定时备份Kubernetes集群常用对象到Yaml文件

有时候我们经常需要用到kubernetes中常用对象的yaml文件, 搞了个简单的脚本来定时备份集群中常用的资源对象保存到yaml文件，以防万一, 毕竟机器的运维权限在别人手上, 指不定什么时候就出问题.

脚本保存的常用对象如下, 需求比较明确, 脚本比较简单，也没啥好说的, 有些为了去掉yaml中不必要的字段使用了正则匹配, 又臭又长.

• Secrets
• Config Maps
• Deployments
• Services
• Ingress
• Persistent Volumes
• Cronjobs

shell脚本如下:

cat kubernetes-object-toyaml.backup

#!/bin/bash
###########
# Global Configurations
#======================
BACKUP_DIR=/usr/local/src/kubernetes-objectyaml-backup
BACKUP_ENCS=$BACKUP_DIR/backup_encs
AWS_CMD=/usr/bin/aws
S3_BUCKET=your-s3-bucket
AWS_ACCESS_KEY_ID=your-aws-access-key
AWS_SECRET_ACCESS_KEY=your-aws-secret-key 
TIME_STAMP=$(date +%Y-%m-%d_%H-%M)
CLUSTER_NAME=your-cluster
#tar secret
KUBE_ARCHIVE_PW=your-tar-password
######################
function get_secret {
  kubectl get secret -n ${1} -o=yaml --export --field-selector type!=kubernetes.io/service-account-token | sed -e '/resourceVersion: "[0-9]\+"/d' -e '/uid: [a-z0-9-]\+/d' -e '/selfLink: [a-z0-9A-Z/]\+/d' -e '/creationTimestamp: "[0-9]\+-[0-9]\+-[0-9]\+T.\+"/d' -e '/kubectl.kubernetes.io\/last-applied-configuration/d' -e '/{"apiVersion":".*}/d'
}

function get_configmap {
  kubectl get configmap -n ${1} -o=yaml --export | sed -e '/resourceVersion: "[0-9]\+"/d' -e '/uid: [a-z0-9-]\+/d' -e '/selfLink: [a-z0-9A-Z/]\+/d' -e '/creationTimestamp: "[0-9]\+-[0-9]\+-[0-9]\+T.\+"/d' -e '/kubectl.kubernetes.io\/last-applied-configuration/d' -e '/{"apiVersion":".*}/d'
}

function get_ingress {
  kubectl get ing -n ${1} -o=yaml --export | sed -e '/status:/,+2d' -e '/\- ip: \([0-9]\{1,3\}\.\)\{3\}[0-9]\{1,3\}/d' -e '/resourceVersion: "[0-9]\+"/d' -e '/uid: [a-z0-9-]\+/d' -e '/selfLink: [a-z0-9A-Z/]\+/d' -e '/creationTimestamp: "[0-9]\+-[0-9]\+-[0-9]\+T.\+"/d' -e '/kubectl.kubernetes.io\/last-applied-configuration/d' -e '/{"apiVersion":".*}/d' -e '/generation: [0-9]\+/d' -e '/field.cattle.io\/publicEndpoints/d' -e '/field.cattle.io\/ingressState/d'
}

function get_service {
  kubectl get service -n ${1} -o=yaml --export | sed -e '/ownerReferences:/,+5d' -e '/resourceVersion: "[0-9]\+"/d' -e '/uid: [a-z0-9-]\+/d' -e '/selfLink: [a-z0-9A-Z/]\+/d' -e '/clusterIP: \([0-9]\{1,3\}\.\)\{3\}[0-9]\{1,3\}/d' -e '/creationTimestamp: "[0-9]\+-[0-9]\+-[0-9]\+T.\+"/d' -e '/kubectl.kubernetes.io\/last-applied-configuration/d' -e '/{"apiVersion":".*}/d'
}

function get_deployment {
  kubectl get deployment -n ${1} -o=yaml --export | sed -e '/deployment\.kubernetes\.io\/revision: "[0-9]\+"/d' -e '/resourceVersion: "[0-9]\+"/d' -e '/uid: [a-z0-9-]\+/d' -e '/selfLink: [a-z0-9A-Z/]\+/d' -e '/status:/,+18d' -e '/creationTimestamp: "[0-9]\+-[0-9]\+-[0-9]\+T.\+"/d' -e '/kubectl.kubernetes.io\/last-applied-configuration/d' -e '/{"apiVersion":".*}/d' -e '/generation: [0-9]\+/d' -e '/field.cattle.io\/publicEndpoints/d'
}

