☰

Vagrant 多次运行内联脚本，以及尚未定义的脚本

tag-icon • tag-icon scripts virtualbox vagrant
Vagrant 多次运行内联脚本，以及尚未定义的脚本

我正在使用 vagrant 启动多个虚拟机，这是我的 vagrantfile（直到第一个主机，删除注释）：

# Installed Vagrant Version: 2.2.14
Vagrant.configure("2") do |config|
  config.vm.box = "bento/ubuntu-20.04"
  config.vm.provider :virtualbox

  mount_new_disk = <<-SCRIPT
  sudo useradd -m -s /bin/bash -G sudo testuser1
  sudo echo -e "123\n123"|sudo passwd testuser1
  sudo mkdir /mnt/da
  sudo mkfs -t ext4 /dev/sdb
  sudo mount /dev/sdb /mnt/da
  echo "/dev/sdb /mnt/da ext4 auto,nofail,noatime,nolock,intr,tcp,actimeo=1800 0 0" |sudo tee -a /etc/fstab
  SCRIPT

  make_default_gw = <<-SCRIPT
  superda=$(sudo ip route show|head -1|awk '{print $3}');if ! [[ $superda =~ "192.168.1.254" ]]; then echo "router is not default, removing";sudo ip route del default via $superda ;else echo "yes it's default, gud";fi
  SCRIPT

  nfs_host = <<-SCRIPT
  sudo apt update
  sudo apt install nfs-kernel-server
  echo "/mnt/da    10.118.8.0/24(rw,sync,no_subtree_check)"|sudo tee -a /etc/exports
  sudo chown nobody:nogroup /mnt/da
  sudo chmod 777 /mnt/da
  sudo systemctl restart nfs-kernel-server
  sudo ufw allow from 10.118.8.0/24 to any port nfs
  sudo ufw allow from 10.118.8.0/24 to any port ssh
  sudo ufw enable
  SCRIPT

  nfs_client = <<-SCRIPT
  sudo apt update
  sudo apt install nfs-common
  sudo mkdir /mnt/nfs
  sudo mount 10.118.8.10:/mnt/da /mnt/nfs
  echo "10.118.8.10:/mnt/da    /mnt/nfs   nfs auto,nofail,noatime,nolock,intr,tcp,actimeo=1800 0 0" |sudo tee -a /etc/fstab
  SCRIPT

  config.vm.define "da1" do |da1|
    da1.vm.synced_folder ".", "/vagrant", disabled: true
    da1.vm.disk :disk, name: "backup", size: "10GB"
      da1.vm.hostname = "da1"
      da1.vm.network "private_network", ip: "10.118.8.10"
    da1.vm.network "public_network", use_dhcp_assigned_default_route: true, netmask: "255.255.255.0", bridge: "Intel(R) I211 Gigabit Network Connection"
    da1.vm.provision :shell, :inline => mount_new_disk
    da1.vm.provision :shell, :inline => make_default_gw
    da1.vm.provision :shell, :inline => nfs_host
      config.ssh.forward_agent = true

  end

  config.vm.define "da2" do |da2|
    da2.vm.synced_folder ".", "/vagrant", disabled: true
    da2.vm.disk :disk, name: "backup", size: "10GB"
      da2.vm.hostname = "da2"
      da2.vm.network "private_network", ip: "10.118.8.11"
    da2.vm.network "public_network", use_dhcp_assigned_default_route: true, netmask: "255.255.255.0", bridge: "Intel(R) I211 Gigabit Network Connection"
    config.vm.provision :shell, :inline => mount_new_disk
    config.vm.provision :shell, :inline => make_default_gw
    config.vm.provision :shell, :inline => nfs_client
    config.ssh.forward_agent = true
  end

  config.vm.define "da3" do |da3|
    da3.vm.synced_folder ".", "/vagrant", disabled: true
    da3.vm.disk :disk, name: "backup", size: "10GB"
    da3.vm.hostname = "da3"
    da3.vm.network "private_network", ip: "10.118.8.12"
      da3.vm.network "public_network", use_dhcp_assigned_default_route: true, netmask: "255.255.255.0", bridge: "Intel(R) I211 Gigabit Network Connection"
#    da3.vm.network "forwarded_port", guest: 22, host: 5555
    config.vm.provision :shell, :inline => mount_new_disk
    config.vm.provision :shell, :inline => make_default_gw
    config.vm.provision :shell, :inline => nfs_client
    config.ssh.forward_agent = true
  end

