|
| 1 | +# OVERRIDE values for HyperPodHelmChart for RIG. |
| 2 | +# This is a YAML-formatted file. |
| 3 | +# Declare variables to be passed into your templates. |
| 4 | + |
| 5 | +# value_safe_name of dependencies - see install_rig_dependencies.sh |
| 6 | +# |
| 7 | +# Note: |
| 8 | +# |
| 9 | +# The format is add-on: {keys:...} . |
| 10 | +# Helm wil AUTOMATICALLY SCOPE the add-on values when resolving each dependency |
| 11 | +# In other words, what is exposed to each dependency template is |
| 12 | +# {{ .Values.tolerations }} , NOT {{ .Values.add-on.tolerations }} |
| 13 | + |
| 14 | + |
| 15 | +# This is a special case. We will not use values.yaml to override the values for this add-on |
| 16 | +# It will still be rendered by Helm using special logic. |
| 17 | +# See install_rig_dependncies.yaml |
| 18 | +#aws-cni: |
| 19 | + |
| 20 | + |
| 21 | +coredns: |
| 22 | + tolerations: |
| 23 | + - effect: NoSchedule |
| 24 | + key: node-role.kubernetes.io/control-plane |
| 25 | + - key: CriticalAddonsOnly |
| 26 | + operator: Exists |
| 27 | + - effect: NoSchedule |
| 28 | + operator: Equal |
| 29 | + key: "sagemaker.amazonaws.com/RestrictedNode" |
| 30 | + value: "Worker" |
| 31 | + nodeSelector: |
| 32 | + "sagemaker.amazonaws.com/instance-group-type": "Restricted" |
| 33 | + |
| 34 | +mpi-operator: |
| 35 | + tolerations: |
| 36 | + - key: "sagemaker.amazonaws.com/node-health-status" |
| 37 | + operator: "Equal" |
| 38 | + value: "Unschedulable" |
| 39 | + effect: "NoSchedule" |
| 40 | + - effect: NoSchedule |
| 41 | + operator: Equal |
| 42 | + key: "sagemaker.amazonaws.com/RestrictedNode" |
| 43 | + value: "Worker" |
| 44 | + nodeSelector: |
| 45 | + "sagemaker.amazonaws.com/instance-group-type": "Restricted" |
| 46 | + |
| 47 | +neuron-device-plugin: |
| 48 | + tolerations: |
| 49 | + - key: CriticalAddonsOnly |
| 50 | + operator: Exists |
| 51 | + - key: "aws.amazon.com/neuron" |
| 52 | + operator: Exists |
| 53 | + effect: NoSchedule |
| 54 | + - key: "sagemaker.amazonaws.com/node-health-status" |
| 55 | + operator: Equal |
| 56 | + value: Unschedulable |
| 57 | + effect: NoSchedule |
| 58 | + - effect: NoSchedule |
| 59 | + operator: Equal |
| 60 | + key: "sagemaker.amazonaws.com/RestrictedNode" |
| 61 | + value: "Worker" |
| 62 | + nodeSelector: |
| 63 | + "sagemaker.amazonaws.com/instance-group-type": "Restricted" |
| 64 | + |
| 65 | +training-operators: |
| 66 | + tolerations: |
| 67 | + - effect: NoSchedule |
| 68 | + operator: Equal |
| 69 | + key: "sagemaker.amazonaws.com/RestrictedNode" |
| 70 | + value: "Worker" |
| 71 | + nodeSelector: |
| 72 | + "sagemaker.amazonaws.com/instance-group-type": "Restricted" |
| 73 | + |
| 74 | +# Overrides for externally-maintained Helm charts |
| 75 | +aws-efa-k8s-device-plugin: |
| 76 | + devicePlugin: |
| 77 | + enabled: true |
| 78 | + supportedInstanceLabels: |
| 79 | + values: |
| 80 | + - ml.c5n.9xlarge |
| 81 | + - ml.c5n.18xlarge |
| 82 | + - ml.g5.8xlarge |
| 83 | + - ml.g5.12xlarge |
| 84 | + - ml.g5.16xlarge |
| 85 | + - ml.g5.24xlarge |
| 86 | + - ml.g5.48xlarge |
| 87 | + - ml.g6.8xlarge |
| 88 | + - ml.g6.12xlarge |
| 89 | + - ml.g6.16xlarge |
| 90 | + - ml.g6.24xlarge |
| 91 | + - ml.g6.48xlarge |
| 92 | + - ml.g6e.8xlarge |
| 93 | + - ml.g6e.12xlarge |
| 94 | + - ml.g6e.16xlarge |
| 95 | + - ml.g6e.24xlarge |
| 96 | + - ml.g6e.48xlarge |
| 97 | + - ml.gr6.8xlarge |
| 98 | + - ml.i3en.large |
