Jivvon

echo -n "Really want to reset this node? (y/N): "
read -r check

if [[ "$check" = "y" ]]; then
  echo "[INFO] Reset kubeadm"
  kubeadm reset -f
  echo "[INFO] Stop kubelet & docker service"
  systemctl stop kubelet docker docker.socket
  echo "[INFO] Remove K8S files"
  rm -rvf /etc/kubernetes/*

Jivvon

prometheus:
  extraScrapeConfigs: |
    - job_name: gpu-metrics
      kubernetes_sd_configs:
      - role: endpoints
      relabel_configs:
      - action: keep
        regex: dcgm-exporter;monitoring
        source_labels:
        - __meta_kubernetes_pod_label_app_kubernetes_io_name

Jivvon

# using velero
# backup k8s resources per namespace without pv snapshot
# required: velero server & client are installed

NAMESPACES=`kubectl get ns | grep -v NAME | awk '{ print $1 }'`
EXCEPT_NAMESPACES=(velero)
BUCKET_LOCATION=default

for NAMESPACE in $NAMESPACES; do
  skip_flag=false

Jivvon

apiVersion: batch/v1beta1
kind: CronJob
metadata:
  name: {{ include "k8s-etcd-backup.fullname" . }}
  labels:
    {{- include "k8s-etcd-backup.labels" . | nindent 4 }}
spec:
  concurrencyPolicy: Allow
  failedJobsHistoryLimit: 1
  jobTemplate: