Fix KubernetesJobOperator.get_or_create_pod() #52885
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
boring-cyborg
bot
added
area:providers
provider:cncf-kubernetes
|
Congratulations on your first Pull Request and welcome to the Apache Airflow community! If you have any issues or are unsure about any anything please check our Contributors' Guide (https://github.com/apache/airflow/blob/main/contributing-docs/README.rst)
|
stephen-bracken
requested review from
hussein-awala and
jedcunningham
as code owners
dominikhei
reviewed
Jul 4, 2025
|
|
||
| def get_or_create_pod(self, pod_request_obj, context): | ||
| pod = None | ||
| while pod is None: |
Contributor
There was a problem hiding this comment.
Doesn't this risk an infinite loop?
Contributor
Author
There was a problem hiding this comment.
Good point. I was relying on the task timeout to interrupt this, which is a naive assumption. I can add a static timeout, or add a retry with backoff here or use the activeDeadlineSeconds from the job spec?
dominikhei
reviewed
Jul 11, 2025
| @patch(JOB_OPERATORS_PATH.format("KubernetesJobOperator.build_job_request_obj")) | ||
| @patch(JOB_OPERATORS_PATH.format("KubernetesJobOperator.create_job")) | ||
| @patch(HOOK_CLASS) | ||
| def test_pod_creation_timeout(self, mock_hook, mock_create_job, mock_build_job_request_obj, mock_find_pod): |
Contributor
There was a problem hiding this comment.
If I am not wrong this tests waits the whole 10s you used as a timeout. Could you mock the behavior, such that the test does not run for 10s or set it to 0?
Contributor
Author
There was a problem hiding this comment.
sure, I'll set it to 1
dominikhei
reviewed
Jul 11, 2025
Contributor
dominikhei
left a comment
dominikhei
left a comment
There was a problem hiding this comment.
I don’t have a strong preference between introducing a new parameter or defaulting to a few retries with exponential backoff. cc: @hussein-awala
apache#52908) The file might be a left-over from earlier api generation and it is already .gitignored, so it does not appear in the source tarball, but sdist and wheel generation does not look at .gitignored files (for a good reason - we might want to include some generated files in those distributions) - so we should exclude it manually similarly as we already do with node_modules for example.
Co-authored-by: Amogh Desai <amoghrajesh1999@gmail.com>
…pache#52909) I noticed that we did not have `pre-commit-uv` in the image that was building the distribution packages. Adding it should slightly speed up building such an image for isolated package preparation. The `uv` has been upgraded along the way, also the comment that we should automate updating the versions in release_management_command.py has been removed - because we already have automation in place. The missing pieces were adding "hatch", "pyyaml", "gitpython" and "rich".
…he#52931) We are still waiting for release of the Lucas's pre-commit after the Lucas-C/pre-commit-hooks#103 has been merged - and for now we need to use `bleeding-edge` for the repo. Also upgrades zizmor and solves using "env." in shell commands that might lead to security issues.
stephen-bracken
requested review from
Fokko,
XD-DENG,
amoghrajesh,
bbovenzi,
bolkedebruin,
bugraoz93,
dstandish,
eladkal,
ephraimbuddy,
gopidesupavan,
jason810496,
josh-fell,
jscheffl,
kaxil,
mik-laj,
mobuchowski,
o-nikolas,
pierrejeambrun,
potiuk,
rawwar,
ryanahamilton,
shahar1,
shubhamraj-git,
uranusjr and
vincbeck
as code owners
Contributor
Author
|
argh I've messed up the diff for this PR - going to close and reopen as fresh |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
20 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fix KubernetesJobOperator.get_or_create_pod() sometimes creating duplicate pods.
during
execute()the KubernetesJobOperator attempts to find the pod associated with the Job object usingself.get_or_create_pod(). If Kubernetes is being slow then the Job object will not create a pod before this method gets called, which will result in the underlyingfind_pod()method returningNone, and a duplicate headless pod being created for this task.This PR overrides the
get_or_create_pod()method inKubernetesJobOperatorto poll for the pod associated with the Job if one is not created immediately.^ Add meaningful description above
Read the Pull Request Guidelines for more information.
In case of fundamental code changes, an 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 a newsfragment file, named
{pr_number}.significant.rstor{issue_number}.significant.rst, in airflow-core/newsfragments.