Make Local Queue label optional

src/codeflare_sdk/utils/generate_yaml.py

@@ -30,6 +30,7 @@
 from os import urandom
 from base64 import b64encode
 from urllib3.util import parse_url
+from kubernetes.client.exceptions import ApiException
 
 
 def read_template(template):
@@ -191,8 +192,11 @@ def get_default_kueue_name(namespace: str):
             namespace=namespace,
             plural="localqueues",
         )
-    except Exception as e:  # pragma: no cover
-        return _kube_api_error_handling(e)
+    except ApiException as e:  # pragma: no cover
+        if e.status == 404 or e.status == 403:
+            return
+        else:
+            return _kube_api_error_handling(e)
     for lq in local_queues["items"]:
         if (
             "annotations" in lq["metadata"]
@@ -201,9 +205,6 @@ def get_default_kueue_name(namespace: str):
             == "true"
         ):
             return lq["metadata"]["name"]
-    raise ValueError(
-        "Default Local Queue with kueue.x-k8s.io/default-queue: true annotation not found please create a default Local Queue or provide the local_queue name in Cluster Configuration"
-    )
 
 
 def local_queue_exists(namespace: str, local_queue_name: str):
@@ -228,7 +229,9 @@ def local_queue_exists(namespace: str, local_queue_name: str):
 
 def add_queue_label(item: dict, namespace: str, local_queue: Optional[str]):
     lq_name = local_queue or get_default_kueue_name(namespace)
-    if not local_queue_exists(namespace, lq_name):
+    if lq_name == None:
+        return
+    elif not local_queue_exists(namespace, lq_name):
         raise ValueError(
             "local_queue provided does not exist or is not in this namespace. Please provide the correct local_queue name in Cluster Configuration"
         )

