milindl opened a new pull request, #1166:
URL: https://github.com/apache/flink-kubernetes-operator/pull/1166
## What is the purpose of the change
The state of the FlinkSessionJob (`status.jobStatus.state` and
`status.state`) is stuck and not updated in case of an UpgradeFailureException
(for instance, if the job ends quickly and the checkpointing interval is set,
but state.checkpoints.dir is not configured).
## Brief change log
- Moves observer.observe() inside try-catch and add explicit catch for
UpgradeFailureException, similar to FlinkDeploymentController.
- `triggerErrorEvent` now accepts a reason similar to
FlinkDeploymentController
## Verifying this change
This change added tests and can be verified as follows:
- testJobFinishedWithNonExternallyAddressableCheckpoint: Submits a session
job, simulates it finishing with a checkpoint whose external pointer is
non-externally-addressable (checkpointing enabled but no
state.checkpoints.dir), and verifies the job status correctly reflects FINISHED
with the error recorded.
- Manual testing using minikube using the exact JIRA reproducer in the
ticket (SleepJob with 5s delay, execution.checkpointing.interval: 2000, no
state.checkpoints.dir):
- FKO 1.15.0: job stuck at RUNNING/STABLE despite Flink REST API showing
FINISHED
- Patched FKO: job correctly transitions to FINISHED/STABLE with error
recorded
## Does this pull request potentially affect one of the following parts:
- Dependencies (does it add or upgrade a dependency): no
- The public API, i.e., is any changes to the `CustomResourceDescriptors`:
no
- Core observer or reconciler logic that is regularly executed: no (I
think)
## Documentation
- Does this pull request introduce a new feature? no
--
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]
