amoghrajesh opened a new pull request, #69651:
URL: https://github.com/apache/airflow/pull/69651
<!-- SPDX-License-Identifier: Apache-2.0
https://www.apache.org/licenses/LICENSE-2.0 -->
<!--
Thank you for contributing!
Please provide above a brief description of the changes made in this pull
request.
Write a good git commit message following this guide:
http://chris.beams.io/posts/git-commit/
Please make sure that your code changes are covered with tests.
And in case of new features or big changes remember to adjust the
documentation.
For user-facing UI changes, please attach before/after screenshots (or a
short
screen recording) so reviewers can assess the visual impact.
Feel free to ping (in general) for the review if you do not see reaction for
a few days
(72 Hours is the minimum reaction time you can expect from volunteers) - we
sometimes miss notifications.
In case of an existing issue, reference it using one of the following:
* closes: #ISSUE
* related: #ISSUE
-->
---
##### Was generative AI tooling used to co-author this PR?
<!--
If generative AI tooling has been used in the process of authoring this PR,
please
change below checkbox to `[X]` followed by the name of the tool, uncomment
the "Generated-by".
-->
- [ ] Yes (please specify the tool below)
<!--
Generated-by: [Tool Name] following [the
guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/05_pull_requests.rst#gen-ai-assisted-contributions)
-->
---
closes: #67236
### What
Through this PR, I am trying to surface which operators can survive a worker
crash/retry (aka durable execution) without redoing expensive work on Airflow
Registry. `ResumableJobMixin` already provides this capability (Spark,
Databricks, so far, Snowflake and other operators in progress), but nothing in
the registry's extraction pipeline or UI could distinguish an operator that
genuinely implements it from one that merely inherits it, and the Registry
itself (`registry/` + `dev/registry/`, in this repo) had no capability concept
at all, only type classification (operator/hook/sensor/etc).
### Current behaviour
Today, browsing a provider page in the Registry shows every
operator/hook/sensor with no indication of crash-recovery support. Inheriting
`ResumableJobMixin` is not proof an operator is crash safe on its own. A
subclass can leave the mixin's methods unimplemented, implement them but never
call `execute_resumable()` from `execute()` (dead capability), or, as in
`SparkSubmitOperator`, only reach it under specific hook configuration.
### Proposed change
The change is a 3 step process
1. **Detection** (`dev/registry/extract_parameters.py`):
`is_durable_capable()` checks three things: the class inherits
`ResumableJobMixin`, none of its abstract methods are left unimplemented (via
`inspect.isabstract()`, now that the mixin's 6 crash-recovery methods are real
`@abstractmethod`s as of #69607, already merged), and `execute_resumable` is
reachable from `execute()`'s source. `load_resumable_job_mixin()` imports the
mixin once and is threaded down through `discover_classes_from_provider()` and
`make_entry()`, mirroring the existing `base_classes` injection pattern.
2. **Schema** (`registry_contract_models.py`): `supports_durable_execution:
bool = False` added to `ModuleContract`.
3. **Frontend** (`registry/src/provider-version.njk`, `main.css`): a small
teal "Durable" badge on module cards when `supports_durable_execution` is true,
with a tooltip pointing at the caveats (`durable=True` default,
config-dependent, no per-run guarantee).
### User implications / backcompat
New field defaults to `False`; no schema break. Note `ModuleContract`
validation uses `exclude_unset` semantics: the field is simply absent (not
present-and-false) on payloads that never set it, e.g. legacy per-version data
from older extraction runs.
