Airflow XComs: Difference between revisions

From NovaOrdis Knowledge Base
Jump to navigation Jump to search
 
(3 intermediate revisions by the same user not shown)
Line 31: Line 31:
</syntaxhighlight>
</syntaxhighlight>


<code>xcom_pull</code> pulls XComs that optionally meet certain criteria. It has the following parameters:
<code>xcom_pull</code> pulls XComs that optionally meet certain criteria. If there is no XCom that match the criteria (no such key for an existing task, no such task, etc.), <code>xcom_pull()</code> returns <code>None</code>.
It has the following parameters:
====<tt>key</tt>====
====<tt>key</tt>====
The XCom's key. If provided, only XComs with matching keys will be returned. The default key is 'return_value', also available as constant <code>XCOM_RETURN_KEY</code>. This key is automatically associated with the XComs returned by the task, as opposed to being pushed with <code>xcom_push()</code>. To remove the key filter, pass <code>None</code>.
The XCom's key. If provided, only XComs with matching keys will be returned. The default key is 'return_value', also available as constant <code>XCOM_RETURN_KEY</code>. This key is automatically associated with the XComs returned by the task, as opposed to being pushed with <code>xcom_push()</code>. To remove the key filter, pass <code>None</code>.
====<tt>task_ids</tt>====
====<tt>task_ids</tt>====
Only XComs from tasks with matching ids will be pulled. Pass <code>None</code> to remove the filter.
Only XComs from tasks with matching ids will be pulled. Pass <code>None</code> to remove the filter. To specify multiple task IDs, provide a non-string iterable.
 
====<tt>dag_id</tt>====
====<tt>dag_id</tt>====
If provided, only pulls XComs from the specified DAG. If <code>None</code> is provided, which is the default, the DAG of the calling task is used.
If provided, only pulls XComs from the specified DAG. If <code>None</code> is provided, which is the default, the DAG of the calling task is used.
Line 41: Line 43:
If provided, only pull XComs with matching indexes. If <code>None</code> (default), this is inferred from the task(s) being pulled.  When pulling one single task (<code>task_ids</code> is <code>None</code> or a str) without  specifying <code>map_indexes</code>, the return value is inferred from whether the specified task is mapped. If not, value from the one single task instance is returned. If the task to pull is mapped, an iterator (not a list) yielding XComs from mapped task instances is returned. In either case, <code>default</code> (<code>None</code> if not specified) is returned if no matching  XComs are found. When pulling multiple tasks (i.e. either <code>task_ids</code> or <code>map_index</code> is a non-str iterable), a list of matching XComs is returned. Elements in the list is ordered by item ordering in <code>task_id</code> and <code>map_index</code>.
If provided, only pull XComs with matching indexes. If <code>None</code> (default), this is inferred from the task(s) being pulled.  When pulling one single task (<code>task_ids</code> is <code>None</code> or a str) without  specifying <code>map_indexes</code>, the return value is inferred from whether the specified task is mapped. If not, value from the one single task instance is returned. If the task to pull is mapped, an iterator (not a list) yielding XComs from mapped task instances is returned. In either case, <code>default</code> (<code>None</code> if not specified) is returned if no matching  XComs are found. When pulling multiple tasks (i.e. either <code>task_ids</code> or <code>map_index</code> is a non-str iterable), a list of matching XComs is returned. Elements in the list is ordered by item ordering in <code>task_id</code> and <code>map_index</code>.
====<tt>include_prior_dates</tt>====
====<tt>include_prior_dates</tt>====
If <code>False</code>, only XComs from the current <code>execution_date</code> are returned. If <code>True</code, XComs from previous dates are returned as well.
If <code>False</code>, only XComs from the current <code>[[Airflow_Concepts#execution_date|execution_date]]</code> are returned. If <code>True</code>, XComs from previous dates are returned as well.
 
