Configuration¶

This section contains detailed information about configuration, for which the relevant API documentation can be found in kedro.config.ConfigLoader.

Configuration root¶

We recommend that you keep all configuration files in the conf directory of a Kedro project. However, if you prefer, point Kedro to any other directory and change the configuration paths by setting the CONF_SOURCE variable in src/<package_name>/settings.py as follows:

CONF_SOURCE = "new_conf"

You can also specify a source directory for the configuration files at run time using the kedro run CLI command with the --conf-source flag as follows:

kedro run --conf-source=<path-to-new-conf-directory>

If you’re using the OmegaConfigLoader you can also read configuration from a compressed file in tar.gz or zip format. The two commands below are examples of how to reference a tar.gz or zip file:

kedro run --conf-source=<path-to-compressed-file>.tar.gz

kedro run --conf-source=<path-to-compressed-file>.zip

To compress your configuration you can either leverage Kedro’s kedro package command which builds the package into the dist/ folder of your project, and creates one .egg file, one .whl file, as well as a tar.gz file containing the project configuration. This compressed version of the config files excludes any files inside your local directory. Alternatively you can run the command below to create a tar.gz file:

tar --exclude=local/*.yml -czf <my_conf_name>.tar.gz --directory=<path-to-conf-dir> <conf-dir>

Or the following command to create a zip file:

zip -x <conf-dir>/local/** -r <my_conf_name>.zip <conf-dir>

Note that for both the tar.gz and zip file the following structure is expected:

<conf_dir>
├── base               <-- the files inside be different, but this is an example of a standard Kedro structure.
│   └── parameters.yml
│   └── catalog.yml
└── local              <-- the top level local directory is required, but no files should be inside when distributed.
└── README.md          <-- optional but included with the default Kedro conf structure.

Local and base configuration environments¶

Kedro-specific configuration (e.g., DataCatalog configuration for IO) is loaded using the ConfigLoader class:

from kedro.config import ConfigLoader
from kedro.framework.project import settings

conf_path = str(project_path / settings.CONF_SOURCE)
conf_loader = ConfigLoader(conf_source=conf_path, env="local")
conf_catalog = conf_loader["catalog"]

This recursively scans for configuration files firstly in the conf/base/ (base being the default environment) and then in the conf/local/ (local being the designated overriding environment) directory according to the following rules:

Either of the following is true:
- filename starts with catalog
- file is located in a sub-directory whose name is prefixed with catalog
And file extension is one of the following: yaml, yml, json, ini, pickle, xml or properties for the ConfigLoader and TemplatedConfigLoader or yaml, yml, or json for the OmegaConfigLoader.

This logic is specified by config_patterns in the ConfigLoader and TemplatedConfigLoader classes. By default those patterns are set as follows for the configuration of catalog, parameters, logging and credentials:

config_patterns = {
    "catalog": ["catalog*", "catalog*/**", "**/catalog*"],
    "parameters": ["parameters*", "parameters*/**", "**/parameters*"],
    "credentials": ["credentials*", "credentials*/**", "**/credentials*"],
    "logging": ["logging*", "logging*/**", "**/logging*"],
}

The configuration patterns can be changed by setting the CONFIG_LOADER_ARGS variable in src/<package_name>/settings.py. You can change the default patterns as well as add additional ones, for example, for Spark configuration files. This example shows how to load parameters if your files are using a params naming convention instead of parameters and how to add patterns to load Spark configuration:

CONFIG_LOADER_ARGS = {
    "config_patterns": {
        "spark": ["spark*/"],
        "parameters": ["params*", "params*/**", "**/params*"],
    }
}

You can also bypass the configuration patterns and set configuration directly on the instance of a config loader class. You can bypass the default configuration (catalog, parameters, credentials, and logging) as well as additional configuration.

from kedro.config import ConfigLoader
from kedro.framework.project import settings

conf_path = str(project_path / settings.CONF_SOURCE)
conf_loader = ConfigLoader(conf_source=conf_path, env="local")

# Bypass configuration patterns by setting the key and values directly on the config loader instance.
conf_loader["catalog"] = {"catalog_config": "something_new"}

Configuration information from files stored in base or local that match these rules is merged at runtime and returned as a config dictionary:

If any two configuration files located inside the same environment path (conf/base/ or conf/local/ in this example) contain the same top-level key, load_config will raise a ValueError indicating that the duplicates are not allowed.
If two configuration files have duplicate top-level keys but are in different environment paths (one in conf/base/, another in conf/local/, for example) then the last loaded path (conf/local/ in this case) takes precedence and overrides that key value. ConfigLoader.get will not raise any errors - however, a DEBUG level log message will be emitted with information on the overridden keys.

