两个常见的Kubernetes问题以及解决方法

applying cgroup … caused: mkdir …no space left on device
或者在describe pod的时候出现cannot allocate memory

cat /sys/fs/cgroup/memory/kubepods/memory.kmem.slabinfo
当出现cat: /sys/fs/cgroup/memory/kubepods/memory.kmem.slabinfo: Input/output error则说明不存在内存泄露的情况
如果存在内存泄露会出现
slabinfo - version: 2.1
# name            <active_objs> <num_objs> <objsize> <objperslab> <pagesperslab> : tunables <limit> <batchcount> <sharedfactor> : slabdata <active_slabs> <num_slabs> <sharedavail>

wget https://dl.google.com/go/go1.12.9.linux-amd64.tar.gz
tar xf go1.12.9.linux-amd64.tar.gz -C /usr/local/

写入bashrc
vim ~/.bashrc 
export GOPATH="/data/Documents"
export GOROOT="/usr/local/go"
export PATH="$GOROOT/bin:$GOPATH/bin:$PATH"
export GO111MODULE=off

验证
source ~/.bashrc 
go env

mkdir -p /data/Documents/src/github.com/opencontainers/
cd /data/Documents/src/github.com/opencontainers/
git clone https://github.com/opencontainers/runc
cd runc/
git checkout v1.0.0-rc9  # 切到v1.0.0-rc9 tag

安装编译组件
sudo yum install libseccomp-devel
make BUILDTAGS='seccomp nokmem'
编译完成之后会在当前目录下看到一个runc的可执行文件，等kubelet编译完成之后会将其替换

mkdir -p /root/k8s/
cd /root/k8s/
git clone https://github.com/kubernetes/kubernetes
cd kubernetes/
git checkout v1.15.3

FROM centos:centos7.3.1611

ENV GOROOT /usr/local/go
ENV GOPATH /usr/local/gopath
ENV PATH /usr/local/go/bin:$PATH

RUN yum install rpm-build which where rsync gcc gcc-c++ automake autoconf libtool make -y 
    && curl -L https://studygolang.com/dl/golang/go1.12.9.linux-amd64.tar.gz | tar zxvf - -C /usr/local

docker run  -it --rm   -v /root/k8s/kubernetes:/usr/local/gopath/src/k8s.io/kubernetes   build-k8s:centos-7.3-go-1.12.9-k8s-1.15.3   bash
cd /usr/local/gopath/src/k8s.io/kubernetes
#编译
GO111MODULE=off KUBE_GIT_TREE_STATE=clean KUBE_GIT_VERSION=v1.15.3 make kubelet GOFLAGS="-tags=nokmem"

mv /usr/bin/kubelet /home/kubelet
mv /usr/bin/docker-runc /home/docker-runc

systemctl stop docker
systemctl stop kubelet

cp kubelet /usr/bin/kubelet
cp kubelet /usr/local/bin/kubelet
cp runc /usr/bin/docker-runc

cat /sys/fs/cgroup/memory/kubepods/burstable/memory.kmem.usage_in_bytes

cat /sys/fs/cgroup/memory/kubepods/memory.kmem.slabinfo

Unable to connect to the server: x509: certificate has expired or is not yet valid

kubeadm alpha certs check-expiration

kubeadm alpha certs renew all --config=kubeadm.yaml
 systemctl restart kubelet
 kubeadm init phase kubeconfig all --config kubeadm.yaml
 然后将生成的配置文件替换，重启kube-apiserver、kube-controller、kube-scheduler、etcd这4个容器即可

首先在/etc/kubernetes/manifests/kube-controller-manager.yaml文件加入配置
spec:
  containers:
  - command:
    - kube-controller-manager
    # 设置证书有效期为10年
    - --experimental-cluster-signing-duration=87600h 
    - --client-ca-file=/etc/kubernetes/pki/ca.crt

kubeadm alpha certs renew all --use-api --config kubeadm.yaml &

# 先拷贝静态Pod资源清单
 cp -r /etc/kubernetes/manifests/ /etc/kubernetes/manifests.bak
 vi /etc/kubernetes/manifests/etcd.yaml
......
spec:
  containers:
  - command:
    - etcd
    # 修改为CA文件
    - --peer-trusted-ca-file=/etc/kubernetes/pki/ca.crt
    - --trusted-ca-file=/etc/kubernetes/pki/ca.crt
......
    volumeMounts:
    - mountPath: /var/lib/etcd
      name: etcd-data
    - mountPath: /etc/kubernetes/pki  # 更改证书目录
      name: etcd-certs
  volumes:
  - hostPath:
      path: /etc/kubernetes/pki  # 将 pki 目录挂载到etcd中去
      type: DirectoryOrCreate
    name: etcd-certs
  - hostPath:
      path: /var/lib/etcd 
      type: DirectoryOrCreate
    name: etcd-data
......

vi /etc/kubernetes/manifests/kube-apiserver.yaml
......
spec:
  containers:
  - command:
    - kube-apiserver
    # 将etcd ca文件修改为默认的ca.crt文件
    - --etcd-cafile=/etc/kubernetes/pki/ca.crt
......

cp /etc/kubernetes/pki/ca.crt /etc/kubernetes/pki/front-proxy-ca.crt
 cp /etc/kubernetes/pki/ca.key /etc/kubernetes/pki/front-proxy-ca.key

相关推荐: 我做了一个项目，惊艳了面试官，当场给我开了50w年薪

Java目前已经形成了庞大的生态圈，应用于各大企业当中，而很多学习Java的同学，在找工作的时候，由于项目实战经验不足，常常无法通过企业的面试筛选。 许多学习Java的同学在学习过程中无法接触到企业真实的开发流程，所以，在面试的时候，他们无法回答Java项目实…