ferruzzi commented on code in PR #53727: URL: https://github.com/apache/airflow/pull/53727#discussion_r2274477432
########## airflow-core/docs/howto/deadline-alerts.rst: ########## @@ -0,0 +1,312 @@ + .. Licensed to the Apache Software Foundation (ASF) under one + or more contributor license agreements. See the NOTICE file + distributed with this work for additional information + regarding copyright ownership. The ASF licenses this file + to you under the Apache License, Version 2.0 (the + "License"); you may not use this file except in compliance + with the License. You may obtain a copy of the License at + + .. http://www.apache.org/licenses/LICENSE-2.0 + + .. Unless required by applicable law or agreed to in writing, + software distributed under the License is distributed on an + "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + KIND, either express or implied. See the License for the + specific language governing permissions and limitations + under the License. + + +Deadline Alerts +=============== + +The :class:`~airflow.sdk.definitions.deadline.DeadlineAlert` feature is the next evolution of +the old SLA. Deadline Alerts allow you to set time thresholds for your DAG runs and automatically +respond when those thresholds are exceeded. You can set up Deadline Alerts by choosing a built-in +reference point, setting an interval, and defining a response using either Airflow's Notifiers or +a custom callback function. + +Creating a Deadline Alert +------------------------- + +Creating a Deadline Alert requires three mandatory and one optional parameter: + +* Reference: When to start counting from +* Interval: How far before or after the reference point to trigger the alert +* Callback: What to do when the deadline is exceeded +* Callback Kwargs: Optional values to pass to the Callback when it is run + +Here is how Deadlines are calculated: + +:: + + [Reference] ------ [Interval] ------> [Deadline] + ^ ^ + | | + Start time Trigger point + +Below is an example DAG implementation. If the DAG has not finished 15 minutes after it was queued, send a Slack message: + +.. code-block:: python + + from datetime import datetime, timedelta + from airflow import DAG + from airflow.sdk.definitions.deadline import AsyncCallback, DeadlineAlert, DeadlineReference + from airflow.providers.slack.notifications.slack_webhook import SlackWebhookNotifier + from airflow.providers.standard.operators.empty import EmptyOperator + + with DAG( + dag_id="deadline_alert_example", + deadline=DeadlineAlert( + reference=DeadlineReference.DAGRUN_QUEUED_AT, + interval=timedelta(minutes=15), + callback=AsyncCallback( + SlackWebhookNotifier, + kwargs={ + "slack_conn_id": "slack_default", + "channel": "#alerts", + "text": "DAG 'slack_deadline_alert' still running after 30 minutes.", + "username": "Airflow Alerts", + }, + ), + ), + ): + EmptyOperator(task_id="example_task") + +The timeline for this example would look like this: + +:: + + |------|-----------|---------|-----------|--------| + Scheduled Queued Started Deadline + 00:00 00:03 00:05 00:18 + +Using Built-in References +------------------------- + +Airflow provides several built-in reference points that you can use with DeadlineAlert: + +``DeadlineReference.DAGRUN_QUEUED_AT`` + Measures time from when the DagRun was queued. Useful for monitoring resource constraints. + +``DeadlineReference.DAGRUN_LOGICAL_DATE`` + References when the DAG run was scheduled to start. For example, setting an interval of + ``timedelta(minutes=15)`` would trigger the alert if the DAG hasn't completed 15 minutes + after it was scheduled to start, regardless of when (or if) it actually began executing. + Useful for ensuring scheduled DAGs complete before their next scheduled run. + +``DeadlineReference.FIXED_DATETIME`` + Specifies a fixed point in time. Useful when DAGs must complete by a specific time. + +Here's an example using a fixed datetime: + +.. code-block:: python + + tomorrow_at_ten = datetime.combine(datetime.now().date() + timedelta(days=1), time(10, 0)) + + with DAG( + dag_id="fixed_deadline_alert", + deadline=DeadlineAlert( + reference=DeadlineReference.FIXED_DATETIME(tomorrow_at_ten), + interval=timedelta(minutes=-30), # Alert 