Giters

ruderal04o / docker-setup

My docker setup scripts

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Docker Setup Script

系统

扩展软件源:

# 对用户启用systemd支持
sudo loginctl enable-linger <用户名>

# For user session
export XDG_RUNTIME_DIR="/run/user/$UID"
export DBUS_SESSION_BUS_ADDRESS="unix:path=${XDG_RUNTIME_DIR}/bus"

# For rootless docker, install uidmap(which contains newuidmap and newgidmap)
grep ^$(whoami): /etc/subuid # Test subuid
grep ^$(whoami): /etc/subgid # Test subgid
export DOCKER_HOST=unix://$XDG_RUNTIME_DIR/docker.sock
# export DOCKERD_ROOTLESS_ROOTLESSKIT_PORT_DRIVER=slirp4netns

启动命令备注

# 带systemd
podman run docker run -d --systemd true IMAGE /sbin/init

## systemd expects to have /run, /run/lock and /tmp on tmpfs
## It also expects to be able to write to /sys/fs/cgroup/systemd and /var/log/journal
## docker run -d --cap-add=SYS_ADMIN -v /sys/fs/cgroup:/sys/fs/cgroup IMAGE /sbin/init
## Mount list come from setupSystemd@libpod/container_internal_linux.go on https://github.com/containers/libpod
docker build --tag router-base -f debian10.router.raw.Dockerfile
docker run -d --name router --cap-add=SYS_ADMIN                             \
        --mount type=tmpfs,target=/run,tmpfs-size=67108864                  \
        --mount type=tmpfs,target=/run/lock,tmpfs-size=67108864             \
        --mount type=tmpfs,target=/tmp                                      \
        --mount type=tmpfs,target=/var/log/journal                          \
        --mount type=bind,source=/sys/fs/cgroup,target=/sys/fs/cgroup       \
        IMAGE /sbin/init
        # --mount type=bind,source=/sys/fs/cgroup/systemd,target=/sys/fs/cgroup/systemd

# 路由
podman build --tag router-base -f debian10.router.raw.Dockerfile
podman run -d --name router --systemd true                                                  \
       --mount type=bind,source=/home/router,target=/home/router                            \
       --mount type=bind,source=/opt/nftables,target=/opt/nftables,ro=true                  \
       --cap-add=NET_ADMIN --network=host router-base /sbin/init

podman run -d --name router --systemd true                                                  \
       --mount type=bind,source=/home/router,target=/home/router                            \
       --cap-add=NET_ADMIN --network=host router-base /lib/systemd/systemd

# @see https://docs.docker.com/engine/reference/builder/#entrypoint for detail about CMD and ENTRYPOINT

# 查看当前内核所有可用的模块
find /lib/modules/$(uname -r) -type f -name '*.ko*' | xargs basename -a | sort | uniq

# 查看已安装的内核所有可用的模块
find /lib/modules/ -type f -name '*.ko*' | awk '{if (match($0, /^\/lib\/modules\/([^\/]+).*\/([^\/]+)\.ko(\.[^\/\.]+)?$/, m)) {print m[1] " : " m[2];}}' | sort | uniq

# 查看和管理当前内核加载的模块信息
insmod/modprobe # 加载
rmmod           # 卸载
lsmod           # 查看系统中所有已经被加载了的所有的模块以及模块间的依赖关系
modinfo         # 获得模块的信息
cat /proc/modules  # 能够显示模块大小、在内核空间中的地址
cat /proc/devices  # 只显示驱动的主设备号，且是分类显示
ls /sys/modules    # 下面存在对应的驱动的目录，目录下包含驱动的分段信息等等。

