tags: worker, docker
docker 是容器的运行环境,管理它的生命周期。kubelet 通过 Container Runtime Interface (CRI) 与 docker 进行交互。
注意:如果没有特殊指明,本文档的所有操作均在 m7-autocv-gpu01 节点上执行,然后远程分发文件和执行命令。
到 https://download.docker.com/linux/static/stable/x86_64/ 页面下载最新发布包:
cd /opt/k8s/work
wget https://download.docker.com/linux/static/stable/x86_64/docker-18.09.0.tgz
tar -xvf docker-18.09.0.tgz
分发二进制文件到所有 worker 节点:
cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
do
echo ">>> ${node_ip}"
scp docker/* root@${node_ip}:/opt/k8s/bin/
ssh root@${node_ip} "chmod +x /opt/k8s/bin/*"
done
cd /opt/k8s/work
cat > docker.service <<"EOF"
[Unit]
Description=Docker Application Container Engine
Documentation=http://docs.docker.io
[Service]
WorkingDirectory=##DOCKER_DIR##
Environment="PATH=/opt/k8s/bin:/bin:/sbin:/usr/bin:/usr/sbin"
EnvironmentFile=-/run/flannel/docker
ExecStart=/opt/k8s/bin/dockerd $DOCKER_NETWORK_OPTIONS
ExecReload=/bin/kill -s HUP $MAINPID
Restart=on-failure
RestartSec=5
LimitNOFILE=infinity
LimitNPROC=infinity
LimitCORE=infinity
Delegate=yes
KillMode=process
[Install]
WantedBy=multi-user.target
EOF
-
EOF 前后有双引号,这样 bash 不会替换文档中的变量,如 $DOCKER_NETWORK_OPTIONS;
-
dockerd 运行时会调用其它 docker 命令,如 docker-proxy,所以需要将 docker 命令所在的目录加到 PATH 环境变量中;
-
flanneld 启动时将网络配置写入
/run/flannel/docker
文件中,dockerd 启动前读取该文件中的环境变量DOCKER_NETWORK_OPTIONS
,然后设置 docker0 网桥网段; -
如果指定了多个
EnvironmentFile
选项,则必须将/run/flannel/docker
放在最后(确保 docker0 使用 flanneld 生成的 bip 参数); -
docker 需要以 root 用于运行;
-
docker 从 1.13 版本开始,可能将 iptables FORWARD chain的默认策略设置为DROP,从而导致 ping 其它 Node 上的 Pod IP 失败,遇到这种情况时,需要手动设置策略为
ACCEPT
:$ sudo iptables -P FORWARD ACCEPT
并且把以下命令写入
/etc/rc.local
文件中,防止节点重启iptables FORWARD chain的默认策略又还原为DROP/sbin/iptables -P FORWARD ACCEPT
完整 unit 见 docker.service
分发 systemd unit 文件到所有 worker 机器:
cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
sed -i -e "s/##DOCKER_DIR##/${DOCKER_DIR}/" docker.service
for node_ip in ${NODE_IPS[@]}
do
echo ">>> ${node_ip}"
scp docker.service root@${node_ip}:/etc/systemd/system/
done
使用国内的仓库镜像服务器以加快 pull image 的速度,同时增加下载的并发数 (需要重启 dockerd 生效):
cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
cat > docker-daemon.json <<EOF
{
"registry-mirrors": ["https://hub-mirror.c.163.com", "https://docker.mirrors.ustc.edu.cn"],
"insecure-registries": ["docker02:35000"],
"max-concurrent-downloads": 20,
"live-restore": true,
"max-concurrent-uploads": 10,
"debug": true,
"data-root": "${DOCKER_DIR}/data",
"exec-root": "${DOCKER_DIR}/exec",
"log-opts": {
"max-size": "100m",
"max-file": "5"
}
}
EOF
分发 docker 配置文件到所有 work 节点:
cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
do
echo ">>> ${node_ip}"
ssh root@${node_ip} "mkdir -p /etc/docker/ ${DOCKER_DIR}/{data,exec}"
scp docker-daemon.json root@${node_ip}:/etc/docker/daemon.json
done
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
do
echo ">>> ${node_ip}"
ssh root@${node_ip} "systemctl stop firewalld && systemctl disable firewalld"
ssh root@${node_ip} "/usr/sbin/iptables -F && /usr/sbin/iptables -X && /usr/sbin/iptables -F -t nat && /usr/sbin/iptables -X -t nat"
ssh root@${node_ip} "/usr/sbin/iptables -P FORWARD ACCEPT"
ssh root@${node_ip} "systemctl daemon-reload && systemctl enable docker && systemctl restart docker"
#ssh root@${node_ip} 'for intf in /sys/devices/virtual/net/docker0/brif/*; do echo 1 > $intf/hairpin_mode; done'
ssh root@${node_ip} "sudo sysctl -p /etc/sysctl.d/kubernetes.conf"
done
- 关闭 firewalld(centos7)/ufw(ubuntu16.04),否则可能会重复创建 iptables 规则;
- 清理旧的 iptables rules 和 chains 规则;
- 开启 docker0 网桥下虚拟网卡的 hairpin 模式;
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
do
echo ">>> ${node_ip}"
ssh root@${node_ip} "systemctl status docker|grep Active"
done
确保状态为 active (running)
,否则查看日志,确认原因:
$ journalctl -u docker
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
do
echo ">>> ${node_ip}"
ssh root@${node_ip} "/usr/sbin/ip addr show flannel.1 && /usr/sbin/ip addr show docker0"
done
确认各 work 节点的 docker0 网桥和 flannel.1 接口的 IP 处于同一个网段中(如下 172.30.112.0/32 位于 172.30.112.1/21 中):
>>> 172.27.128.150
4: flannel.1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UNKNOWN group default
link/ether 0a:28:6f:9b:1e:9a brd ff:ff:ff:ff:ff:ff
inet 172.30.112.0/32 scope global flannel.1
valid_lft forever preferred_lft forever
5: docker0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN group default
link/ether 02:42:ab:92:54:c8 brd ff:ff:ff:ff:ff:ff
inet 172.30.112.1/21 brd 172.30.119.255 scope global docker0
valid_lft forever preferred_lft forever