RHOAIENG-32532: Fix broken E2E tests

Author: kryanbeane

.github/workflows/e2e_tests.yaml

@@ -57,7 +57,7 @@ jobs:
       - name: Set up specific Python version
         uses: actions/setup-python@v5
         with:
-          python-version: '3.11'
+          python-version: '3.12'
           cache: 'pip' # caching pip dependencies
 
       - name: Setup NVidia GPU environment for KinD

.github/workflows/rayjob_e2e_tests.yaml

@@ -1,4 +1,4 @@
-name: rayjob-e2e-with-kueue
+name: rayjob-e2e
 
 on:
   pull_request:

src/codeflare_sdk/common/utils/k8s_utils.py

@@ -7,14 +7,10 @@
 from ..kubernetes_cluster import config_check, _kube_api_error_handling
 
 
-def get_current_namespace():
+def get_current_namespace():  # pragma: no cover
     """
     Retrieves the current Kubernetes namespace.
 
-    This function attempts to detect the current namespace by:
-    1. First checking if running inside a pod (reading from service account namespace file)
-    2. Falling back to reading from the current kubeconfig context
-
     Returns:
         str:
             The current namespace or None if not found.

src/codeflare_sdk/ray/cluster/build_ray_cluster.py

@@ -136,7 +136,6 @@ def build_ray_cluster(cluster: "codeflare_sdk.ray.cluster.Cluster"):
                 "enableIngress": False,
                 "rayStartParams": {
                     "dashboard-host": "0.0.0.0",
-                    "dashboard-port": "8265",
                     "block": "true",
                     "num-gpus": str(head_gpu_count),
                     "resources": head_resources,

src/codeflare_sdk/ray/cluster/cluster.py

@@ -783,6 +783,12 @@ def remove_autogenerated_fields(resource):
                 del resource[key]
             else:
                 remove_autogenerated_fields(resource[key])
+
+        # After cleaning, remove empty metadata sections
+        if "metadata" in resource and isinstance(resource["metadata"], dict):
+            if len(resource["metadata"]) == 0:
+                del resource["metadata"]
+
     elif isinstance(resource, list):
         for item in resource:
             remove_autogenerated_fields(item)

src/codeflare_sdk/ray/cluster/test_config.py

@@ -1,4 +1,4 @@
-# Copyright 2022-2025 IBM, Red Hat
+# Copyright 2024 IBM, Red Hat
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

tests/e2e/cluster_apply_kind_test.py

@@ -32,12 +32,13 @@ def test_cluster_apply(self):
             num_workers=1,
             head_cpu_requests="500m",
             head_cpu_limits="1",
-            head_memory_requests="1Gi",
-            head_memory_limits="2Gi",
+            head_memory_requests="5",
+            head_memory_limits="6",
             worker_cpu_requests="500m",
             worker_cpu_limits="1",
-            worker_memory_requests="1Gi",
-            worker_memory_limits="2Gi",
+            worker_memory_requests="3",
+            worker_memory_limits="4",
+            image="rayproject/ray:2.47.1",
             write_to_file=True,
             verify_tls=False,
         )
@@ -65,12 +66,13 @@ def test_cluster_apply(self):
             num_workers=2,
             head_cpu_requests="500m",
             head_cpu_limits="1",
-            head_memory_requests="1Gi",
-            head_memory_limits="2Gi",
+            head_memory_requests="5",
+            head_memory_limits="6",
             worker_cpu_requests="500m",
             worker_cpu_limits="1",
-            worker_memory_requests="1Gi",
-            worker_memory_limits="2Gi",
+            worker_memory_requests="3",
+            worker_memory_limits="4",
+            image="rayproject/ray:2.47.1",
             write_to_file=True,
             verify_tls=False,
         )
@@ -111,12 +113,13 @@ def test_apply_invalid_update(self):
             num_workers=1,
             head_cpu_requests="500m",
             head_cpu_limits="1",
-            head_memory_requests="1Gi",
-            head_memory_limits="2Gi",
+            head_memory_requests="5",
+            head_memory_limits="6",
             worker_cpu_requests="500m",
             worker_cpu_limits="1",
-            worker_memory_requests="1Gi",
-            worker_memory_limits="2Gi",
+            worker_memory_requests="3",
+            worker_memory_limits="4",
+            image="rayproject/ray:2.47.1",
             write_to_file=True,
             verify_tls=False,
         )
@@ -135,14 +138,15 @@ def test_apply_invalid_update(self):
             name=cluster_name,
             namespace=namespace,
             num_workers=2,
-            head_cpu_requests="1",
-            head_cpu_limits="2",  # Changing CPU limits (immutable)
-            head_memory_requests="1Gi",
-            head_memory_limits="2Gi",
+            head_cpu_requests="500m",
+            head_cpu_limits="1",
+            head_memory_requests="5",
+            head_memory_limits="6",
             worker_cpu_requests="500m",
             worker_cpu_limits="1",
-            worker_memory_requests="1Gi",
-            worker_memory_limits="2Gi",
+            worker_memory_requests="1",
+            worker_memory_limits="2",
+            image="rayproject/ray:2.47.1",
             write_to_file=True,
             verify_tls=False,
         )

tests/e2e/rayjob/ray_version_validation_oauth_test.py

@@ -11,10 +11,7 @@
     ManagedClusterConfig,
 )
 
-# This test validates Ray version compatibility checking for RayJob with cluster lifecycling scenarios
 
-
-@pytest.mark.openshift
 class TestRayJobRayVersionValidationOauth:
     def setup_method(self):
         initialize_kubernetes_client(self)

tests/test_cluster_yamls/appwrapper/unit-test-all-params.yaml

@@ -39,7 +39,6 @@ spec:
           rayStartParams:
             block: 'true'
             dashboard-host: 0.0.0.0
-            dashboard-port: '8265'
             num-gpus: '1'
             resources: '"{\"TPU\": 2}"'
           serviceType: ClusterIP

tests/test_cluster_yamls/kueue/aw_kueue.yaml

@@ -33,7 +33,6 @@ spec:
           rayStartParams:
             block: 'true'
             dashboard-host: 0.0.0.0
-            dashboard-port: '8265'
             num-gpus: '0'
             resources: '"{}"'
           serviceType: ClusterIP