You are viewing a plain text version of this content. The canonical link for it is here.

Posted to commits@airflow.apache.org by "amoghrajesh (via GitHub)" <gi...@apache.org> on 2026/05/26 11:16:04 UTC

[PR] Unify task/asset state storage between Core API and Execution API [airflow]

amoghrajesh opened a new pull request, #67547:
URL: https://github.com/apache/airflow/pull/67547

    <!-- 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.
   
   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".
   -->
   
   - [x] Yes: claude sonnet 4.6
   
   <!--
   Generated-by: [Tool Name] following [the guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/05_pull_requests.rst#gen-ai-assisted-contributions)
   -->
   
   ### What problem are we solving?
   
   Task and asset state written by workers (via the Execution API) is stored as JSON-encoded strings in the DB — e.g., the integer `42` is stored as `"42"`, and the string `"hello"` as `'"hello"'`. The core API read path returned the raw DB string without decoding it, so callers saw `'"hello"'` instead of `"hello"` and `"42"` (a string) instead of `42` (an integer). 
   
   The write path had the inverse bug: values posted to the core API were stored without `json.dumps`, so a worker reading them back via the execution API would hit a `JSONDecodeError`.
   
   
   ### Current behaviour
   - GET `/dags/{dag_id}/dagRuns/{run_id}/taskInstances/{task_id}/states` and the equivalent asset state endpoints return raw JSON-encoded strings in the value field (e.g. "\"hello\"", "42" as a string).
   - PUT on these endpoints stores the value as-is without JSON-encoding, so values written via the Core API are unreadable by workers.
   - The value field on request/response models is typed as str, with no null or size validation.
   
   Classic example is if I run this task:
   
   ```python
       @task
       def my_task(**context: Context):
           task_state = context["task_state"]
   
           task_state.set("job_id", "12345")
           task_state.set("secret-dict", {"key": "value"})
           task_state.set("int_value", 42)
   
           print("Fetching task states I stored earlier")
   
           print("job_id:", task_state.get("job_id"))
           print("secret-dict:", task_state.get("secret-dict"))
           print("int_value:", task_state.get("int_value"))
   
       my_task()
   ```
   
   Core API call to get all task states returned this:
   
   ```shell
   {
       "task_states": [
           {
               "key": "int_value",
               "value": "42",
               "updated_at": "2026-05-26T10:07:01.872139Z",
               "expires_at": "2026-06-25T10:07:01.868443Z"
           },
           {
               "key": "job_id",
               "value": "\"12345\"",
               "updated_at": "2026-05-26T10:07:01.855922Z",
               "expires_at": "2026-06-25T10:07:01.851283Z"
           },
           {
               "key": "secret-dict",
               "value": "{\"key\": \"value\"}",
               "updated_at": "2026-05-26T10:07:01.864312Z",
               "expires_at": "2026-06-25T10:07:01.860810Z"
           }
       ],
       "total_entries": 3
   }
   ```
   
   ### Proposed change
   - `value: str` widened to `value: JsonValue` on all four core API datamodels (TaskStateBody, TaskStateResponse, AssetStateBody, AssetStateResponse).
   - Write paths now call `json.dumps(body.value)` before passing to the backend, matching the execution API as introduced in https://github.com/apache/airflow/pull/67418
   - Read paths now call `json.loads(r.value)` when constructing responses, so callers receive native types (int, dict, list, bool, str).
   - Body models now have validators rejecting `null`, `non-finite floats (NaN, Inf, -Inf)`, and values whose serialized form exceeds 65535 bytes (preserving the old max_length constraint).
   
   NOTE: The UI PR: https://github.com/apache/airflow/pull/67292 sends in raw values as entered by user, so no impact there.
   
   ### Testing
   
   
   #### Test 1: Set using task execution, Get using core API and task execution should have same response type
   
   <img width="1708" height="855" alt="image" src="https://github.com/user-attachments/assets/f9708969-bf6b-4a3f-8df4-dc797728479c" />
   
   
   <img width="1301" height="1213" alt="image" src="https://github.com/user-attachments/assets/9006df0a-8ea6-4c11-9fc9-f30730e702ed" />
   
   
   #### Test 2: Value set using core API, getting via task execution and core API should have same response type
   
   Set values using curl for this task:
   ```python
       @task
       def my_task(**context: Context):
           task_state = context["task_state"]
   
           print("Fetching task states that are set by API")
   
           print("str_value:", task_state.get("str_value"), type(task_state.get("str_value")))
           print("dict-value:", task_state.get("dict-value"), type(task_state.get("dict-value")))
           print("int_value:", task_state.get("int_value"), type(task_state.get("int_value")))
   
       my_task()
   ```
   
   <img width="1708" height="855" alt="image" src="https://github.com/user-attachments/assets/8b91ba8d-3b6f-4892-8040-deda6270f0c1" />
   
   
   <img width="1455" height="1098" alt="image" src="https://github.com/user-attachments/assets/5f9c42c1-8bcc-401b-bb54-aa28975c32e7" />
   
   
   
   ---
   
   * 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: commits-unsubscribe@airflow.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org