Skip to content

Dell RHEL SOS Reports for OpenShift

1. Mirror sos-report image [1]

  1. From a connected machine:

    podman pull registry.redhat.io/rhel9/support-tools:latest
    podman images | grep support-tools
    
    podman save -o rhel9-support-tools.tar registry.redhat.io/rhel9/support-tools:latest
    
    tar tf rhel9-support-tools.tar | head
    

  2. Copy over to Disconnected machine.

  3. Retag and push to artifactory:

    # load
    podman load -i rhel9-support-tools.tar
    
    # retag to Artifactory
    podman tag \
      registry.redhat.io/rhel9/support-tools:latest \
      artifactory.local/link-to-repo/rhel9-support-tools:latest
    
    # push
    podman push artifactory.local/link-to-repo/rhel9-support-tools:latest
    

2. Generating SOS Report

  1. Create toolboxrc:
    vi /root/.toolboxrc
    
    REGISTRY=[custom-private-registry.example.com:5000]
    IMAGE=rhel9/support-tools
    
  2. Spawn toolbox container:
    toolbox
    
sos report -e openshift -e openshift_ovn -e openvswitch -e podman -e crio -k crio.all=on -k crio.logs=on -k podman.all=on -k podman.logs=on -k networking.ethtool-namespaces=off --all-logs --plugin-timeout=600

cat /host/var/tmp/sosreport-XXXXX.tar.xz > /tmp/sosreport-XXXXX.tar.xz
  1. SCP it out:
    scp core@<nodename>:/var/tmp/sosreport-XXXXX.tar.xz /tmp/sosreport-XXXXX.tar.xz
    

3. rocm-smi and rocminfo

apiVersion: v1
kind: Pod
metadata:
  name: rocm-smi-test
  namespace: default
spec:
  containers:
  - name: rocm-test
    image: rocm/dev-ubuntu-22.04:latest
    command: ["/bin/bash", "-c", "--"]
    args: ["while true; do sleep 3600; done;"]
    resources:
      limits:
        amd.com/gpu: 1
    securityContext:
      privileged: true
  restartPolicy: Never
apiVersion: v1
kind: Pod
metadata:
  name: rocminfo
spec:
  restartPolicy: Never
  containers:
  - name: test
    image: docker.io/rocm/pytorch:latest
    command: ["/bin/sh","-c"]
    args: ["rocminfo || /usr/bin/rocminfo || true"]
    resources:
      requests: { amd.com/gpu: 1 }
      limits:   { amd.com/gpu: 1 }