配置firewalld

apt/dnf/yum install firewalld;
systemctl enable firewalld;
systemctl start firewalld;
vim /etc/firewalld/firewalld.conf ; # using FirewallBackend=nftables
firewall-cmd --permanent --add-service=dhcp;
firewall-cmd --permanent --add-service=dhcpv6;
firewall-cmd --permanent --add-service=dhcpv6-client;
firewall-cmd --permanent --add-service=dns;
firewall-cmd --permanent --add-service=ssh;
# firewall-cmd --permanent --add-service=custom services;
# firewall-cmd --permanent --add-port=custom tcp port/tcp;
# firewall-cmd --permanent --add-port=custom udp port/udp;
firewall-cmd --reload ;
firewall-cmd --list-all ;

文档地址备注

更新内核备注

## CentOS - add elrepo from http://elrepo.org/tiki/tiki-index.php
yum/dnf --enablerepo=elrepo-kernel install kernel-ml kernel-ml-core kernel-ml-modules kernel-ml-devel kernel-ml-modules-extra

if [[ "x$NEED_KERNEL_TOOLS" != "x" ]] || [[ "x$NEED_REBUILD_GLIBC" != "x" ]]; then
    yum/dnf remove -y kernel-headers kernel-tools kernel-tools-libs ;
    yum/dnf --enablerepo=elrepo-kernel install -y kernel-ml-tools kernel-ml-tools-libs ;
fi
if [[ "x$NEED_REBUILD_GLIBC" != "x" ]]; then
    yum/dnf remove -y kernel-headers ;
    yum/dnf --enablerepo=elrepo-kernel install -y kernel-ml-headers ;
    # install gcc again
    yum/dnf install -y gcc gcc-c++ libtool ;
    yum/dnf --enablerepo=elrepo-kernel update -y;
fi

### Update boot order - CentOS/RHEL 8 only
for KERNEL_PATH in /boot/vmlinuz-* ; do 
    printf "============ %s ============\n%s\n" "$KERNEL_PATH" "$(grubby --info=$KERNEL_PATH)";
done
grubby --make-default ;
echo "Current boot -> Kernel: $(grubby --default-kernel), Index: $(grubby --default-index)" ;

### Update boot order - CentOS/RHEL 7 only
awk -F\' '$1=="menuentry " {print $2}' /etc/grub2.cfg ;
SELECT_KERNEL_INDEXS=($(awk -F\' '$1=="menuentry " {print $2}' /etc/grub2.cfg | grep -n '[(]\s*[4-9]\.' | cut -d: -f 1));
if [ ${#SELECT_KERNEL_INDEXS} -gt 0 ]; then
    ((SELECT_KERNEL_INDEX=${SELECT_KERNEL_INDEXS[0]}-1));
    grub2-set-default $SELECT_KERNEL_INDEX ;
    echo "Now, you can reboot to use kernel $(awk -F\' '$1=="menuentry " {print $2}' /etc/grub2.cfg | grep '[(]\s*[4-9]\.' | head -n 1) ";
    echo -e "Please to rerun this script if you want to update componments in this system, or use \\033[33;1myum --enablerepo=elrepo-kernel update -y\\033[0m to update system";
else
    echo -e "\\033[31;1mError: kernel 4.X or upper not found.\\033[0m";
fi

## Ubuntu - download from https://kernel.ubuntu.com/~kernel-ppa/mainline/ or run scripts below
apt search "linux-image-" | awk '$0 ~ /linux-image-[0-9\.-]+-generic/ {print $0}' ;
apt install linux-image-<VERSION>-generic

## Debian
apt search linux-image;
apt install linux-image-<VERSION>

Docker备注

yum/apt install -y docker-ce 或 dnf install -y podman

# 手动设置docker需要的网桥
sudo brctl addbr docker0
sudo ip addr add 192.168.10.1/24 dev docker0
sudo ip link set dev docker0 up
ip addr show docker0

# 代理必须在启动脚本加环境变量
HTTP_PROXY=$http_proxy
HTTPS_PROXY=$https_proxy
NO_PROXY=$no_proxy

# /etc/docker/daemon.json 里可配不用验证证书的服务和存储位置
{
    "graph": "/data/docker-data",
    "storage-driver": "overlay",
    "insecure-registries" : [ "docker.io" ]
}

