Configuration#

Note

By default, if no configuration is provided, Lithops will use the Localhost backend to run the functions.

To work with Lithops on the Cloud, you must configure at least one compute backend and one storage backend. Lithops can work both with the leading cloud providers, and using on-premise or Kubernetes platforms. You have multiple options to choose compute and storage backends based on your needs.

Lithops configuration can be provided either using a configuration file, or in runtime via a python dictionary.

Compute and Storage backends#

Choose your compute and storage engines from the table below:

Compute backends

Storage Backends

Configuration File#

To configure Lithops through a configuration file you have multiple options:

  1. Create a new file called config in the ~/.lithops folder.

  2. Create a new file called .lithops_config in the root directory of your project from where you will execute your Lithops scripts.

3. Create a new file called config in the /etc/lithops/ folder (i.e: /etc/lithops/config). Useful for sharing the config file on multi-user machines.

4. Create the config file in any other location and configure the LITHOPS_CONFIG_FILE system environment variable indicating the absolute or relative location of the configuration file:

LITHOPS_CONFIG_FILE=<CONFIG_FILE_LOCATION>

Configuration keys in runtime#

An alternative mode of configuration is to use a Python dictionary. This option allows to pass all the configuration details as part of the Lithops invocation in runtime. You can see an entire list of configuration keys at the Configuration Reference section.

Here is an example of providing configuration keys for IBM Cloud Functions and IBM Cloud Object Storage:

import lithops


config = {'lithops': {'backend': 'ibm_cf', 'storage': 'ibm_cos'},
          'ibm': {'region': 'REGION',
                  'iam_api_key': 'IAM_API_KEY',
                  'resource_group_id': 'RESOURCE_GROUP_ID'}
          'ibm_cos': {'storage_bucket': 'STORAGE_BUCKET'}}

def hello_world(name):
    return 'Hello {}!'.format(name)

if __name__ == '__main__':
    fexec = lithops.FunctionExecutor(config=config)
    fexec.call_async(hello_world, 'World')
    print(fexec.get_result())

Configuration Reference#

Lithops Config Keys#

Group

Key

Default

Mandatory

Additional info

lithops

backend

ibm_cf

no

Compute backend implementation. IBM Cloud Functions is the default. If not set, Lithops will check the mode and use the backend set under the serverless or standalone sections described below.

lithops

storage

ibm_cos

no

Storage backend implementation. IBM Cloud Object Storage is the default.

lithops

data_cleaner

True

no

If set to True, then the cleaner will automatically delete all the temporary data that was written into storage_bucket/lithops.jobs.

lithops

monitoring

storage

no

Monitoring system implementation. One of: storage or rabbitmq.

lithops

monitoring_interval

2

no

Monitoring check interval in seconds in case of storage monitoring.

lithops

data_limit

4

no

Max (iter)data size (in MB). Set to False for unlimited size.

lithops

execution_timeout

1800

no

Functions will be automatically killed if they exceed this execution time (in seconds). Alternatively, it can be set in the call_async(), map() or map_reduce() calls using the timeout parameter.

lithops

include_modules

[]

no

Explicitly pickle these dependencies. All required dependencies are pickled if default empty list. No one dependency is pickled if it is explicitly set to None.

lithops

exclude_modules

[]

no

Explicitly keep these modules from pickled dependencies. It is not taken into account if you set include_modules.

lithops

log_level

INFO

no

Logging level. One of: WARNING, INFO, DEBUG, ERROR, CRITICAL, Set to None to disable logging.

lithops

log_format

%(asctime)s [%(levelname)s] %(name)s -- %(message)

no

Logging format string.

lithops

log_stream

ext://sys.stderr

no

Logging stream. eg.: ext://sys.stderr, ext://sys.stdout

lithops

log_filename

no

Path to a file. log_filename has preference over log_stream.