30 minutes before the reference. + callback=AsyncCallback( + SlackWebhookNotifier, + kwargs={ + "slack_conn_id": "slack_default", + "channel": "#alerts", + "text": "DAG 'slack_deadline_alert' still running after 30 minutes.", + "username": "Airflow Alerts", + }, + ), + ), + ): + EmptyOperator(task_id="example_task") + +The timeline for this example would look like this: + +:: + + |------|----------|---------|------------|--------| + Queued Start Deadline Reference + 09:15 09:17 09:30 10:00 + +.. note:: + Note that since the interval is a negative value, the deadline is before the reference in this case. + +Using Callbacks +--------------- + +When a deadline is exceeded, the callback is executed. You can use an existing :doc:`Notifier </howto/notifications>` +or create a custom callback function. A callback must be either an :class:`~airflow.sdk.definitions.deadline.AsyncCallback` +or a :class:`~airflow.sdk.definitions.deadline.SyncCallback`. + +Using Built-in Notifiers +^^^^^^^^^^^^^^^^^^^^^^^^ + +Here's an example using the Slack Notifier if the DagRun has not finished within 30 minutes of it being queued: + +.. code-block:: python + + with DAG( + dag_id="slack_deadline_alert", + deadline=DeadlineAlert( + reference=DeadlineReference.DAGRUN_QUEUED_AT, + interval=timedelta(minutes=30), + callback=AsyncCallback( + SlackNotifier, + kwargs={ + "slack_conn_id": "slack_default", + "channel": "#alerts", + "text": "DAG 'slack_deadline_alert' still running after 30 minutes.", + "username": "Airflow Alerts", + }, + ), + ), + ): + EmptyOperator(task_id="example_task") + +Creating Custom Callbacks +^^^^^^^^^^^^^^^^^^^^^^^^^ + +You can create custom callbacks for more complex handling. The ``callback_kwargs`` specified in +the ``DeadlineAlert`` are passed to the callback function, if any are provided. **Synchronous callbacks** +(standard python methods) can be defined in the dag bundle and are run in the Executor. **Asynchronous +callbacks** must be defined somewhere in the Triggerer's system path. + +.. note:: + Regarding Async Custom Deadline callbacks: + + * Async callbacks are executed by the Triggerer, so users must ensure they are importable by the Triggerer. + * One easy way to do this is to place the callback as a top-level method in a new file in the plugins folder. + * The Triggerer will need to be restarted when a callback is added or changed in order to reload the file. + +A **custom synchronous callback** might look like this: + +.. code-block:: python + + from datetime import timedelta + + from airflow import DAG + from airflow.providers.standard.operators.empty import EmptyOperator + from airflow.sdk.definitions.deadline import DeadlineAlert, DeadlineReference, SyncCallback + + + def custom_synchronous_callback(**kwargs): + """Handle deadline violation with custom logic.""" + print(f"Deadline exceeded for DAG {kwargs.get("dag_id")}!") + print(f"Alert type: {kwargs.get("alert_type")}") + # Additional custom handling here + + + with DAG( + dag_id="custom_deadline_alert", + deadline=DeadlineAlert( + reference=DeadlineReference.DAGRUN_QUEUED_AT, + interval=timedelta(minutes=15), + callback=SyncCallback( + custom_synchronous_callback, + kwargs={"alert_type": "time_exceeded", "dag_id": "custom_deadline_alert"}, + ), + ), + ): + EmptyOperator(task_id="example_task") + +A **custom asynchronous callback** is only slightly more work. Note in the following example that +the custom callback code is placed in a separate file, and must be imported in the DAG file. + +.. code-block:: python + + # Place this method in `/files/plugins/deadline_callbacks.py` + async def custom_async_callback(**kwargs): + """Handle deadline violation with custom logic.""" + print(f"Deadline exceeded for DAG {kwargs.get("dag_id")}!") + print(f"Alert type: {kwargs.get("alert_type")}") + # Additional custom handling here + + + # Place this in a dag file Review Comment: Sure, I'll make the change in the next revision, thanks. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: commits-unsubscr...@airflow.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
