cloudera.services.ml_project_model_deployment module – Manage a Cloudera Machine Learning (CML) project model deployment

Note

This module is part of the cloudera.services collection (version 1.0.0).

It is not included in ansible-core. To check whether it is installed, run ansible-galaxy collection list.

To install it, use: ansible-galaxy collection install git+https://github.com/cloudera-labs/cloudera.services.git.

To use it in a playbook, specify: cloudera.services.ml_project_model_deployment.

New in cloudera.services 1.0.0

Synopsis

• Start, stop, restart, or delete a Cloudera Machine Learning (CML) project model deployment.

• Deployments are immutable; a running deployment is never updated in place. Use state=restarted to stop the current deployment and start a new one on the build.

• The module supports check_mode.

Parameters

Parameter	Comments
api_key aliases: token string / required	The CML API key (bearer token) used to authenticate to the Workspace API. If not set, the value of the CML_API_KEY environment variable is used.
build_id string	The unique identifier of the build to deploy. If not set, the most recent successful build is used.
client_cert path	The path to a client certificate for authenticating to the API endpoint.
client_key path	The path to a client key for authenticating to the API endpoint.
cpu integer	The vCPU allocated to the deployment.
debug aliases: debug_endpoints boolean	A flag to enable debug logging of the module’s execution. Choices: false ← (default) true
env aliases: env_vars dictionary	Environment variables to set on the deployment.
force boolean	A flag to force a refresh of the API request, ignoring any cached results. Choices: false ← (default) true
force_basic_auth boolean	A flag to force basic authentication for API requests. Choices: false ← (default) true
gpu aliases: nvidia_gpus integer	The count of Nvidia GPUs allocated to the deployment.
http_agent aliases: user_agent string	The User-Agent string to send with API requests. Default: "cloudera-services-module"
id aliases: deployment_id string	The unique identifier of an existing deployment.
memory integer	The RAM allocated to the deployment, in GB.
model_id string	The unique identifier of the enclosing model. Mutually exclusive with model_name.
model_name string	The name of the enclosing model. Mutually exclusive with model_id.
page_size aliases: default_page_size integer	The number of items to return per page in a paginated API response. Default: 100
project_id string	The unique identifier of the enclosing project. Mutually exclusive with project_name.
project_name string	The name of the enclosing project. Mutually exclusive with project_id.
state string	The declarative state of the deployment. present is an alias for started. Choices: "started" ← (default) "present" "stopped" "restarted" "absent"
timeout aliases: timeout_seconds integer	The timeout in seconds for any API requests. Default: 60
url aliases: endpoint, endpoint_url, workspace_url string / required	The base URL of the CML Workspace API endpoint, including the port if necessary. If not set, the value of the CML_ENDPOINT environment variable is used.
url_password string	The password for authenticating to the API endpoint.
url_username string	The username for authenticating to the API endpoint.
use_gssapi boolean	A flag to enable or disable GSSAPI authentication for API requests. Choices: false ← (default) true
use_proxy boolean	A flag to enable or disable the use of a proxy for API requests. Choices: false true ← (default)
validate_certs boolean	A flag to enable or disable SSL certificate validation for API requests. Choices: false true ← (default)

Examples

- name: Deploy the latest build of a model
  cloudera.services.ml_project_model_deployment:
    url: "https://ml-workspace.example.com"
    api_key: "{{ cml_api_key }}"
    project_name: my-project
    model_name: fraud-detector
    cpu: 1
    memory: 2
    state: started

- name: Stop a deployment
  cloudera.services.ml_project_model_deployment:
    project_id: "{{ project_id }}"
    model_id: "{{ model_id }}"
    id: "{{ deployment_id }}"
    state: stopped

- name: Restart (stop then redeploy) a model
  cloudera.services.ml_project_model_deployment:
    project_id: "{{ project_id }}"
    model_id: "{{ model_id }}"
    state: restarted

- name: Delete a deployment
  cloudera.services.ml_project_model_deployment:
    project_id: "{{ project_id }}"
    model_id: "{{ model_id }}"
    id: "{{ deployment_id }}"
    state: absent

Return Values

Common return values are documented here, the following are the fields unique to this module:

Key	Description
deployment dictionary	The CML model deployment details. Returned: always
build_id string	The identifier of the enclosing build. Returned: when available
cpu float	The vCPU allocated to the deployment. Returned: when available
created_at string	The timestamp when the deployment was created. Returned: when available
deployer dictionary	Details of the user that deployed the model. Returned: when available
environment dictionary	The environment variables of the deployment. Returned: when available
id string	The unique identifier of the deployment. Returned: always
memory float	The RAM allocated to the deployment, in GB. Returned: when available
model_id string	The identifier of the enclosing model. Returned: when available
nvidia_gpus integer	The count of Nvidia GPUs allocated to the deployment. Returned: when available
project_id string	The identifier of the enclosing project. Returned: when available
replicas integer	The replica count of the deployment. Returned: when available
status string	The status of the deployment. Returned: when available
updated_at string	The timestamp when the deployment was last updated. Returned: when available
sdk_out string	Returns the captured REST API log. Returned: when supported
sdk_out_lines list / elements=string	Returns a list of each line of the captured REST API log. Returned: when supported

Authors

• Webster Mudge (@wmudge)

Collection Links

• Issue Tracker
• Repository (Sources)
• Submit a bug report
• Request a feature