Add inter-pod anti-affinity rule in master node spec to schedule node…

…s on different cluster GPU/CPU nodes
opendatahub-io · Dec 20, 2024 · a325059 · a325059
1 parent 9cc9a6f
commit a325059
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 2 deletions.
diff --git a/tests/kfto/kfto_mnist_training_test.go b/tests/kfto/kfto_mnist_training_test.go
@@ -104,7 +104,26 @@ func createKFTOPyTorchMnistJob(test Test, namespace string, config corev1.Config
 					Replicas:      Ptr(int32(1)),
 					RestartPolicy: kftov1.RestartPolicyOnFailure,
 					Template: corev1.PodTemplateSpec{
+						ObjectMeta: metav1.ObjectMeta{
+							Labels: map[string]string{
+								"app": "kfto-mnist",
+							},
+						},
 						Spec: corev1.PodSpec{
+							Affinity: &corev1.Affinity{
+								PodAntiAffinity: &corev1.PodAntiAffinity{
+									RequiredDuringSchedulingIgnoredDuringExecution: []corev1.PodAffinityTerm{
+										{
+											LabelSelector: &metav1.LabelSelector{
+												MatchLabels: map[string]string{
+													"app": "kfto-mnist",
+												},
+											},
+											TopologyKey: "kubernetes.io/hostname",
+										},
+									},
+								},
+							},
 							Containers: []corev1.Container{
 								{
 									Name:            "pytorch",

diff --git a/tests/kfto/resources/mnist.py b/tests/kfto/resources/mnist.py
@@ -1,5 +1,3 @@
-from __future__ import print_function
-
 import argparse
 import os