OCP - Baremetal

• Get IPMI status using iDrac/IPMI interface

for i in {2..10}; do ipmitool -I lanplus -H 10.5.35.$i -U root -P p4ssW0rD power status; done

• Get/Debug Ironic status os Master nodes using dev-scripts

cd ${HOME}/repos/dev-scripts
export OS_CLOUD=metal3-bootstrap
openstack baremetal node list
openstack baremetal node show openshift-master-0

• Use Racadm to get info form Baremetal Chasis using iDrac/IPMI interface

podman pull justinclayton/racadm
alias racadm='podman run --rm -it -v "/root":/shared justinclayton/racadm'
racadm -r "10.5.35.2" -u root -p "p4ssW0rD" getsysinfo

• Or you could use ssh against chassis to do that

alias racadm="sshpass -e ssh -oStrictHostKeyChecking=no root@$10.5.35.2 racadm"
racadm getsysinfo

• Follow an OCP installation with watch

watch "tail -n 4 clusterconfigs/.openshift_install.log; oc get po -A -o wide | grep -v -E 'Running|Complete';oc get bmh -A; oc get machines -A; oc get nodes"

• Deleting "Bad" BareMetal Nodes

  1. Create bmh resource and wait until it's Ready
  2. Scale up machineset to start deployment
  3. Do not Delete bmh
  4. Annotate "bad" or failed nodes with machine.openshift.io/cluster-api-delete-machine=yes
  5. Scale down replicas to match the number of existing worker nodes
  6. This will automatically delete "bad" nodes

• Get the Kubeconfig from a Hive secret and put in a file

oc get secret -n hive mgmt-spoke1-0-s55zb-admin-kubeconfig -o jsonpath='{.data.kubeconfig}' | base64 -d > ~/repos/hive/spoke1/kubeconfig_spoke1

• Get the Kubeadmin password from a Hive secret and put it in a file

oc get secret mgmt-spoke1-0-s55zb-admin-password -o jsonpath='{.data.password}' | base64 -d > ~/repos/hive/spoke1/kube