kubelet启动报错：kubelet.service: Failed with result 'exit

最新推荐文章于 2025-05-13 20:28:19 发布

那记忆微凉

最新推荐文章于 2025-05-13 20:28:19 发布

阅读量2.3w

点赞数 2

CC 4.0 BY-SA版权

分类专栏： Kubernetes

本文链接：https://blog.youkuaiyun.com/BigData_Mining/article/details/87905804

Kubernetes 专栏收录该内容

62 篇文章

订阅专栏

本文记录了在Kubernetes集群中，kubelet服务启动失败的问题排查与解决过程。问题出现于CgroupDriver配置不一致，kubelet与docker的CgroupDriver默认分别为cgroupfs与systemd。通过调整kubelet的CgroupDriver为cgroupfs，成功解决了启动失败的问题。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

qyh@qyh-master:/home/kubernetes$ systemctl status kubelet
● kubelet.service - Kubernetes Kubelet
   Loaded: loaded (/lib/systemd/system/kubelet.service; enabled; vendor preset: 
   Active: failed (Result: exit-code) since Sun 2019-02-24 17:34:38 CST; 8s ago
  Process: 13392 ExecStart=/opt/kubernetes/bin/kubelet ${KUBE_LOGTOSTDERR} ${KUB
 Main PID: 13392 (code=exited, status=255)

2月 24 17:34:38 qyh-master kubelet[13392]:       --vmodule moduleSpec           
2月 24 17:34:38 qyh-master kubelet[13392]:       --volume-plugin-dir string     
2月 24 17:34:38 qyh-master kubelet[13392]:       --volume-stats-agg-period durat
2月 24 17:34:38 qyh-master kubelet[13392]: F0224 17:34:38.463222   13392 server.
2月 24 17:34:38 qyh-master systemd[1]: kubelet.service: Service hold-off time ov
2月 24 17:34:38 qyh-master systemd[1]: kubelet.service: Scheduled restart job, r
2月 24 17:34:38 qyh-master systemd[1]: Stopped Kubernetes Kubelet.
2月 24 17:34:38 qyh-master systemd[1]: kubelet.service: Start request repeated t
2月 24 17:34:38 qyh-master systemd[1]: kubelet.service: Failed with result 'exit
2月 24 17:34:38 qyh-master systemd[1]: Failed to start Kubernetes Kubelet.

------------------------------------------------------------------------------------------------------------------

查看docker和kublelt的Cgroup Driver是否一致。这个是不一致造成的。改成一致。可以看出报错的原因是kubelet与docker的cgroup driver 不一致导致

qyh@qyh-master:/home/kubernetes$ sudo docker info|grep Cgroup
Cgroup Driver: cgroupfs

----------------------------------------------------------------------------------------------------------------------

qyh@qyh-master:/home/kubernetes$ journalctl -f -u kubelet
-- Logs begin at Sun 2019-02-24 15:21:08 CST. --
2月 24 16:53:25 qyh-master kubelet[7850]:       --vmodule moduleSpec                                                                                        comma-separated list of pattern=N settings for file-filtered logging
2月 24 16:53:25 qyh-master kubelet[7850]:       --volume-plugin-dir string                                                                                  The full path of the directory in which to search for additional third party volume plugins (default "/usr/libexec/kubernetes/kubelet-plugins/volume/exec/")
2月 24 16:53:25 qyh-master kubelet[7850]:       --volume-stats-agg-period duration                                                                          Specifies interval for kubelet to calculate and cache the volume disk usage for all pods and volumes.  To disable volume calculations, set to 0. (default 1m0s) (DEPRECATED: This parameter should be set via the config file specified by the Kubelet's --config flag. See https://kubernetes.io/docs/tasks/administer-cluster/kubelet-config-file/ for more information.)
2月 24 16:53:25 qyh-master kubelet[7850]: F0224 16:53:25.446480    7850 server.go:154] unknown command:
2月 24 16:53:25 qyh-master systemd[1]: kubelet.service: Service hold-off time over, scheduling restart.
2月 24 16:53:25 qyh-master systemd[1]: kubelet.service: Scheduled restart job, restart counter is at 5.
2月 24 16:53:25 qyh-master systemd[1]: Stopped Kubernetes Kubelet.
2月 24 16:53:25 qyh-master systemd[1]: kubelet.service: Start request repeated too quickly.
2月 24 16:53:25 qyh-master systemd[1]: kubelet.service: Failed with result 'exit-code'.
2月 24 16:53:25 qyh-master systemd[1]: Failed to start Kubernetes Kubelet.

--------------------------------------------------------

docker相比1.10增加了KernelMemory变量和CgroupDriver变量，KernelMemory变量表示是否设置linux内核内存限制，CgroupDriver变量表示使用哪个Cgroup驱动，有两种驱动，分别是cgroupfs和systemd，默认使用cgroupfs。那么解决该问题的方法是：要么修改docker 的cgroup driver ，要么改kubelet的 cgroup driver.。建议改kubelet的值，因为docker本身有些镜像在使用的话，直接改会破坏镜像

--------------------------------------------------------------------

直接更改kubelet.sh

# 其他参数
KUBELET_ARGS="--cgroup-driver=cgroupfs"
EOF

KUBE_PROXY_OPTS=" \${KUBE_LOGTOSTDERR}     \\
 \${KUBE_LOG_LEVEL}       \\
 \${NODE_ADDRESS}         \\
 \${NODE_PORT}            \\
 \${NODE_HOSTNAME}        \\
 \${KUBELET_KUBECONFIG}   \\
 \${KUBE_ALLOW_PRIV}      \\
 \${KUBELET_DNS_IP}       \\
 \${KUBELET_DNS_DOMAIN}   \\
 \${KUBELET_ARGS}"