[jira] [Resolved] (AIRFLOW-6397) Check for `sub_process` before trying to get pid in bash operator on kill

["Jarek Potiuk (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/AIRFLOW-6397?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jarek Potiuk resolved AIRFLOW-6397.
-----------------------------------
    Fix Version/s: 2.0.0
       Resolution: Fixed

> Check for `sub_process` before trying to get pid in bash operator on kill
> -------------------------------------------------------------------------
>
>                 Key: AIRFLOW-6397
>                 URL: https://issues.apache.org/jira/browse/AIRFLOW-6397
>             Project: Apache Airflow
>          Issue Type: Bug
>          Components: operators
>    Affects Versions: 1.10.7
>            Reporter: Daniel Standish
>            Assignee: Daniel Standish
>            Priority: Major
>             Fix For: 2.0.0
>
>
> test {{test_mark_success_no_kill}} in {{TestLocalTaskJob}} is very flakey
> i found that one reason is that the test may attempt to kill the task before the subprocess is created and stored as an attribute
> For example:
> {code}
> [2019-12-29 15:03:51,963] {bash_operator.py:116} INFO - Running command: sleep 600
> [2019-12-29 15:03:51,967] {helpers.py:315} INFO - Sending Signals.SIGTERM to GPID 57093
> [2019-12-29 15:03:51,968] {taskinstance.py:913} ERROR - Received SIGTERM. Terminating subprocesses.
> [2019-12-29 15:03:51,970] {bash_operator.py:143} INFO - Sending SIGTERM signal to bash process group
> [2019-12-29 15:03:51,970] {taskinstance.py:913} ERROR - Received SIGTERM. Terminating subprocesses.
> [2019-12-29 15:03:51,982] {bash_operator.py:143} INFO - Sending SIGTERM signal to bash process group
> [2019-12-29 15:03:51,982] {taskinstance.py:1078} ERROR - 'BashOperator' object has no attribute 'sub_process'
> Traceback (most recent call last):
>   File "/Users/dstandish/code/airflow/airflow/models/taskinstance.py", line 945, in _run_raw_task
>     result = task_copy.execute(context=context)
>   File "/Users/dstandish/code/airflow/airflow/operators/bash_operator.py", line 123, in execute
>     preexec_fn=pre_exec)
>   File "/Users/dstandish/.pyenv/versions/3.7.5/lib/python3.7/subprocess.py", line 800, in __init__
>     restore_signals, start_new_session)
>   File "/Users/dstandish/.pyenv/versions/3.7.5/lib/python3.7/subprocess.py", line 1505, in _execute_child
>     part = os.read(errpipe_read, 50000)
>   File "/Users/dstandish/code/airflow/airflow/models/taskinstance.py", line 914, in signal_handler
>     task_copy.on_kill()
>   File "/Users/dstandish/code/airflow/airflow/operators/bash_operator.py", line 144, in on_kill
>     os.killpg(os.getpgid(self.sub_process.pid), signal.SIGTERM)
> AttributeError: 'BashOperator' object has no attribute 'sub_process'
> [2019-12-29 15:03:51,988] {taskinstance.py:1123} INFO - Marking task as FAILED.dag_id=test_mark_success, task_id=task1, execution_date=20160101T000000, start_date=20191229T230351, end_date=20191229T230351
> [2019-12-29 15:03:52,022] {helpers.py:281} INFO - Process psutil.Process(pid=57093, status='terminated') (57093) terminated with exit code 1
> {code}



--
This message was sent by Atlassian Jira
(v8.3.4#803005)