When using the default ConfigLoader or the TemplatedConfigLoader, any top-level keys that start with _ are considered hidden (or reserved) and are ignored after the config is loaded. Those keys will neither trigger a key duplication error nor appear in the resulting configuration dictionary. However, you can still use such keys, for example, as YAML anchors and aliases.

Additional configuration environments¶

In addition to the two built-in local and base configuration environments, you can create your own. Your project loads conf/base/ as the bottom-level configuration environment but allows you to overwrite it with any other environments that you create, such as conf/server/ or conf/test/. To use additional configuration environments, run the following command:

kedro run --env=test

If no env option is specified, this will default to using the local environment to overwrite conf/base.

If, for some reason, your project does not have any other environments apart from base, i.e. no local environment to default to, you must customise the configuration loader you’re using to take default_run_env="base" in the constructor and then specify your custom config loader subclass in src/<package_name>/settings.py under the CONFIG_LOADER_CLASS key. Below is an example of such a custom class. If you’re using the TemplatedConfigLoader or the OmegaConfigLoader you need to use either of those as the class you are subclassing.

# src/<package_name>/custom_config.py

from kedro.config import ConfigLoader
from typing import Any, Dict


class CustomConfigLoader(ConfigLoader):
    def __init__(
        self,
        conf_source: str,
        env: str = None,
        runtime_params: Dict[str, Any] = None,
        default_run_env: str = "base",
    ):
        super().__init__(
            conf_source=conf_source,
            env=env,
            runtime_params=runtime_params,
            default_run_env=default_run_env,
        )

And then you can import your CustomConfigLoader from settings.py:

# settings.py
from package_name.custom_configloader import CustomConfigLoader

CONFIG_LOADER_CLASS = CustomConfigLoader

If you set the KEDRO_ENV environment variable to the name of your environment, Kedro will load that environment for your kedro run, kedro ipython, kedro jupyter notebook and kedro jupyter lab sessions:

export KEDRO_ENV=test

Note

If you both specify the KEDRO_ENV environment variable and provide the --env argument to a CLI command, the CLI argument takes precedence.

Template configuration¶

Kedro also provides an extension TemplatedConfigLoader class that allows you to template values in configuration files. To apply templating in your project, set the CONFIG_LOADER_CLASS constant in your src/<package_name>/settings.py:

from kedro.config import TemplatedConfigLoader  # new import

CONFIG_LOADER_CLASS = TemplatedConfigLoader

Globals¶

When using the TemplatedConfigLoader you can provide values in the configuration template through a globals file or dictionary.

Let’s assume the project contains a conf/base/globals.yml file with the following contents:

bucket_name: "my_s3_bucket"
key_prefix: "my/key/prefix/"

datasets:
    csv: "pandas.CSVDataSet"
    spark: "spark.SparkDataSet"

folders:
    raw: "01_raw"
    int: "02_intermediate"
    pri: "03_primary"
    fea: "04_feature"

To point your TemplatedConfigLoader to the globals file, add it to the the CONFIG_LOADER_ARGS variable in src/<package_name>/settings.py:

CONFIG_LOADER_ARGS = {"globals_pattern": "*globals.yml"}

Alternatively, you can declare which values to fill in the template through a dictionary. This dictionary could look like the below:

{
    "bucket_name": "another_bucket_name",
    "non_string_key": 10,
    "key_prefix": "my/key/prefix",
    "datasets": {"csv": "pandas.CSVDataSet", "spark": "spark.SparkDataSet"},
    "folders": {
        "raw": "01_raw",
        "int": "02_intermediate",
        "pri": "03_primary",
        "fea": "04_feature",
    },
}

