csp33 opened a new issue, #53866:
URL: https://github.com/apache/airflow/issues/53866
### Apache Airflow version
3.0.3
### If "Other Airflow 2 version" selected, which one?
_No response_
### What happened?
I have Airflow deployed in kubernetes, using git-sync to manage DAG files.
Sometimes I encounter that certain DAGs appear and disappear from the UI.
When checking the DAG processor logs, I find the following:
```
{"timestamp":"2025-07-29T09:19:16.431084","level":"info","event":"Filling up
the DagBag from
/opt/airflow/dags/repo/src/data_warehouse/ETL_CLIENTES/ETL_CLIENTES.py","logger":"airflow.models.dagbag.DagBag"}
{"timestamp":"2025-07-29T09:19:16.431508","level":"debug","event":"Importing
/opt/airflow/dags/repo/src/data_warehouse/ETL_CLIENTES/ETL_CLIENTES.py","logger":"airflow.models.dagbag.DagBag"}
{"timestamp":"2025-07-29T09:19:21.997978","level":"debug","event":"Initializing
Providers Manager[hook_lineage_writers]","logger":"airflow.providers_manager"}
{"timestamp":"2025-07-29T09:19:21.998101","level":"debug","event":"Initializing
Providers Manager[taskflow_decorators]","logger":"airflow.providers_manager"}
{"timestamp":"2025-07-29T09:19:21.998227","level":"debug","event":"Initialization
of Providers Manager[taskflow_decorators] took 0.00
seconds","logger":"airflow.providers_manager"}
{"timestamp":"2025-07-29T09:19:21.998263","level":"debug","event":"Initialization
of Providers Manager[hook_lineage_writers] took 0.00
seconds","logger":"airflow.providers_manager"}
{"timestamp":"2025-07-29T09:19:22.128794","level":"debug","event":"Loaded
DAG <DAG: obx_etl_clientes>","logger":"airflow.models.dagbag.DagBag"}
{"timestamp":"2025-07-29T09:19:22.128975","level":"debug","event":"Processing
Callback
Request","request":"{\"filepath\":\"data_warehouse/ETL_CLIENTES/ETL_CLIENTES.py\",\"bundle_name\":\"dags-folder\",\"bundle_version\":null,\"msg\":\"Executor
CeleryExecutor(parallelism=128) reported that the task instance <TaskInstance:
obx_etl_clientes.load_final_tables.snf_consolidate_clients_data
scheduled__2025-07-22T20:40:00+00:00 [running]> finished with state failed, but
the task instance's state attribute is running. Learn more:
https://airflow.apache.org/docs/apache-airflow/stable/troubleshooting.html#task-state-changed-externally\",\"ti\":{\"id\":\"019833dd-55c3-7bb5-b160-6fa27770249b\",\"task_id\":\"load_final_tables.snf_consolidate_clients_data\",\"dag_id\":\"obx_etl_clientes\",\"run_id\":\"scheduled__2025-07-22T20:40:00+00:00\",\"try_number\":1,\"map_index\":-1,\"hostname\":\"172.30.26.106\",\"context_carrier\":{}},\"task_callback_type\":null,\"type\":\"TaskCallbackRequest\"}","logge
r":"task"}
{"timestamp":"2025-07-29T09:19:22.200887Z","level":"warning","event":"Process
exited abnormally","exit_code":126,"logger":"processor"}
{"timestamp":"2025-07-29T09:19:27.542789","level":"info","event":"Filling up
the DagBag from
/opt/airflow/dags/repo/src/data_warehouse/ETL_CLIENTES/ETL_CLIENTES.py","logger":"airflow.models.dagbag.DagBag"}
```
It don't understand why an external task state change should affect the
parsing process of the DAG.
Also, when checking the DAG Processor pod logs I can see the following:
```
--- Supervised process Last chance exception handler ---
Traceback (most recent call last):
File
"/app/.venv/lib/python3.11/site-packages/airflow/sdk/execution_time/supervisor.py",
line 376, in _fork_main
target()
File
"/app/.venv/lib/python3.11/site-packages/airflow/dag_processing/processor.py",
line 126, in _parse_file_entrypoint
result = _parse_file(msg, log)
^^^^^^^^^^^^^^^^^^^^^
File
"/app/.venv/lib/python3.11/site-packages/airflow/dag_processing/processor.py",
line 142, in _parse_file
_execute_callbacks(bag, msg.callback_requests, log)
File
"/app/.venv/lib/python3.11/site-packages/airflow/dag_processing/processor.py",
line 182, in _execute_callbacks
raise NotImplementedError(
NotImplementedError: Haven't coded Task callback yet -
https://github.com/apache/airflow/issues/44354!
```
### What you think should happen instead?
Instead of making the DAG parse process fail, parsing should run without
errors and the task instance should be marked as failed.
### How to reproduce
Happens intermittently
### Operating System
Airflow in K8s
### Versions of Apache Airflow Providers
_No response_
### Deployment
Official Apache Airflow Helm Chart
### Deployment details
_No response_
### Anything else?
_No response_
### Are you willing to submit PR?
- [ ] Yes I am willing to submit a PR!
### Code of Conduct
- [x] I agree to follow this project's [Code of
Conduct](https://github.com/apache/airflow/blob/main/CODE_OF_CONDUCT.md)
--
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]