| 99 | + - ml.i3en.xlarge |
| 100 | + - ml.i3en.2xlarge |
| 101 | + - ml.i3en.3xlarge |
| 102 | + - ml.i3en.6xlarge |
| 103 | + - ml.i3en.12xlarge |
| 104 | + - ml.i3en.24xlarge |
| 105 | + - ml.m7i.large |
| 106 | + - ml.m7i.xlarge |
| 107 | + - ml.m7i.2xlarge |
| 108 | + - ml.m7i.4xlarge |
| 109 | + - ml.m7i.8xlarge |
| 110 | + - ml.m7i.12xlarge |
| 111 | + - ml.m7i.16xlarge |
| 112 | + - ml.m7i.24xlarge |
| 113 | + - ml.m7i.48xlarge |
| 114 | + - ml.p4d.24xlarge |
| 115 | + - ml.p4de.24xlarge |
| 116 | + - ml.p5.48xlarge |
| 117 | + - ml.p5e.48xlarge |
| 118 | + - ml.p5en.48xlarge |
| 119 | + - ml.r7i.large |
| 120 | + - ml.r7i.xlarge |
| 121 | + - ml.r7i.2xlarge |
| 122 | + - ml.r7i.4xlarge |
| 123 | + - ml.r7i.8xlarge |
| 124 | + - ml.r7i.12xlarge |
| 125 | + - ml.r7i.16xlarge |
| 126 | + - ml.r7i.24xlarge |
| 127 | + - ml.r7i.48xlarge |
| 128 | + - ml.trn1.32xlarge |
| 129 | + - ml.trn1n.32xlarge |
| 130 | + - ml.trn2.48xlarge |
| 131 | + tolerations: |
| 132 | + - key: CriticalAddonsOnly |
| 133 | + operator: Exists |
| 134 | + - effect: NoSchedule |
| 135 | + key: aws.amazon.com/efa |
| 136 | + operator: Exists |
| 137 | + - key: sagemaker.amazonaws.com/node-health-status |
| 138 | + operator: "Equal" |
| 139 | + value: "Unschedulable" |
| 140 | + effect: "NoSchedule" |
| 141 | + - effect: NoSchedule |
| 142 | + operator: Equal |
| 143 | + key: "sagemaker.amazonaws.com/RestrictedNode" |
| 144 | + value: "Worker" |
| 145 | + nodeSelector: |
| 146 | + "sagemaker.amazonaws.com/instance-group-type": "Restricted" |
| 147 | + |
| 148 | +nvidia-device-plugin: |
| 149 | + devicePlugin: |
| 150 | + enabled: true |
| 151 | + allowDefaultNamespace: true |
| 152 | + namespaceOverride: "kube-system" |
| 153 | + affinity: |
| 154 | + nodeAffinity: |
| 155 | + requiredDuringSchedulingIgnoredDuringExecution: |
| 156 | + nodeSelectorTerms: |
| 157 | + - matchExpressions: |
| 158 | + - key: node.kubernetes.io/instance-type |
| 159 | + operator: In |
| 160 | + values: |
| 161 | + - ml.g5.xlarge |
| 162 | + - ml.g5.2xlarge |
| 163 | + - ml.g5.4xlarge |
| 164 | + - ml.g5.8xlarge |
| 165 | + - ml.g5.12xlarge |
| 166 | + - ml.g5.16xlarge |
| 167 | + - ml.g5.24xlarge |
| 168 | + - ml.g5.48xlarge |
| 169 | + - ml.g6.xlarge |
| 170 | + - ml.g6.2xlarge |
| 171 | + - ml.g6.4xlarge |
| 172 | + - ml.g6.8xlarge |
| 173 | + - ml.g6.16xlarge |
| 174 | + - ml.g6.12xlarge |
| 175 | + - ml.g6.24xlarge |
| 176 | + - ml.g6.48xlarge |
| 177 | + - ml.g6e.xlarge |
| 178 | + - ml.g6e.2xlarge |
| 179 | + - ml.g6e.4xlarge |
| 180 | + - ml.g6e.8xlarge |
| 181 | + - ml.g6e.12xlarge |
| 182 | + - ml.g6e.16xlarge |
| 183 | + - ml.g6e.24xlarge |
| 184 | + - ml.g6e.48xlarge |
| 185 | + - ml.gr6.4xlarge |
| 186 | + - ml.gr6.8xlarge |
| 187 | + - ml.p4d.24xlarge |
| 188 | + - ml.p4de.24xlarge |
| 189 | + - ml.p5.48xlarge |
| 190 | + - ml.p5e.48xlarge |
| 191 | + - ml.p5en.48xlarge |
| 192 | + tolerations: |
| 193 | + - key: nvidia.com/gpu |
| 194 | + operator: Exists |
| 195 | + effect: NoSchedule |
| 196 | + - key: sagemaker.amazonaws.com/node-health-status |
| 197 | + operator: Equal |
| 198 | + value: Unschedulable |
| 199 | + effect: NoSchedule |
| 200 | + - effect: NoSchedule |
| 201 | + operator: Equal |
| 202 | + key: "sagemaker.amazonaws.com/RestrictedNode" |
| 203 | + value: "Worker" |
| 204 | + nodeSelector: |
| 205 | + "sagemaker.amazonaws.com/instance-group-type": "Restricted" |
0 commit comments