Buster Python SDK

The official Python SDK for Buster.

Installation

pip install buster-sdk

Quick Start

Set your API key and add Buster to your Airflow DAG:

export BUSTER_API_KEY="your-secret-key"

from datetime import datetime
from airflow import DAG
from airflow.sdk import task
from buster import Client

client = Client()

with DAG(
    dag_id="my_pipeline",
    start_date=datetime(2024, 1, 1),
    schedule="@daily",
    catchup=False,
    default_args={
        "on_failure_callback": client.airflow.v3.task_on_failure,
    },
    on_failure_callback=client.airflow.v3.dag_on_failure,
) as dag:

    @task
    def extract():
        # Your extraction logic
        pass

    @task
    def transform():
        # Your transformation logic
        pass

    extract() >> transform()

Configuration

Client Parameters

Parameter	Type	Default	Description
buster_api_key	str	None	Your Buster API key. If not provided, uses BUSTER_API_KEY environment variable (recommended).
debug	str	None	Enable debug logging: "off", "error", "warn", "info", "debug".
env	str	"production"	Target environment: "production", "staging", "development". Rarely needed.
api_version	str	"v2"	API version. Currently only "v2" is supported. Rarely needed.
airflow_config	dict	None	Airflow-specific configuration (see Airflow Integration section).

Configuration Examples

Basic:

from buster import Client

client = Client()  # Uses BUSTER_API_KEY environment variable

With debug logging:

client = Client(debug="info")

With explicit API key:

client = Client(buster_api_key="your-secret-key")

Integrations

Airflow

Monitor and debug your Airflow DAGs by automatically reporting task and DAG failures to Buster.

Basic Setup

Use default_args to report all task failures in your DAG:

from datetime import datetime
from airflow import DAG
from airflow.sdk import task
from buster import Client

client = Client()

with DAG(
    dag_id="my_pipeline",
    start_date=datetime(2024, 1, 1),
    schedule="@daily",
    catchup=False,
    default_args={
        "on_failure_callback": client.airflow.v3.task_on_failure,
    },
    on_failure_callback=client.airflow.v3.dag_on_failure,
) as dag:

    @task
    def my_task():
        # Your task logic
        pass

    my_task()

Per-Task Callbacks

For more granular control, attach callbacks to specific tasks:

from airflow import DAG
from airflow.sdk import task
from buster import Client

client = Client()

with DAG(dag_id="my_dag", ...) as dag:
    @task(on_failure_callback=client.airflow.v3.task_on_failure)
    def critical_task():
        # Only this task reports failures
        pass

Plugin Integration

For centralized error reporting across all DAGs without modifying individual DAG files, use an Airflow plugin. This approach automatically captures failures from all DAGs in your Airflow instance.

Create a plugin file in your Airflow plugins directory (e.g., plugins/buster_plugin.py):

import sys
from airflow.plugins_manager import AirflowPlugin
from airflow.listeners import hookimpl
from airflow.utils.state import TaskInstanceState
from airflow.models.dagrun import DagRun
from buster import Client

client = Client()

try:
    from airflow.sdk.execution_time.task_runner import RuntimeTaskInstance
except ImportError:
    from airflow.models.taskinstance import TaskInstance as RuntimeTaskInstance

@hookimpl
def on_task_instance_failed(
    previous_state: TaskInstanceState,
    task_instance: RuntimeTaskInstance,
    error: str | BaseException | None,
):
    """Event listener for task instance failures."""
    client.airflow.v3.plugin_task_on_failure(
        previous_state=previous_state,
        task_instance=task_instance,
        error=error,
    )

@hookimpl
def on_dag_run_failed(dag_run: DagRun, msg: str):
    """Event listener for DAG run failures."""
    client.airflow.v3.plugin_dag_on_failure(
        dag_run=dag_run,
        msg=msg,
    )

class BusterPlugin(AirflowPlugin):
    name = "buster_plugin"
    listeners = [sys.modules[__name__]]

Benefits of plugin approach:

• Centralized error reporting for all DAGs
• No need to modify individual DAG files
• Automatically captures failures from new DAGs
• Easier to maintain and update

Airflow Configuration Options

Configure Airflow-specific behavior using the airflow_config parameter:

Option	Type	Default	Description
send_when_retries_exhausted	bool	True	If True, only reports errors when the task has exhausted all retries. This should rarely be set to false