### Testing
#### Non UI surface testing
1. Tested if the `extract-data` works fine for databricks as well as spark:
Spark:
```shell
breeze registry extract-data --provider apache-spark
```
Result:
```json
{
"id": "apache-spark-spark_submit-SparkSubmitOperator",
"name": "SparkSubmitOperator",
"type": "operator",
"import_path":
"airflow.providers.apache.spark.operators.spark_submit.SparkSubmitOperator",
"module_path": "airflow.providers.apache.spark.operators.spark_submit",
"short_description": "Wrap the spark-submit binary to kick off a
spark-submit job; requires \"spark-submit\" binary in the PATH.",
"docs_url":
"https://airflow.apache.org/docs/apache-airflow-providers-apache-spark/stable/_api/airflow/providers/apache/spark/operators/spark_submit/index.html#airflow.providers.apache.spark.operators.spark_submit.SparkSubmitOperator";,
"source_url":
"https://github.com/apache/airflow/blob/providers-apache-spark/6.2.0/providers/apache/spark/src/airflow/providers/apache/spark/operators/spark_submit.py#L69";,
"category": "apache-spark",
"provider_id": "apache-spark",
"provider_name": "Apache Spark",
"supports_durable_execution": true
},
```
Databricks:
```shell
breeze registry extract-data --provider databricks
```
Output:
```json
Databricks
{
"id": "databricks-databricks-DatabricksRunNowOperator",
"name": "DatabricksRunNowOperator",
"type": "operator",
"import_path":
"airflow.providers.databricks.operators.databricks.DatabricksRunNowOperator",
"module_path": "airflow.providers.databricks.operators.databricks",
"short_description": "Runs an existing Spark job run to Databricks
using the api/2.2/jobs/run-now API endpoint.",
"docs_url":
"https://airflow.apache.org/docs/apache-airflow-providers-databricks/stable/_api/airflow/providers/databricks/operators/databricks/index.html#airflow.providers.databricks.operators.databricks.DatabricksRunNowOperator";,
"source_url":
"https://github.com/apache/airflow/blob/providers-databricks/7.17.0/providers/databricks/src/airflow/providers/databricks/operators/databricks.py#L983";,
"category": "databricks",
"provider_id": "databricks",
"provider_name": "Databricks",
"supports_durable_execution": true
},
{
"id": "databricks-databricks-DatabricksSubmitRunOperator",
"name": "DatabricksSubmitRunOperator",
"type": "operator",
"import_path":
"airflow.providers.databricks.operators.databricks.DatabricksSubmitRunOperator",
"module_path": "airflow.providers.databricks.operators.databricks",
"short_description": "Submits a Spark job run to Databricks using the
api/2.2/jobs/runs/submit API endpoint.",
"docs_url":
"https://airflow.apache.org/docs/apache-airflow-providers-databricks/stable/_api/airflow/providers/databricks/operators/databricks/index.html#airflow.providers.databricks.operators.databricks.DatabricksSubmitRunOperator";,
"source_url":
"https://github.com/apache/airflow/blob/providers-databricks/7.17.0/providers/databricks/src/airflow/providers/databricks/operators/databricks.py#L559";,
"category": "databricks",
"provider_id": "databricks",
"provider_name": "Databricks",
"supports_durable_execution": true
},
```
#### UI Testing
Build the UI with `pnpm install`, `pnpm dev` and visit the registry page
locally buillt.
This is how it looks for spark:
<img width="868" height="153" alt="image"
src="https://github.com/user-attachments/assets/a93f5f35-6b64-4d44-911b-ea689cd10087";
/>
<img width="868" height="153" alt="image"
src="https://github.com/user-attachments/assets/045c097f-875a-4b83-8d13-f5f66843e5af";
/>
### What's next
- A registry wide (cross-provider) `durable-execution` filter was considered
and dropped for now in favor of just the badge; revisit once more providers
adopt the mixin.
---
* Read the **[Pull Request
Guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/05_pull_requests.rst#pull-request-guidelines)**
for more information. Note: commit author/co-author name and email in commits
become permanently public when merged.
* For fundamental code changes, an Airflow Improvement Proposal
([AIP](https://cwiki.apache.org/confluence/display/AIRFLOW/Airflow+Improvement+Proposals))
is needed.
* When adding dependency, check compliance with the [ASF 3rd Party License
Policy](https://www.apache.org/legal/resolved.html#category-x).
* For significant user-facing changes create newsfragment:
`{pr_number}.significant.rst`, in
[airflow-core/newsfragments](https://github.com/apache/airflow/tree/main/airflow-core/newsfragments).
You can add this file in a follow-up commit after the PR is created so you
know the PR number.
--
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: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
