gpudirect-tcpx/optmem-max-ds.yaml

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: optmem-max-ds
  namespace: kube-system
  labels:
    k8s-app: optmem-max-ds
spec:
  selector:
    matchLabels:
      k8s-app: optmem-max-ds
  template:
    metadata:
      labels:
        name: optmem-max-ds
        k8s-app: optmem-max-ds
    spec:
      affinity:
        nodeAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
              - matchExpressions:
                  - key: cloud.google.com/gke-accelerator
                    operator: In
                    values:
                      - nvidia-h100-80gb
      hostNetwork: true
      tolerations:
        - operator: "Exists"
      initContainers:
        - name: optmem-max
          image: gke.gcr.io/gke-distroless/bash
          securityContext:
            privileged: true
          command:
            - /bin/bash
            - -c
            - "set -x; cat /proc/sys/net/core/optmem_max; echo 131072 > /proc/sys/net/core/optmem_max"
      containers:
        - name: pause
          image: registry.k8s.io/pause:3.9
          securityContext:
            runAsUser: 2023
            runAsGroup: 2023
            allowPrivilegeEscalation: false
            capabilities:
              drop: ["all"]
            seccompProfile:
              type: RuntimeDefault
            readOnlyRootFilesystem: true