 


If there is no such XCom key for an existing task, or no such task, <code>xcom_pull()</code> returns <code>None</code>.
===<tt>task_ids</tt>===
===<tt>task_ids</tt>===
===To Clarify===
===To Clarify===

Latest revision as of 03:10, 18 July 2022

External

Internal

Overview

XComs is one of the methods tasks use to exchange data. Tasks communicate using inputs and outputs, and the XComs ("cross-communications") mechanism is an implementation of this pattern. By default, tasks are entirely isolated and may be running on entirely different machines so when they exchange data, the data must be serializable.

An XCom is identified by a key, which is the XCom's name, as well as the task_id and dag_id it came from. The XCom can have any serializable value, however it must be relatively small. If there is need to exchange large amounts of data, this is usually done uploading and downloading large files from a storage service.

XComs are explicitly “pushed” and “pulled” to/from their storage using the xcom_push() and xcom_pull() methods on task instances. For more details see Ingesting Input into a Task with xcom_pull() and Exposing Task Output with xcom_push() or via the Return Value below. The XComs are stored in the xcom table and they need to be explicitly deleted after use, otherwise they'll leak in the table.

Variables are an alternative mechanism for tasks to share data. However, variables are global and should be used for overall configuration that covers the entire installation. To pass data to and from tasks, XComs are preferable.

Also see:

DAG run execution context

Programming Model

Also see:

Airflow Programming Model

Other examples:

https://github.com/apache/airflow/blob/main/airflow/example_dags/example_xcom.py

Ingesting Input into a Task with xcom_pull()

Data externally generated by a preceding task can be ingested by a task with xcom_pull:

@task
def task_b(ti=None):
    v = ti.xcom_pull(key="return_value", task_ids='task_a')
    print(v)

xcom_pull pulls XComs that optionally meet certain criteria. If there is no XCom that match the criteria (no such key for an existing task, no such task, etc.), xcom_pull() returns None.

It has the following parameters:

key

The XCom's key. If provided, only XComs with matching keys will be returned. The default key is 'return_value', also available as constant XCOM_RETURN_KEY. This key is automatically associated with the XComs returned by the task, as opposed to being pushed with xcom_push(). To remove the key filter, pass None.

task_ids

Only XComs from tasks with matching ids will be pulled. Pass None to remove the filter. To specify multiple task IDs, provide a non-string iterable.

dag_id

If provided, only pulls XComs from the specified DAG. If None is provided, which is the default, the DAG of the calling task is used.

map_indexes

If provided, only pull XComs with matching indexes. If None (default), this is inferred from the task(s) being pulled. When pulling one single task (task_ids is None or a str) without specifying map_indexes, the return value is inferred from whether the specified task is mapped. If not, value from the one single task instance is returned. If the task to pull is mapped, an iterator (not a list) yielding XComs from mapped task instances is returned. In either case, default (None if not specified) is returned if no matching XComs are found. When pulling multiple tasks (i.e. either task_ids or map_index is a non-str iterable), a list of matching XComs is returned. Elements in the list is ordered by item ordering in task_id and map_index.

include_prior_dates

If False, only XComs from the current execution_date are returned. If True, XComs from previous dates are returned as well.

task_ids

To Clarify

  • There is also a static method of xcom_pull(). When to use that?

Exposing Task Output via the Return Value or with xcom_push()

Simply returning a value out of a @task function automatically exposes it as a "return_value" XCom.

@task
def task_a():
    return "something"

is equivalent with:

@task
def task_a(ti=None):
    ti.xcom_push('return_value', 'something')

Deleting XComs

TO PROCESS: https://stackoverflow.com/questions/46707132/how-to-delete-xcom-objects-once-the-dag-finishes-its-run-in-airflow

XComArg

When you call a TaskFlow function in the DAG file, rather than executing it, you will get an object representing the XCom for the result (an XComArg, that you can use as inputs to downstream tasks and operators.

Operations

The values of the created XComs, tabulated by timestamp, DAG ID, Task ID, key and value, are available via Admin → XComs.

Backends

https://airflow.apache.org/docs/apache-airflow/stable/concepts/xcoms.html#custom-xcom-backends