Table Of Contents

  1. Docs
  2. ABEJA Run SDK documentation
  3. API Client

API Client

API Client represents a low level interface for Run API. All API operation can be performed using this.

class abeja.runs.APIClient(credential: dict | None = None, timeout: int | None = None, max_retry_count: int | None = None)

A low-level client for Run API

from abeja.run import APIClient

api_client = APIClient()
create_run(organization_id: str, deployment_id: str, version_id: str, input_data: Dict[str, str] | None = None, output_template: Dict[str, str] | None = None, distribution: int | None = None, retry_count: int | None = None, environment: Dict[str, str] | None = None, model_id: str | None = None) → dict

submit a run

API reference: POST /organizations/<organization_id>/deployments/<deployment_id>/runs

Request Syntax:
organization_id = "1234567890123"
deployment_id = "1111111111111"
version_id = "ver-1111111111111111"
model_id = "3333333333333"
input_data = {
    "$datalake:1": "2222222222222/20180101T112233-44444444-5555-6666-7777-888888888888"
}
distribution = 0
retry_count = 0
environment = {
    "EXAMPLE_ENV": "example"
}
response = api_client.create_run(
    organization_id, deployment_id, version_id, input_data,
    model_id=model_id, distribution=distribution, retry_count=retry_count, environment=environment)
Params:

  • organization_id (str): The ID of the organization.

  • deployment_id (str): The ID of the deployment.

  • version_id (str): The ID of the version of a model.

  • model_id (str): training model identifier

  • input_data (dict): [optional] specifies input data in json format for INPUT. The format is like following {"<operator>:<protocol_version>": "<operator_payload>"}. The value specified in INPUT is passed to the first argument of the handler of the user-defined code.

  • output_data (dict): [optional] specifies output data in json format for OUTPUT. The format is like following {"<operator>:<protocol_version>": "<operator_payload>"}.

  • distribution (int): [optional] specifies the id whose remainder determine distributed queue, as default, round-robin is used.

  • retry_count (int): [optional] specifies the number of attempts from 0 to 9, defaults to 5.

  • environment (int): [optional] specifies the environment variables which can be referred from user code.

Return type:

dict

Returns:

Response Syntax:

{
    "deployment_id": "1111111111111",
    "models": {
        "alias": "3333333333333"
    },
    "model_version": "0.0.3",
    "model_version_id": "ver-1111111111111111",
    "input_data": {
        "$datalake:1": "2222222222222/20180101T112233-44444444-5555-6666-7777-888888888888"
    },
    "output_template": null,
    "retry_count": 7,
    "run_id": "run-3333333333333333",
    "status": "SUBMITTED",
    "user_env_vars": {
        "DEBUG": "debug"
    },
    "distribution": 0,
    "created_at": "2018-01-01T00:00:00.000000Z",
    "modified_at": "2018-01-01T00:00:00.000000Z"
}
Raises:

  • Unauthorized: Authentication failed

  • NotFound:

  • Forbidden:

  • InternalServerError:

get_run(organization_id: str, deployment_id: str, run_id: str) → dict

get a run

API reference: GET /organizations/<organization_id>/deployments/<deployment_id>/runs/<run_id>

Request Syntax:
organization_id = "1234567890123"
deployment_id = "1111111111111"
run_id = "run-3333333333333333"
response = api_client.get_run(organization_id, deployment_id, run_id)
Params:

  • organization_id (str): The ID of the organization.

  • deployment_id (str): The ID of the deployment.

  • run_id (str): The ID of the run.

Return type:

dict

Returns:

Response Syntax:

{
    "deployment_id": "1111111111111",
    "models": {
        "alias": "3333333333333"
    },
    "model_version": "0.0.3",
    "model_version_id": "ver-1111111111111111",
    "input_data": {
        "$datalake:1": "2222222222222/20180101T112233-44444444-5555-6666-7777-888888888888"
    },
    "output_template": null,
    "retry_count": 7,
    "run_id": "run-3333333333333333",
    "status": "SUBMITTED",
    "user_env_vars": {
        "DEBUG": "debug"
    },
    "created_at": "2018-01-01T00:00:00.000000Z",
    "modified_at": "2018-01-01T00:00:00.000000Z"
}

Response Structure:

  • deployment_id (str): The ID of the deployment.

  • models (dict): combination of alias and model_id

  • model_version (str): The version of the model.

  • model_version_id (str): The ID of the version.

  • input_data (str): input data in json format for INPUT.

  • output_template (str): output data in json format for OUTPUT.

  • retry_count (int): the number of attempts.

  • run_id (str): The ID of the run.

  • status (str): The status of the run, which is each one of SUBMITTED, PENDING, RUNNABLE, STARTING, RUNNING, SUCCEEDED, FAILED.

  • user_env_vars (str): environment variables which are available in user code.

  • created_at (str): created date of the run.

