-none-any.whl"
    )

    _SPARK_NLP_VISUAL_WHEEL_URI = (
        "https://pypi.johnsnowlabs.com/{secret}/spark-ocr/"
        f"spark_ocr-{settings.raw_version_ocr}-py3-none-any.whl"
    )

    deps = [
        f"johnsnowlabs_for_databricks=={settings.raw_version_jsl_lib}",
        _get_pinned_requirement("pyspark"),

View on GitHub (pinned to 6a27f2decc)

Solutions

Set the healthcare secret env var (from John Snow Labs)
Or set the visual NLP secret env var if using Visual NLP
Contact John Snow Labs to obtain a secret if you don't have one

Example fix

// before
mlflow.johnsnowlabs.save_model(...)
// after
export JOHNSNOWLABS_HEALTHCARE_SECRET=xxxx
mlflow.johnsnowlabs.save_model(...)

Defensive patterns

Strategy: validation

Validate before calling

import os
if not (os.environ.get("JOHNSNOWLABS_HEALTHCARE_SECRET") or os.environ.get("JOHNSNOWLABS_VISUAL_SECRET")):
    raise RuntimeError("Set either the JSL healthcare or visual NLP secret env var")

Try / catch

try:
    requirements = mlflow.johnsnowlabs.get_default_pip_requirements()
except Exception as e:
    if "environment variable" in str(e):
        raise RuntimeError("Provide a JSL healthcare/visual secret from John Snow Labs") from e
    raise

Prevention

Inject both license and secret env vars together in CI/Docker
Know which JSL product (healthcare vs visual) you use and set the matching secret
Contact John Snow Labs early to obtain secrets for new environments

When it happens

Trigger: Calling get_default_conda_env / get_default_pip_requirements (indirectly via save_model/log_model metadata) without either secret env var set.

Common situations: Saving a JSL model without healthcare/visual NLP credentials; shared environments where only the license.json was provided but not the download secret.

Understand the failure class

Background: "environment variable is not set" and "Missing keys in environment" errors: what missing required env var messages mean and how to fix them — this error's family across 28 libraries.

Related errors

AI-assisted analysis of mlflow/mlflow@6a27f2decc (2026-08-29). Data as JSON: /api/errors/3f1fb5f9c3f6cca5. Report an issue: GitHub.