To point your TemplatedConfigLoader to the globals dictionary, add it to the CONFIG_LOADER_ARGS variable in src/<package_name>/settings.py:

CONFIG_LOADER_ARGS = {
    "globals_dict": {
        "bucket_name": "another_bucket_name",
        "non_string_key": 10,
        "key_prefix": "my/key/prefix",
        "datasets": {"csv": "pandas.CSVDataSet", "spark": "spark.SparkDataSet"},
        "folders": {
            "raw": "01_raw",
            "int": "02_intermediate",
            "pri": "03_primary",
            "fea": "04_feature",
        },
    }
}

If you specify both globals_pattern and globals_dict in CONFIG_LOADER_ARGS, the contents of the dictionary resulting from globals_pattern are merged with the globals_dict dictionary. In case of conflicts, the keys from the globals_dict dictionary take precedence.

Now the templating can be applied to the configuration. Here is an example of a templated conf/base/catalog.yml file:

raw_boat_data:
    type: "${datasets.spark}"  # nested paths into global dict are allowed
    filepath: "s3a://${bucket_name}/${key_prefix}/${folders.raw}/boats.csv"
    file_format: parquet

raw_car_data:
    type: "${datasets.csv}"
    filepath: "s3://${bucket_name}/data/${key_prefix}/${folders.raw}/${filename|cars.csv}"  # default to 'cars.csv' if the 'filename' key is not found in the global dict

Under the hood, TemplatedConfigLoader uses JMESPath syntax to extract elements from the globals dictionary.

Jinja2 support¶

From version 0.17.0, TemplateConfigLoader also supports the Jinja2 template engine alongside the original template syntax. Below is an example of a catalog.yml file that uses both features:

{% for speed in ['fast', 'slow'] %}
{{ speed }}-trains:
    type: MemoryDataSet

{{ speed }}-cars:
    type: pandas.CSVDataSet
    filepath: s3://${bucket_name}/{{ speed }}-cars.csv
    save_args:
        index: true

{% endfor %}

When parsing this configuration file, TemplateConfigLoader will:

Read the catalog.yml and compile it using Jinja2
Use a YAML parser to parse the compiled config into a Python dictionary
Expand ${bucket_name} in filepath using the globals_pattern and globals_dict arguments for the TemplateConfigLoader instance, as in the previous examples

The output Python dictionary will look as follows:

{
    "fast-trains": {"type": "MemoryDataSet"},
    "fast-cars": {
        "type": "pandas.CSVDataSet",
        "filepath": "s3://my_s3_bucket/fast-cars.csv",
        "save_args": {"index": True},
    },
    "slow-trains": {"type": "MemoryDataSet"},
    "slow-cars": {
        "type": "pandas.CSVDataSet",
        "filepath": "s3://my_s3_bucket/slow-cars.csv",
        "save_args": {"index": True},
    },
}

Warning

Although Jinja2 is a very powerful and extremely flexible template engine, which comes with a wide range of features, we do not recommend using it to template your configuration unless absolutely necessary. The flexibility of dynamic configuration comes at a cost of significantly reduced readability and much higher maintenance overhead. We believe that, for the majority of analytics projects, dynamically compiled configuration does more harm than good.

Configuration with OmegaConf¶

OmegaConf is a Python library for configuration. It is a YAML-based hierarchical configuration system with support for merging configurations from multiple sources. From Kedro 0.18.5 you can use the OmegaConfigLoader which uses OmegaConf under the hood to load data.

Note

OmegaConfigLoader is under active development and will be available from Kedro 0.18.5. New features will be added in future releases. Let us know if you have any feedback about the OmegaConfigLoader or ideas for new features.

The OmegaConfigLoader can load YAML and JSON files. Acceptable file extensions are .yml, .yaml, and .json. By default, any configuration files used by the config loaders in Kedro are .yml files.