  • modified_at (str): modified date of the run.

Raises:

  • Unauthorized: Authentication failed

  • NotFound:

  • Forbidden:

  • InternalServerError:

get_run_logs(organization_id: str, deployment_id: str, run_id: str, next_token: str | None = None, start_time: datetime | None = None, end_time: datetime | None = None) → dict

get logs of the run

API reference: GET /organizations/<organization_id>/deployments/<deployment_id>/runs/<run_id>/logs

Request Syntax:
organization_id = "1234567890123"
deployment_id = "1111111111111"
run_id = "run-3333333333333333"
response = api_client.get_run_logs(organization_id, deployment_id, run_id)
Params:

  • organization_id (str): The ID of the organization.

  • deployment_id (str): The ID of the deployment.

  • run_id (str): The ID of the run.

  • next_token (str): [optional] specifies the following logs

  • start_time (datetime): [optional] specifies the start utc datetime included in data series.

  • end_time (datetime): [optional] specifies the end utc datetime included in data series.

Return type:

dict

Returns:

Response Syntax:

{
    "events": [
        {
            "message": "...",
            "timestamp": "2018-01-01T00:00:00.000000Z"
        }
    ],
    "next_token": "..."
}

Response Structure:

  • events (list): list of messages emitted from the run. the messages are sorted in ascending order.

  • next_token (str): token for the next page of logs. If no more logs, this key does not exist.

Raises:

  • Unauthorized: Authentication failed

  • NotFound:

  • Forbidden:

  • InternalServerError:

get_run_recent_logs(organization_id: str, deployment_id: str, run_id: str, next_forward_token: str | None = None, next_backward_token: str | None = None, start_time: datetime | None = None, end_time: datetime | None = None) → dict

get recent logs of the run

API reference: GET /organizations/<organization_id>/deployments/<deployment_id>/runs/<run_id>/recentlogs

Request Syntax:
organization_id = "1234567890123"
deployment_id = "1111111111111"
run_id = "run-3333333333333333"
response = api_client.get_run_recent_logs(organization_id, deployment_id, run_id)
Params:

  • organization_id (str): The ID of the organization.

  • deployment_id (str): The ID of the deployment.

  • run_id (str): The ID of the run.

  • next_forward_token (str): [optional] token for the next page of logs

  • next_backward_token (str): [optional] token for the next previous of logs

  • start_time (datetime): [optional] specifies the start utc datetime included in data series.

  • end_time (datetime): [optional] specifies the end utc datetime included in data series.

Return type:

dict

Returns:

Response Syntax:

{
    "events": [
        {
            "message": "...",
            "timestamp": "2018-01-01T00:00:00.000000Z"
        }
    ],
    "next_backward_token": "...",
    "next_forward_token": "..."
}

Response Structure:

  • events (list): list of messages emitted from the run. the messages are sorted in descending order.

  • next_backward_token (str): token for the previous page of logs. If no more logs, this key exists and returns empty list.

  • next_forward_token (str): token for the next page of logs. If no more logs, this key exists and returns empty list.

Raises:

  • Unauthorized: Authentication failed

  • NotFound:

  • Forbidden:

  • InternalServerError:

get_runs(organization_id: str, deployment_id: str, next_page_token: str | None = None, scan_forward: bool | None = None) → dict

get runs

API reference: GET /organizations/<organization_id>/deployments/<deployment_id>/runs/

Request Syntax:
organization_id = "1234567890123"
deployment_id = "1111111111111"
response = api_client.get_runs(organization_id, deployment_id)
Params:

  • organization_id (str): The ID of the organization.

  • deployment_id (str): The ID of the deployment.

  • next_page_token (str): [optional] Token for offset of entries.

  • scan_forward (str): [optional] By default False, the sort order is descending. To reverse the order, set this to True

Return type:

dict

Returns:

Response Syntax:

{
    "entries": [
        {
            "deployment_id": "1111111111111",
            "models": {
                "alias": "3333333333333"
            },
            "model_version": "0.0.3",
            "model_version_id": "ver-1111111111111111",
            "input_data": {
                "$datalake:1": "2222222222222/20180101T112233-44444444-5555-6666-7777-888888888888"
            },
            "output_template": null,
            "retry_count": 7,
            "run_id": "run-3333333333333333",
            "status": "SUBMITTED",
            "user_env_vars": {
                "DEBUG": "debug"
            },
            "created_at": "2018-01-01T00:00:00.000000Z",
            "modified_at": "2018-01-01T00:00:00.000000Z"
        }
    ],
    "next_page_token": "..."
}
Raises:

  • Unauthorized: Authentication failed

  • NotFound:

  • Forbidden:

  • InternalServerError:

ABEJA Run SDK documentation
ABEJA Trigger SDK documentation
© Copyright 2018, ABEJA, Inc.. Created using Sphinx.