Question: How to setup the client to access cluster wide (custom) resources

[MartijnStraatman]

I added a controller to our Kubernetes cluster which operates at the cluster wide level. I want to create object with this client.
So I created java code from the controller's crd's.

When i try to create an object, the client seems to be namespace-scoped by default.
For example:
https://xxxx:xxxx/apis/capsule.clastix.io/v1beta1/namespaces/default/tenants

This returns a 404 because this path does not exist at the api-server.

I expect it to be:
https://xxxx:xxxx/apis/capsule.clastix.io/v1beta1/tenants

How can I instrument the client so it won't use the namespace while creating the url.

        KubernetesClient client = new KubernetesClientBuilder().build();

        MixedOperation<io.clastix.capsule.v1beta1.Tenant,
                KubernetesResourceList<io.clastix.capsule.v1beta1.Tenant>, Resource<io.clastix.capsule.v1beta1.Tenant>>
                tenantClient = client.resources(io.clastix.capsule.v1beta1.Tenant.class);
       
       .....
         
       Object bla = tenantClient.resource(t).create();

I noticed while creating the java classes with the java-generator-cli-6.4.0.sh the Tenant class implements the io.fabric8.kubernetes.api.model.Namespaced interface. If I remove the interface, things work as expected..

Thx

[andreaTP]

Hi @MartijnStraatman , sorry for the delay, I missed this issue.
With a script like the:

#! /bin/bash
set -x

CRD=$1
CRD_NAME="${2:-default}"

# .XXXXXXX makes it cross-compatible Mac/Linux
TMP_DIR=$(mktemp -d -t $CRD_NAME.XXXXXXX)

cp ${CRD} $TMP_DIR/$CRD_NAME.yaml
mkdir -p $TMP_DIR/src

jbang io.fabric8:java-generator-cli:6.4.1 -u https://raw.githubusercontent.com/clastix/capsule/master/charts/capsule/crds/tenant-crd.yaml --target=$TMP_DIR/src

cat <<EOF >> $TMP_DIR/$CRD_NAME.java
//DEPS io.fabric8:kubernetes-client:6.2.0
//DEPS io.fabric8:generator-annotations:6.2.0
//DEPS io.sundr:builder-annotations:0.90.4
//DEPS org.projectlombok:lombok:1.18.24
//DEPS io.sundr:builder-annotations:0.90.4
//DEPS javax.validation:validation-api:2.0.1.Final
//SOURCES src/**.java
EOF

(
  cd $TMP_DIR
  jbang export portable --force $TMP_DIR/$CRD_NAME.java
  jbang --class-path $TMP_DIR/$CRD_NAME.jar -i
)

rm -rf $TMP_DIR

you have access to a shell session where you can reproduce the issue:

import io.fabric8.kubernetes.client.*
var client = new KubernetesClientBuilder().build();
client.resources(io.clastix.capsule.v1beta1.Tenant.class).create(new io.clastix.capsule.v1beta1.Tenant())

You get the following StackTrace:

|  Exception io.fabric8.kubernetes.client.KubernetesClientException: Failure executing: POST at: https://127.0.0.1:49635/apis/capsule.clastix.io/v1beta1/namespaces/default/tenants. Message: Not Found.
|        at KubernetesClientException.copyAsCause (KubernetesClientException.java:238)
|        at OperationSupport.waitForResult (OperationSupport.java:517)
|        at OperationSupport.handleResponse (OperationSupport.java:551)
|        at OperationSupport.handleResponse (OperationSupport.java:535)
|        at OperationSupport.handleCreate (OperationSupport.java:328)
|        at BaseOperation.handleCreate (BaseOperation.java:675)
|        at BaseOperation.handleCreate (BaseOperation.java:88)
|        at CreateOnlyResourceOperation.create (CreateOnlyResourceOperation.java:42)
|        at (#3:1)
|  Caused by: io.fabric8.kubernetes.client.KubernetesClientException: Failure executing: POST at: https://127.0.0.1:49635/apis/capsule.clastix.io/v1beta1/namespaces/default/tenants. Message: Not Found.
|        at OperationSupport.requestFailure (OperationSupport.java:709)
|        at OperationSupport.requestFailure (OperationSupport.java:689)
|        at OperationSupport.assertResponseCode (OperationSupport.java:640)
|        at OperationSupport.lambda$handleResponse$0 (OperationSupport.java:576)
|        at CompletableFuture$UniApply.tryFire (CompletableFuture.java:642)
|        at CompletableFuture.postComplete (CompletableFuture.java:506)
|        at CompletableFuture.complete (CompletableFuture.java:2073)
|        at OperationSupport.lambda$retryWithExponentialBackoff$2 (OperationSupport.java:618)
|        at CompletableFuture.uniWhenComplete (CompletableFuture.java:859)
|        at CompletableFuture$UniWhenComplete.tryFire (CompletableFuture.java:837)
|        at CompletableFuture.postComplete (CompletableFuture.java:506)
|        at CompletableFuture.complete (CompletableFuture.java:2073)
|        at OkHttpClientImpl$4.onResponse (OkHttpClientImpl.java:277)
|        at RealCall$AsyncCall.execute (RealCall.java:203)
|        at NamedRunnable.run (NamedRunnable.java:32)
|        at ThreadPoolExecutor.runWorker (ThreadPoolExecutor.java:1128)
|        at ThreadPoolExecutor$Worker.run (ThreadPoolExecutor.java:628)
|        at Thread.run (Thread.java:829)

cc. @manusa @shawkins

[manusa]

kubernetes-client/java-generator/core/src/main/java/io/fabric8/java/generator/nodes/JCRObject.java

Line 134 in 591d421

clz.addImplementedType("io.fabric8.kubernetes.api.model.Namespaced");

Should not always be added only for those CRDs with scope Namespaced

[andreaTP]

Thanks @manusa , yes the bug has been identified, will fix it in a while.

[MartijnStraatman]

thx!!

[andreaTP]

Thank you for reporting! Don't know how we could have missed it 😅

[MartijnStraatman]

We are building some kind of self-service api for users to create tenants on multi-tenant k8s clusters. This is a clusterscoped resource, hence we stumbled upon it.