现在脚本mount_new_disk和make_default_gw应该在所有主机上运行，​​而nfs_host应该仅在上运行da1，并且nfs_client应该仅在上运行da[23]。

但是对于某些 readon，每个脚本在每个主机上运行，​​并且它们也运行多次。

请注意，我所遵循的教程说要写config.vm.provision :shell, :inline => mount_new_disk，但我也尝试了，da1.vm.provision :shell, :inline => mount_new_disk结果相同。

例如：我将echo 'RUNNING SCRIPT START: ...'在每个脚本的顶部和echo 'RUNNING SCRIPT END: ...'底部写一个来查看哪个脚本运行了以及运行了多少次（捕获输出然后对其进行 grepped）：

unu@here:~/yes$ grep -n 'RUNNING SCRIPT' opt
44:    da1: RUNNING SCRIPT START: mount_new_disk
62:    da1: RUNNING SCRIPT STOP: mount_new_disk
65:    da1: RUNNING SCRIPT START: make_default_gw
67:    da1: RUNNING SCRIPT STOP: make_default_gw
70:    da1: RUNNING SCRIPT START: nfs_client
120:    da1: RUNNING SCRIPT STOP: nfs_client
123:    da1: RUNNING SCRIPT START: mount_new_disk
137:    da1: RUNNING SCRIPT STOP: mount_new_disk
140:    da1: RUNNING SCRIPT START: make_default_gw
142:    da1: RUNNING SCRIPT STOP: make_default_gw
145:    da1: RUNNING SCRIPT START: nfs_client
174:    da1: RUNNING SCRIPT STOP: nfs_client
177:    da1: RUNNING SCRIPT START: mount_new_disk
191:    da1: RUNNING SCRIPT STOP: mount_new_disk
194:    da1: RUNNING SCRIPT START: make_default_gw
196:    da1: RUNNING SCRIPT STOP: make_default_gw
199:    da1: RUNNING SCRIPT START: nfs_host
261:    da1: RUNNING SCRIPT STOP: nfs_host

正如您所看到的，它运行了所有脚本（甚至是那些在其自己的配置部分中未提及的脚本），并且多次运行了它们。

有趣的是，在 nfs 客户端 da2 上，它没有运行该nfs_host脚本，但它仍然运行了其他所有内容两次：

unu@here:~/yes$ grep -n 'RUNNING SCRIPT' noua
41:    da2: RUNNING SCRIPT START: mount_new_disk
61:    da2: RUNNING SCRIPT STOP: mount_new_disk
64:    da2: RUNNING SCRIPT START: make_default_gw
66:    da2: RUNNING SCRIPT STOP: make_default_gw
69:    da2: RUNNING SCRIPT START: nfs_client
118:    da2: RUNNING SCRIPT STOP: nfs_client
121:    da2: RUNNING SCRIPT START: mount_new_disk
134:    da2: RUNNING SCRIPT STOP: mount_new_disk
137:    da2: RUNNING SCRIPT START: make_default_gw
139:    da2: RUNNING SCRIPT STOP: make_default_gw
142:    da2: RUNNING SCRIPT START: nfs_client
171:    da2: RUNNING SCRIPT STOP: nfs_client

现在我知道我可以将所有这些都放在配置部分本身中，如下所示：

  config.vm.define "da1" do |da1|
    da1.vm.synced_folder ".", "/vagrant", disabled: true
    da1.vm.disk :disk, name: "backup", size: "10GB"
      da1.vm.hostname = "da1"
      da1.vm.network "private_network", ip: "10.118.8.10"
    da1.vm.network "public_network", use_dhcp_assigned_default_route: true, netmask: "255.255.255.0", bridge: "Intel(R) I211 Gigabit Network Connection"
    da1.vm.provision :shell, :inline => mount_new_disk
    da1.vm.provision :shell, :inline => make_default_gw
    da1.vm.provision :shell, :inline => nfs_host
    da1.vm.provision :shell, :inline => <<-SCRIPT
    echo "yohoho and a bottle of rum!"
    SCRIPT
      config.ssh.forward_agent = true

但是另一种方式，只需在需要时调用它，就更方便了。

有人知道为什么会发生这种情况吗？以及如何解决它？

答案1

发现问题：我必须config.vm.provision :shell, :inline => script_name根据配置块用适当的虚拟机名称进行替换。

所以da1我做到了da1.vm.provision :shell, :inline => script_name，因为da2我做到了da2.vm.provision :shell, :inline => script_name，因为da3我做到了da3.vm.provision :shell, :inline => script_name。

相关内容