tests/test-case-no-kueue-no-aw.yaml

@@ -0,0 +1,153 @@
+apiVersion: ray.io/v1
+kind: RayCluster
+metadata:
+  annotations:
+    app.kubernetes.io/managed-by: test-prefix
+  labels:
+    controller-tools.k8s.io: '1.0'
+  name: unit-test-no-kueue
+  namespace: ns
+spec:
+  autoscalerOptions:
+    idleTimeoutSeconds: 60
+    imagePullPolicy: Always
+    resources:
+      limits:
+        cpu: 500m
+        memory: 512Mi
+      requests:
+        cpu: 500m
+        memory: 512Mi
+    upscalingMode: Default
+  enableInTreeAutoscaling: false
+  headGroupSpec:
+    enableIngress: false
+    rayStartParams:
+      block: 'true'
+      dashboard-host: 0.0.0.0
+      num-gpus: '0'
+    serviceType: ClusterIP
+    template:
+      spec:
+        containers:
+        - image: quay.io/rhoai/ray:2.23.0-py39-cu121
+          imagePullPolicy: Always
+          lifecycle:
+            preStop:
+              exec:
+                command:
+                - /bin/sh
+                - -c
+                - ray stop
+          name: ray-head
+          ports:
+          - containerPort: 6379
+            name: gcs
+          - containerPort: 8265
+            name: dashboard
+          - containerPort: 10001
+            name: client
+          resources:
+            limits:
+              cpu: 2
+              memory: 8G
+              nvidia.com/gpu: 0
+            requests:
+              cpu: 2
+              memory: 8G
+              nvidia.com/gpu: 0
+          volumeMounts:
+          - mountPath: /etc/pki/tls/certs/odh-trusted-ca-bundle.crt
+            name: odh-trusted-ca-cert
+            subPath: odh-trusted-ca-bundle.crt
+          - mountPath: /etc/ssl/certs/odh-trusted-ca-bundle.crt
+            name: odh-trusted-ca-cert
+            subPath: odh-trusted-ca-bundle.crt
+          - mountPath: /etc/pki/tls/certs/odh-ca-bundle.crt
+            name: odh-ca-cert
+            subPath: odh-ca-bundle.crt
+          - mountPath: /etc/ssl/certs/odh-ca-bundle.crt
+            name: odh-ca-cert
+            subPath: odh-ca-bundle.crt
+        imagePullSecrets:
+        - name: unit-test-pull-secret
+        volumes:
+        - configMap:
+            items:
+            - key: ca-bundle.crt
+              path: odh-trusted-ca-bundle.crt
+            name: odh-trusted-ca-bundle
+            optional: true
+          name: odh-trusted-ca-cert
+        - configMap:
+            items:
+            - key: odh-ca-bundle.crt
+              path: odh-ca-bundle.crt
+            name: odh-trusted-ca-bundle
+            optional: true
+          name: odh-ca-cert
+  rayVersion: 2.23.0
+  workerGroupSpecs:
+  - groupName: small-group-unit-test-no-kueue
+    maxReplicas: 2
+    minReplicas: 2
+    rayStartParams:
+      block: 'true'
+      num-gpus: '7'
+    replicas: 2
+    template:
+      metadata:
+        annotations:
+          key: value
+        labels:
+          key: value
+      spec:
+        containers:
+        - image: quay.io/rhoai/ray:2.23.0-py39-cu121
+          lifecycle:
+            preStop:
+              exec:
+                command:
+                - /bin/sh
+                - -c
+                - ray stop
+          name: machine-learning
+          resources:
+            limits:
+              cpu: 4
+              memory: 6G
+              nvidia.com/gpu: 7
+            requests:
+              cpu: 3
+              memory: 5G
+              nvidia.com/gpu: 7
+          volumeMounts:
+          - mountPath: /etc/pki/tls/certs/odh-trusted-ca-bundle.crt
+            name: odh-trusted-ca-cert
+            subPath: odh-trusted-ca-bundle.crt
+          - mountPath: /etc/ssl/certs/odh-trusted-ca-bundle.crt
+            name: odh-trusted-ca-cert
+            subPath: odh-trusted-ca-bundle.crt
+          - mountPath: /etc/pki/tls/certs/odh-ca-bundle.crt
+            name: odh-ca-cert
+            subPath: odh-ca-bundle.crt
+          - mountPath: /etc/ssl/certs/odh-ca-bundle.crt
+            name: odh-ca-cert
+            subPath: odh-ca-bundle.crt
+        imagePullSecrets:
+        - name: unit-test-pull-secret
+        volumes:
+        - configMap:
+            items:
+            - key: ca-bundle.crt
+              path: odh-trusted-ca-bundle.crt
+            name: odh-trusted-ca-bundle
+            optional: true
+          name: odh-trusted-ca-cert
+        - configMap:
+            items:
+            - key: odh-ca-bundle.crt
+              path: odh-ca-bundle.crt
+            name: odh-trusted-ca-bundle
+            optional: true
+          name: odh-ca-cert

tests/unit_test.py

@@ -285,6 +285,28 @@ def test_cluster_creation(mocker):
     )
 
 
+def test_cluster_no_kueue_no_aw(mocker):
+    mocker.patch("kubernetes.client.ApisApi.get_api_versions")
+    mocker.patch(
+        "kubernetes.client.CustomObjectsApi.get_cluster_custom_object",
+        return_value={"spec": {"domain": "apps.cluster.awsroute.org"}},
+    )
+    mocker.patch("kubernetes.client.CustomObjectsApi.list_namespaced_custom_object")
+    mocker.patch("os.environ.get", return_value="test-prefix")
+    config = createClusterConfig()
+    config.appwrapper = False
+    config.name = "unit-test-no-kueue"
+    config.write_to_file = True
+    cluster = Cluster(config)
+    assert cluster.app_wrapper_yaml == f"{aw_dir}unit-test-no-kueue.yaml"
+    assert cluster.config.local_queue == None
+    assert filecmp.cmp(
+        f"{aw_dir}unit-test-no-kueue.yaml",
+        f"{parent}/tests/test-case-no-kueue-no-aw.yaml",
+        shallow=True,
+    )
+
+
 def test_create_app_wrapper_raises_error_with_no_image():
     config = createClusterConfig()
     config.image = ""  # Clear the image to test error handling
@@ -2799,6 +2821,7 @@ def test_rjc_list_jobs(ray_job_client, mocker):
 
 # Make sure to always keep this function last
 def test_cleanup():
+    os.remove(f"{aw_dir}unit-test-no-kueue.yaml")
     os.remove(f"{aw_dir}unit-test-cluster.yaml")
     os.remove(f"{aw_dir}test.yaml")
     os.remove(f"{aw_dir}raytest2.yaml")