To use the OmegaConfigLoader in your project, set the CONFIG_LOADER_CLASS constant in your src/<package_name>/settings.py:

from kedro.config import OmegaConfigLoader  # new import

CONFIG_LOADER_CLASS = OmegaConfigLoader

Templating for parameters¶

Templating or variable interpolation, as it’s called in OmegaConf, for parameters works out of the box if one condition is met: the name of the file that contains the template values must follow the same config pattern specified for parameters. By default, the config pattern for parameters is: ["parameters*", "parameters*/**", "**/parameters*"]. Suppose you have one parameters file called parameters.yml containing parameters with omegaconf placeholders like this:

model_options:
  test_size: ${data.size}
  random_state: 3

and a file containing the template values called parameters_globals.yml:

data:
  size: 0.2

Since both of the file names (parameters.yml and parameters_globals.yml) match the config pattern for parameters, the OmegaConfigLoader will load the files and resolve the placeholders correctly.

Custom template resolvers¶

Omegaconf provides functionality to register custom resolvers for templated values. You can use these custom resolves within Kedro by extending the OmegaConfigLoader class. The example below illustrates this:

from kedro.config import OmegaConfigLoader
from omegaconf import OmegaConf
from typing import Any, Dict


class CustomOmegaConfigLoader(OmegaConfigLoader):
    def __init__(
        self,
        conf_source: str,
        env: str = None,
        runtime_params: Dict[str, Any] = None,
    ):
        super().__init__(
            conf_source=conf_source, env=env, runtime_params=runtime_params
        )

        # Register a customer resolver that adds up numbers.
        self.register_custom_resolver("add", lambda *numbers: sum(numbers))

    @staticmethod
    def register_custom_resolver(name, function):
        """
        Helper method that checks if the resolver has already been registered and registers the
        resolver if it's new. The check is needed, because omegaconf will throw an error
        if a resolver with the same name is registered twice.
        Alternatively, you can call `register_new_resolver()` with  `replace=True`.
        """
        if not OmegaConf.has_resolver(name):
            OmegaConf.register_new_resolver(name, function)

You can then use the custom “add” resolver in your parameters.yml as follows:

model_options:
  test_size: ${add:1,2,3}
  random_state: 3

Environment variables for credentials¶

The OmegaConfigLoader enables you to load credentials from environment variables. To achieve this you have to use the omegaconf oc.env resolver. This is an example of how you can access credentials from environment variables in credentials.yml:

dev_s3:
  client_kwargs:
    aws_access_key_id: ${oc.env:AWS_ACCESS_KEY_ID}
    aws_secret_access_key: ${oc.env:AWS_SECRET_ACCESS_KEY}

Note

Note that you can only use the resolver in credentials.yml and not in catalog or parameter files. This is because we do not encourage the usage of environment variables for anything other than credentials.

Parameters¶

Load parameters¶

Parameters project configuration can be loaded by any of the configuration loader classes: ConfigLoader, TemplatedConfigLoader, and OmegaConfigLoader. The following examples will all make use of the default ConfigLoader class.

from kedro.config import ConfigLoader
from kedro.framework.project import settings

conf_path = str(project_path / settings.CONF_SOURCE)
conf_loader = ConfigLoader(conf_source=conf_path, env="local")
parameters = conf_loader["parameters"]

This will load configuration files from any subdirectories in conf that have a filename starting with parameters, or are located inside a folder with name starting with parameters.

Note

Since local is set as the environment, the configuration path conf/local takes precedence in the example above. Hence any overlapping top-level keys from conf/base will be overwritten by the ones from conf/local.

Calling conf_loader[key] in the example above will throw a MissingConfigException error if no configuration files match the given key. If this is a valid workflow for your application, you can handle it as follows:

from kedro.config import ConfigLoader, MissingConfigException
from kedro.framework.project import settings

conf_path = str(project_path / settings.CONF_SOURCE)
conf_loader = ConfigLoader(conf_source=conf_path, env="local")

try:
    parameters = conf_loader["parameters"]
except MissingConfigException:
    parameters = {}

Note

The kedro.framework.context.KedroContext class uses the approach above to load project parameters.

Parameters can then be used on their own or fed in as function inputs.

Specify parameters at runtime¶

Kedro also allows you to specify runtime parameters for the kedro run CLI command. To do so, use the --params command line option and specify a comma-separated list of key-value pairs that will be added to KedroContext parameters and made available to pipeline nodes. Each key-value pair is split on the first colon or equals sign. Following examples are both valid commands:

kedro run --params=param_key1:value1,param_key2:2.0  # this will add {"param_key1": "value1", "param_key2": 2} to parameters dictionary

kedro run --params=param_key1=value1,param_key2=2.0

Values provided in the CLI take precedence and overwrite parameters specified in configuration files. Parameter keys are always treated as strings. Parameter values are converted to a float or an integer number if the corresponding conversion succeeds; otherwise, they are also treated as string.

