Kubernetes-client: KubernetesServer CRUD mode and informers

[lalithsuresh]

I'd like to make informers work with the CRUD API. The current behavior seems to break with the reflector code within the informer. For example:

    @Test
    void testCrudInformer() throws InterruptedException {
        final KubernetesServer server = new KubernetesServer(true, true);
        server.before();

        KubernetesClient client = server.getClient();
        SharedInformerFactory factory = client.informers();
        SharedIndexInformer<Pod> podInformer = factory.sharedIndexInformerFor(Pod.class, PodList.class, 4000);
        podInformer.addEventHandler(
                new ResourceEventHandler<Pod>() {
                    @Override
                    public void onAdd(Pod obj) {
                    }

                    @Override
                    public void onUpdate(Pod oldObj, Pod newObj) {
                    }

                    @Override
                    public void onDelete(Pod oldObj, boolean deletedFinalStateUnknown) {
                    }
                });
        factory.startAllRegisteredInformers();
        Thread.sleep(5000);
        factory.stopAllRegisteredInformers();
    }

Leads to the following NPE in the reflector....

Jun 22, 2020 7:15:46 PM okhttp3.mockwebserver.MockWebServer$2 execute
INFO: MockWebServer[58658] starting to accept connections
19:15:46.956 [informer-controller-Pod] INFO  io.fabric8.kubernetes.client.informers.cache.Controller - informer#Controller: ready to run resync and reflector runnable
19:15:46.967 [informer-controller-Pod] INFO  io.fabric8.kubernetes.client.informers.cache.Reflector - Started ReflectorRunnable watch for class io.fabric8.kubernetes.api.model.Pod
Jun 22, 2020 7:15:47 PM okhttp3.mockwebserver.MockWebServer$3 processOneRequest
INFO: MockWebServer[58658] received request: GET /api/v1/namespaces/test/pods HTTP/1.1 and responded: HTTP/1.1 200 OK
19:15:47.397 [informer-controller-Pod] WARN  io.fabric8.kubernetes.client.informers.cache.Controller - Reflector list-watching job exiting because the thread-pool is shutting down
 java.util.concurrent.RejectedExecutionException: Error while starting ReflectorRunnable watch
    at io.fabric8.kubernetes.client.informers.cache.Reflector.listAndWatch(Reflector.java:85) ~[kubernetes-client-4.10.1.jar:?]
    at io.fabric8.kubernetes.client.informers.cache.Controller.run(Controller.java:112) ~[kubernetes-client-4.10.1.jar:?]
    at java.lang.Thread.run(Thread.java:835) [?:?]
Caused by: java.lang.NullPointerException
    at io.fabric8.kubernetes.client.informers.cache.Reflector.reListAndSync(Reflector.java:95) ~[kubernetes-client-4.10.1.jar:?]
    at io.fabric8.kubernetes.client.informers.cache.Reflector.listAndWatch(Reflector.java:80) ~[kubernetes-client-4.10.1.jar:?]
    ... 2 more

Which in turn happens because the list type (PodList) that the reflector gets back has a null metadata object:

  private void reListAndSync() {
    final L list = getList();
    final String latestResourceVersion = list.getMetadata().getResourceVersion(); // NPE: Metadata is null
    ...

[rohanKanojia]

Right now Reflector seems to be failing since list returned by mockwebserver seems to have null metadata:

https://github.com/fabric8io/kubernetes-client/blob/55cf52c4e27adafa52052cea9cbca6d1ede1b07b/kubernetes-client/src/main/java/io/fabric8/kubernetes/client/informers/cache/Reflector.java#L95

However, a real Kubernetes server returns metadata too:

~/go/src/github.com/fabric8io/kubernetes-client/kubernetes-tests : $ kubectl get pods -ojson
{
    "apiVersion": "v1",
    "items": [],
    "kind": "List",
    "metadata": {
        "resourceVersion": "",
        "selfLink": ""
    }
}

Anyone who wants to fix this issue needs to update this method with metadata updated:

https://github.com/fabric8io/kubernetes-client/blob/55cf52c4e27adafa52052cea9cbca6d1ede1b07b/kubernetes-server-mock/src/main/java/io/fabric8/kubernetes/client/server/mock/KubernetesResponseComposer.java#L39-L44

[lalithsuresh]

Thanks @rohanKanojia. I'll send over a PR soon.