Dennis-Mircea opened a new pull request, #1122:
URL: https://github.com/apache/flink-kubernetes-operator/pull/1122
## What is the purpose of the change
`FlinkConfigManager` already discovers whether the `FlinkStateSnapshot` CRD
is installed at operator startup (via the boolean it accepts from
`FlinkOperator`) and stores the result in a final field. Despite that, both
`FlinkDeploymentController` and `FlinkSessionJobController` were calling
`KubernetesClientUtils.isCrdInstalled(FlinkStateSnapshot.class)` again from
their `prepareEventSources` methods. That helper does not just hit the API
server, it also constructs a brand-new `KubernetesClient` (`new
KubernetesClientBuilder().build()`) with its own TLS handshake and auth chain,
then issues a `list().getItems()` call. The result is three independent API
roundtrips and three transient clients to answer the same question whose answer
is already cached on `FlinkConfigManager`. This PR exposes that flag through a
getter and switches both controllers to use it.
## Brief change log
- Added `FlinkConfigManager#isSnapshotCrdInstalled()` (plain Java getter,
no Lombok) returning the existing `snapshotCrdInstalled` field.
- Replaced
`KubernetesClientUtils.isCrdInstalled(FlinkStateSnapshot.class)` with
`flinkConfigManager.isSnapshotCrdInstalled()` in
`FlinkDeploymentController#prepareEventSources` and
`FlinkSessionJobController#prepareEventSources`.
- Removed the now-unused imports `KubernetesClientUtils` and
`FlinkStateSnapshot` from both controllers.
- Added a `FlinkConfigManager` field + constructor parameter on
`FlinkSessionJobController` (it did not have one before). Updated
`FlinkOperator#registerSessionJobController` and
`TestingFlinkSessionJobController` to pass it in.
- While in `FlinkDeploymentController`, specialized two method signatures
from raw `Context` to `Context<FlinkDeployment>` (`cleanup` and `reconcile`),
resolving raw-type warnings the IDE was flagging in the file we were already
editing.
## Verifying this change
This change is already covered by existing tests, such as
`FlinkDeploymentControllerTest`, `FlinkSessionJobControllerTest`,
`FlinkStateSnapshotControllerTest`, `FlinkConfigManagerTest`, and
`FlinkOperatorTest`.
Behavioral equivalence: `prepareEventSources` is invoked once at JOSDK
controller registration. The `snapshotCrdInstalled` flag is computed once at
`FlinkOperator` construction (right before `FlinkConfigManager` is built). Both
points happen during operator startup with no possibility of the CRD state
changing between them, so the answer used by both controllers is identical to
today's behavior. The only difference is that we no longer build two transient
`KubernetesClient` instances and issue two extra `list()` calls.
## 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
## Documentation
- Does this pull request introduce a new feature? no
- If yes, how is the feature documented? not applicable
--
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]