function get_cronjob {
  kubectl get cronjob -n ${1} -o=yaml --export | sed -e '/status:/,+1d' -e '/resourceVersion: "[0-9]\+"/d' -e '/uid: [a-z0-9-]\+/d' -e '/selfLink: [a-z0-9A-Z/]\+/d' -e '/creationTimestamp: "[0-9]\+-[0-9]\+-[0-9]\+T.\+"/d' -e '/kubectl.kubernetes.io\/last-applied-configuration/d' -e '/{"apiVersion":".*}/d'
}

function get_pvc {
  kubectl get pvc -n ${1} -o=yaml --export | sed -e '/control\-plane\.alpha\.kubernetes\.io\/leader\:/d' -e '/resourceVersion: "[0-9]\+"/d' -e '/uid: [a-z0-9-]\+/d' -e '/selfLink: [a-z0-9A-Z/]\+/d' -e '/creationTimestamp: "[0-9]\+-[0-9]\+-[0-9]\+T.\+"/d' -e '/kubectl.kubernetes.io\/last-applied-configuration/d' -e '/{"apiVersion":".*}/d'
}

function get_pv {
  for pvolume in `kubectl get pvc -n ${1} -o=custom-columns=:.spec.volumeName` 
  do
     kubectl get pv -o=yaml --export --field-selector metadata.name=${pvolume} | sed -e '/resourceVersion: "[0-9]\+"/d' -e '/uid: [a-z0-9-]\+/d' -e '/selfLink: [a-z0-9A-Z/]\+/d' -e '/creationTimestamp: "[0-9]\+-[0-9]\+-[0-9]\+T.\+"/d' -e '/kubectl.kubernetes.io\/last-applied-configuration/d' -e '/{"apiVersion":".*}/d'
  done
}

function export_ns {
  mkdir -p ${BACKUP_DIR}/${CLUSTER_NAME}/
  cd ${BACKUP_DIR}/${CLUSTER_NAME}/
  for namespace in `kubectl get namespaces --no-headers=true | awk '{ print $1 }' | grep -v -e "cattle-prometheus" -e "cattle-system" -e "kube-system" -e "kube-public"`
  do
     echo "Namespace: $namespace"
     echo "+++++++++++++++++++++++++"
     mkdir -p $namespace

     for object_kind in configmap ingress service secret deployment cronjob pvc
     do
       if kubectl get ${object_kind} -n ${namespace} 2>&1 | grep "No resources" > /dev/null; then
         echo "No resources found for ${object_kind} in ${namespace}"
       else
         get_${object_kind} ${namespace} > ${namespace}/${object_kind}.${namespace}.yaml &&  echo "${object_kind}.${namespace}";
         
         if [ ${object_kind} = "pvc" ]; then
           get_pv ${namespace} > ${namespace}/pv.${namespace}.yaml &&  echo "pv.${namespace}";
         fi
       fi
     done
     echo "+++++++++++++++++++++++++"
  done
}

###########################################################
## Archiving k8s data with password to upload it to AWS S3.
## This password is available on our password manager.
############################################################
function archive_ns {
  cd ${BACKUP_DIR}
  mkdir -p ${BACKUP_ENCS}
  # tar with openssl
  tar cz ${CLUSTER_NAME} | openssl enc -aes-256-cbc -e -k ${KUBE_ARCHIVE_PW} > ${BACKUP_ENCS}/${CLUSTER_NAME}-${TIME_STAMP}.tar.gz.enc
  # untar with openssl
  # https://www.howtoing.com/encrypt-decrypt-files-tar-openssl-linux/
  # openssl enc -d -aes256-cbc -k ${KUBE_ARCHIVE_PW} -in xxx.ooo.tar.gz.enc | tar xz -C test
}

# Upload Backups
#===============
function upload_backup_to_s3 {
  ${AWS_CMD} s3 cp ${BACKUP_DIR}/${CLUSTER_NAME}-${TIME_STAMP}.tar.gz.enc s3://${S3_BUCKET}/${CLUSTER_NAME}/
  if [ $? -eq 0 ]; then
    echo "${CLUSTER_NAME}-${TIME_STAMP}.tar.gz.enc is successfully uploaded"
    rm -rf ${BACKUP_DIR}/${CLUSTER_NAME} ${BACKUP_DIR}/k8s-data-${TIME_STAMP}.tar.gz.enc
  else
    echo "${CLUSTER_NAME}-${TIME_STAMP}.tar.gz.enc failed to be uploaded"
    exit 1
  fi
}

# Clean 7 days ago
function clean_7days_ago {
  find ${BACKUP_ENCS} -mtime +7 -name "*.tar.gz.enc" -exec /bin/rm -rf {} \;
}


###########
export_ns
archive_ns
upload_backup_to_s3
clean_7days_ago

在另一master节点下执行:

bash kubernetes-object-toyaml.backup

最后加入系统crontab即可

11 22 * * * bash /your/dir/kubernetes-object-toyaml.backup

再或者加个rsync同步到其它节点上，双保险.

参考文章:

转载请注明出处https://izsk.me