-
Notifications
You must be signed in to change notification settings - Fork 16.4k
Enable task run setting to be able reinitialise #19624
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Enable task run setting to be able reinitialise #19624
Conversation
eef4c2e to
956852b
Compare
|
The PR most likely needs to run full matrix of tests because it modifies parts of the core of Airflow. However, committers might decide to merge it quickly and take the risk. If they don't merge it quickly - please rebase it to the latest main at your convenience, or amend the last commit of the PR, and push it with --force-with-lease. |
956852b to
679a638
Compare
|
@potiuk can you please add "debug-ci-resources" label on this? |
|
Added/closed/reopened to retest |
679a638 to
0d823bf
Compare
|
FYI: I am now using this PR to debug CI failures. Once done will file a new one for the original purpose |
fd48ba8 to
ef8a10a
Compare
|
Closing to open a new one as this got a bit messy due to debug |
When task is executed using "StandardTaskRunner._start_by_exec()" (when fork is not available) it's using old DONOT_MODIFY_HANDLERS config variable loaded during settings.py load and not the one from externally supplied --cfg-path param
TL;DR
When "tasks run" command is executed it then internally re-executed as below command using external process (non fork when it's not available).
[2021-11-16 15:28:16,818] {base_task_runner.py:144} INFO - Running: ['airflow', 'tasks', 'run', 'test_logging_dag', 'test_task', 'test_run', '--job-id', '102', '--raw', '--subdir', 'DAGS_FOLDER', '--cfg-path', '/tmp/tmpfv4cj9py', '--error-file', '/tmp/tmphjtaj5gi']Here
'--cfg-path', '/tmp/tmpfv4cj9py'is a copy of active config settings of parent process. When command is executed settings.py file is loaded as one of the first things but external config (/tmp/tmpfv4cj9py) is load later within task_command.py->def task_run.And when it does load it does not re-initialize this setting.
This causes DONOT_MODIFY_HANDLERS to be initialized to a default value rather than one that is defined in the /tmp/tmpfv4cj9py
The PR fixes it by moving initialization into settings.py/configure_vars function where it can be called when needed.
PS:
I think this issue is not limited to this variable but all other similar GLOBAL variables that is initialized during settings.py load.
But I think we need to check each of them separately before making similar changes.
^ Add meaningful description above
Read the Pull Request Guidelines for more information.
In case of fundamental code change, Airflow Improvement Proposal (AIP) is needed.
In case of a new dependency, check compliance with the ASF 3rd Party License Policy.
In case of backwards incompatible changes please leave a note in UPDATING.md.