Question

我出于研究目的使用vagrant来部署多个VM环境，到目前为止它还很棒。但是现在我需要将每个docker容器固定到一个特定的CPU核心，但我不知道如何使用vagrant来做到这一点。我知道我可以使用＆＃34; args＆＃34; Vagrantfile上的子句传递＆＃34; - cpuset＆＃34; docker run命令的参数，但我不知道如何在循环中使用它，因为我启动了多个容器，我需要将每个容器固定到不同的CPU核心（例如，node1引脚到核心＃0，节点2引脚到核心＃1等。

我目前的Vagrantfile如下，没有CPU Pinning的东西：

# Vagrantfile API/syntax version. Don't touch unless you know what you're doing!
VAGRANTFILE_API_VERSION = "2"

# choose how many machines the cluster will contain
N_VMS = 32

Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
  config.vm.box = "tknerr/baseimage-ubuntu-14.04"
  config.vm.network "private_network", ip: "192.168.121.2"
  config.vm.provider "docker" do |v|
    v.has_ssh = true
  end

  hosts_file = []
  1.upto(N_VMS) do |i|
    config.vm.define vm_name = "node#{i}" do |config|
      config.vm.hostname = vm_name
    end
  end

  script = <<-SCRIPT
      apt-get -y update
      apt-get -y install libcr-dev mpich2 mpich2-doc arp-scan openssh-server nano make
  SCRIPT
  script.sub! 'N_VMS', N_VMS.to_s
  config.vm.provision "shell", inline: script

end

Answer 1

假设你想在循环中配置vagrant vm，因为vagrantfile是基于ruby语言的，所以你可以将开发可用性包含在vagrantfile中，这里有一个例子，你可以添加你的“--cpuset”在vm define中配置。

# -*- mode: ruby -*-
# vi: set ft=ruby :
# read vm and chef configurations from JSON files
nodes_config = (JSON.parse(File.read("nodes.json")))['nodes']
VAGRANTFILE_API_VERSION = "2"
Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
    nodes_config.each do |node|
    node_name   = node[0] # name of node
    node_values = node[1] # content of node
            # puts node_name
    # puts node_values['box']
    config.vm.define node_name do |config|
        config.vm.box = node_values['box']
        config.vm.hostname = node_name
        config.vm.network :private_network, ip: node_values['ip']
                    config.vm.provider :virtualbox do |vb|
            vb.customize ["modifyvm", :id, "--memory", node_values['memory']]
            vb.customize ["modifyvm", :id, "--name", node_name]
        end
    end 
end

nodes.json定义vm，您可以定义

{
  "nodes": {
    "jenkins.example.com": {
      "info": "jenkins master server",
      "box": "../Vgrant-boxes/centos65_virtualbox_50G.box",
      "ip": "192.168.35.101",
      "ports": [],
      "memory": 512
    },
    "node01.example.com": {
      "info": "tomcat app host server",
      "box": "../Vgrant-boxes/centos65_virtualbox_50G.box",
      "ip": "192.168.35.121",
      "ports": [],
      "memory": 512
    },
    "node02.example.com": {
      "info": "jboss app host server",
      "box": "../Vgrant-boxes/centos65_virtualbox_50G.box",
      "ip": "192.168.35.122",
      "ports": [],
      "memory": 512
    },
    "node03.example.com": {
      "info": "oracle xe server",
      "box": "../Vgrant-boxes/centos65_virtualbox_50G.box",
      "ip": "192.168.35.123",
      "ports": [],
      "memory": 512
    },
    "node04.example.com": {
      "info": "artifactory server",
      "box": "../Vgrant-boxes/centos65_virtualbox_50G.box",
      "ip": "192.168.35.124",
      "ports": [],
      "memory": 512
    }
  }
}

Answer 2

最后，我在找错了地方。添加“--cpuset-cpus”的正确位置在config.vm.define块中。

代码结束如下：

# Vagrantfile API/syntax version. Don't touch unless you know what you're doing!
VAGRANTFILE_API_VERSION = "2"

# choose how many machines the cluster will contain
N_VMS = 32

Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
  config.vm.box = "tknerr/baseimage-ubuntu-14.04"
  config.vm.network "private_network", ip: "192.168.121.2"
  config.vm.provider "docker" do |v|
    v.has_ssh = true
  end

  2.upto(N_VMS+1) do |i|
    config.vm.define vm_name = "node#{i}" do |config|

      # CPU PINNING CONFIG    
      config.vm.provider "docker" do |docker|
        docker.create_args = ['--cpuset-cpus=' + ((i/2)-1).to_s]
      end

      config.vm.hostname = vm_name
    end
  end

  script = <<-SCRIPT
      apt-get -y update
      apt-get -y install libcr-dev mpich2 mpich2-doc arp-scan openssh-server nano make
  SCRIPT
  script.sub! 'N_VMS', N_VMS.to_s
  i=1
  config.vm.provision "shell", inline: script

end

Vagrant：使用vagrant将docker运行参数传递给多个容器

2 个答案: