airflow-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <>
Subject [jira] [Commented] (AIRFLOW-1488) Add a sensor operator to wait on DagRuns
Date Fri, 07 Dec 2018 02:11:00 GMT


ASF GitHub Bot commented on AIRFLOW-1488:

ybendana opened a new pull request #4291: [AIRFLOW-1488] Add the DagRunSensor operator
   ExternalTaskSensor allows one to wait on any valid combination of
   (dag_id, task_id). It is desirable though to be able to wait on entire
   DagRuns, as opposed to specific task instances in those DAGs.
   This pull request adds a new sensor in contrib called DagRunSensor.
   This version is a different approach from previous pull requests
   addressing the same issue.  In this case, the DagRunSensor takes a
   trigger_task_id, the id of a task that triggers DagRuns.  The trigger
   task returns a list of run_ids of the DagRuns it triggered and the
   DagRunSensor polls their status. For this purpose the
   TriggerDagRunOperator was modified so that it stores the run_id of the
   triggered DagRun.
   Make sure you have checked _all_ steps below.
   ### Jira
   - [ ] My PR addresses the following [Airflow Jira](
issues and references them in the PR title. For example, "\[AIRFLOW-XXX\] My Airflow PR"
     - In case you are fixing a typo in the documentation you can prepend your commit with
\[AIRFLOW-XXX\], code changes always need a Jira issue.
   ### Description
   - [ ] Here are some details about my PR, including screenshots of any UI changes:
   See the above description.  I also copied quite a bit from the previous pull requests on
this issue #2500 and #3234.  Unlike those attempts, the focus here is on triggered DagRuns.
 I realize it may not be generally applicable to everyone's workflows but it has been very
useful for us. At one point I considered using subdags but since they have had their issues
I think this a good alternative.
   ### Tests
   - [ ] My PR adds the following unit tests __OR__ does not need testing for this extremely
good reason:
   I included and  Currently, I'm running
into an issue executing the test in the Docker environment because the sensor task has to
wait on the triggered DagRun.  The DagRun is triggered but is never scheduled.  I think this
is because the unit test uses the SequentialExecutor and SQLite.  If I run the scheduler manually,
the triggered DagRun executes. I'm not sure what to do about this. For now, I've commented
the successful DagRun test and have only the failed DagRun test active.
   ### Commits
   - [ ] My commits all reference Jira issues in their subject lines, and I have squashed
multiple commits if they address the same issue. In addition, my commits follow the guidelines
from "[How to write a good git commit message](":
     1. Subject is separated from body by a blank line
     1. Subject is limited to 50 characters (not including Jira issue reference)
     1. Subject does not end with a period
     1. Subject uses the imperative mood ("add", not "adding")
     1. Body wraps at 72 characters
     1. Body explains "what" and "why", not "how"
   ### Documentation
   - [ ] In case of new functionality, my PR adds documentation that describes how to use
     - When adding new operators/hooks/sensors, the autoclass documentation generation needs
to be added.
     - All the public functions and the classes in the PR contain docstrings that explain
what it does
   ### Code Quality
   - [ ] Passes `flake8`

This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:

> Add a sensor operator to wait on DagRuns
> ----------------------------------------
>                 Key: AIRFLOW-1488
>                 URL:
>             Project: Apache Airflow
>          Issue Type: New Feature
>          Components: contrib, operators
>            Reporter: Yati
>            Assignee: Yati
>            Priority: Major
> The [ExternalTaskSensor|]
operator already allows for encoding dependencies on tasks in external DAGs. However, when
you have teams, each owning multiple small-to-medium sized DAGs, it is desirable to be able
to wait on an external DagRun as a whole. This allows the owners of an upstream DAG to refactor
their code freely by splitting/squashing task responsibilities, without worrying about dependent
DAGs breaking.
> I'll now enumerate the easiest ways of achieving this that come to mind:
> * Make all DAGs always have a join DummyOperator in the end, with a task id that follows
some convention, e.g., "{{ dag_id }}.__end__".
> * Make ExternalTaskSensor poke for a DagRun instead of TaskInstances when the external_task_id
argument is None.
> * Implement a separate DagRunSensor operator.
> After considerations, we decided to implement a separate operator, which we've been using
in the team for our workflows, and I think it would make a good addition to contrib.

This message was sent by Atlassian JIRA

View raw message