Use besteffort minValuePolicy in Helm and add minValues support for training and titan pools (#574)

Author: chithreshazad

tests/assets/karpenter/ai-ml-training-large.yaml

@@ -40,13 +40,15 @@ spec:
         values:
         - linux
       - key: karpenter.k8s.aws/instance-size
+        minValues: 4
         operator: In
         values:
         - large
         - xlarge
         - 2xlarge
         - 4xlarge
       - key: karpenter.k8s.aws/instance-family
+        minValues: 33
         operator: In
         values:
         - c5

tests/assets/karpenter/titan-pool.yaml

@@ -34,13 +34,15 @@ spec:
         values:
         - on-demand
       - key: karpenter.k8s.aws/instance-size
+        minValues: 4
         operator: In
         values:
         - large
         - xlarge
         - 2xlarge
         - 4xlarge
       - key: karpenter.k8s.aws/instance-family
+        minValues: 33
         operator: In
         values:
         - c5

tests/tekton-resources/tasks/setup/karpenter/helm-karpenter-install.yaml

@@ -92,6 +92,7 @@ spec:
         --set "tolerations[0].value=karpenter" \
         --set "tolerations[0].operator=Equal" \
         --set "dnsPolicy=Default" \
+        --set "settings.minValuesPolicy=BestEffort" \
         --set-string "controller.env[0].name=AWS_ENDPOINT_URL_EKS" \
         --set-string "controller.env[0].value=$(params.endpoint)" \
         --set-string "controller.env[1].name=KUBE_CLIENT_QPS" \