生成Kubernetes secret

# Generate secret with kubectl

# === docker-registry ===
# export KUBECONFIG=$HOME/.kube/config
#  Or add --kubeconfig=PATH_TO_CONFIG_FILE.conf after all kubectl command
kubectl config view
kubectl create secret docker-registry my-secret-name --docker-username=USERNAME --docker-password=PASSWORD --docker-email=ci@DOMAIN --docker-server=docker.io \
  --dry-run -o yaml > my-secret-name.yaml
kubectl -n NAMESPACE apply -f my-secret-name.yaml # Deploy, optional
# kubectl -n NAMESPACE get secret # Check

# === generic - kubernetes.io/dockerconfigjson ===
# export KUBECONFIG=$HOME/.kube/config
mkdir "$PWD/temporary-docker-config"
echo "PASSWARD" | docker --config "$PWD/temporary-docker-config" login docker.io -u USERNAME --password-stdin
kubectl create secret generic NAME --type=kubernetes.io/dockerconfigjson --from-file=.dockerconfigjson=$PWD/temporary-docker-config/config.json
rm -rf "$PWD/temporary-docker-config"

systemd/journald 备注

日志

echo "$LOG_CONTENT" | systemd-cat -t TARGET_NAME -p info ; # Write log
journalctl -t TARGET_NAME ; # Review log

Configure GC of journal

# Usage
journalctl --disk-usage
du -sh /var/log/journal/

# Rotate and cleanup
journalctl --rotate

# Clear journal log older than x days
journalctl --vacuum-time=2d

# Restrict logs to a certain size
journalctl --vacuum-size=100M

# Restrict number of log files
journalctl --vacuum-files=5

Auto cleanup: edit /etc/systemd/journald.conf

systemctl restart systemd-journald

常见错误

  • podman启动报 Error: OCI runtime error: container_linux.go:380: starting container process caused: error adding seccomp filter rule for syscall bdflush: requested action matches default action of filter

启动时增加 --security-opt seccomp=unconfined 参数

请先确保 grep CONFIG_SECCOMP= /boot/config-$(uname -r) 输出 CONFIG_SECCOMP=y

docker-compose.yml

version: "3.9"  # optional since v1.27.0
services:
  samba-pix:
    image: samba-server:latest
    security_opt:
      - seccomp=unconfined
      - label=disable
    ports:
      - "139:139/TCP"
      - "445:445/TCP"
      - "137:137/UDP"
      - "138:138/UDP"
    volumes:
      - type: bind
        source: /home/owent/docker-data/samba
        target: /data/samba
      - type: bind
        source: /sys/fs/cgroup
        target: /sys/fs/cgroup
    tmpfs:
      - /run:exec,mode=1777,size=67108864
      - /run/lock:exec,mode=1777,size=67108864
      - /tmp:exec,mode=1777
      - /var/log/journal:exec,mode=1777
  • podman启动访问后bind的目录提示 Permission denied

启动时增加 --security-opt label=disable 参数或关闭 selinux （修改 /etc/selinux/config 后重启）

  • docker内mount目录提示 pam_open_session: Permission denied 和 policy plugin failed session initialization

启动时增加 --cap-add CAP_SYS_RESOURCE 。

请注意rootless模式下 ~/.config/systemd/user/docker.service 或其他类似systemd配置内的限制不能大于 /etc/security/limits.d/ 和 cat /etc/security/limits.conf 的配置。

  • docker内mount nfs提示 mount.nfs: Operation not permitted

启动时增加 --cap-add CAP_SYS_ADMIN 。

rootless模式下建议母机mount完共享到子机。编辑 /etc/fstab 增加 NFS_REMOTE:/NFS_REMOTE_PATH /NFS_LOCAL_PATH nfs rw,nolock 0 0 ，然后手动mount一下: sudo mount -t nfs -o rw,nolock NFS_REMOTE:/NFS_REMOTE_PATH /NFS_LOCAL_PATH

About

My docker setup scripts

License:MIT License

Languages

Language:Shell 88.1%Language:Dockerfile 11.9%