If any extra parameter key and/or value contains spaces, wrap the whole option contents in quotes:

kedro run --params="key1=value with spaces,key2=value"

Since key-value pairs are split on the first colon or equals sign, values can contain colons/equals signs, but keys cannot. These are valid CLI commands:

kedro run --params=endpoint_url:https://endpoint.example.com

kedro run --params=endpoint_url=https://endpoint.example.com

Use parameters¶

Say you have a set of parameters you’re playing around with that specify modelling hyperparameters. You can declare these in one place, for instance conf/base/parameters.yml, so that you isolate your changes in one central location.

step_size: 1
learning_rate: 0.01

You can now use the params: prefix to reference these parameters in the node definition:

def increase_volume(volume, step):
    return volume + step


# in pipeline definition
node(
    func=increase_volume,
    inputs=["input_volume", "params:step_size"],
    outputs="output_volume",
)

You can also group your parameters into nested structures and, using the same method above, load them by top-level key:

step_size: 1
model_params:
    learning_rate: 0.01
    test_data_ratio: 0.2
    number_of_train_iterations: 10000

def train_model(data, model):
    lr = model["learning_rate"]
    test_data_ratio = model["test_data_ratio"]
    iterations = model["number_of_train_iterations"]
    ...


# in pipeline definition
node(
    func=train_model,
    inputs=["input_data", "params:model_params"],
    outputs="output_data",
)

Alternatively, you can also pass parameters to the node inputs and get access to the entire collection of values inside the node function.

def increase_volume(volume, params):
    step = params["step_size"]
    return volume + step


# in pipeline definition
node(
    func=increase_volume, inputs=["input_volume", "parameters"], outputs="output_volume"
)

In both cases, under the hood parameters are added to the Data Catalog through the method add_feed_dict() in DataCatalog, where they live as MemoryDataSets. This method is also what the KedroContext class uses when instantiating the catalog.

Note

You can use add_feed_dict() to inject any other entries into your DataCatalog as per your use case.

Credentials¶

For security reasons, we strongly recommend you to not commit any credentials or other secrets to the Version Control System. Hence, by default any file inside the conf/ folder (and its subfolders) containing credentials in its name will be ignored via .gitignore and not committed to your git repository.

Credentials configuration can be loaded the same way as any other project configuration using any of the configuration loader classes: ConfigLoader, TemplatedConfigLoader, and OmegaConfigLoader. The following examples will all make use of the default ConfigLoader class.

from kedro.config import ConfigLoader
from kedro.framework.project import settings

conf_path = str(project_path / settings.CONF_SOURCE)
conf_loader = ConfigLoader(conf_source=conf_path, env="local")
credentials = conf_loader["credentials"]

This will load configuration files from conf/base and conf/local whose filenames start with credentials, or that are located inside a folder with a name that starts with credentials.

Note

Since local is set as the environment, the configuration path conf/local takes precedence in the example above. Hence, any overlapping top-level keys from conf/base will be overwritten by the ones from conf/local.

Calling conf_loader[key] in the example above throws a MissingConfigException error if no configuration files match the given key. If this is a valid workflow for your application, you can handle it as follows:

from kedro.config import ConfigLoader, MissingConfigException
from kedro.framework.project import settings

conf_path = str(project_path / settings.CONF_SOURCE)
conf_loader = ConfigLoader(conf_source=conf_path, env="local")

try:
    credentials = conf_loader["credentials"]
except MissingConfigException:
    credentials = {}

Note

The kedro.framework.context.KedroContext class uses the approach above to load project credentials.

Credentials configuration can then be used on its own or fed into the DataCatalog.

AWS credentials¶

When you work with AWS credentials on datasets, you are not required to store AWS credentials in the project configuration files. Instead, you can specify them using environment variables AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, and, optionally, AWS_SESSION_TOKEN. Please refer to the official documentation for more details.