Basic Vagrantfile for multiple groups of VMs

In case you want to prepare multiple sets of machines quickly using Vagrant, ready for different setups, this might be something for you:

## -*- mode: ruby -*-
## vi: set ft=ruby :

require 'ipaddr'

###############################
# CUSTOM CONFIGURATION START
###############################

# lab_name is the name of the lab where all the files will be organized.
lab_name = "lab_bigdata"

# here is where you download your software, so it will be available to the VMs.
sw_path  = "C:\\Users\\ludov\\Downloads\\Software"

# cluster(s) definition
clusters = [
  {
  :prefix  => "hadoop", 				# prefix: VMs will be named prefix01, prefix02, etc
  :domain  => "ludovicocaldara.net",	# domain name
  :box     => "ludodba/ol7.3-base",		# base box, either "ludodba/ol7.3-base" or "ludodba/ubu1604"
  :nodes   => 3,						# number of nodes for this cluster
  :cpu     => 1,
  :mem     => 2048,
  :publan  => IPAddr.new("192.168.56.0/24"), 	# public lan for the cluster
  :publan_start => 121							# starting IP, each VM will increment it by one
  },
  {
  :prefix  => "kafka",							# eventually, continue with another cluster!
  :domain  => "ludovicocaldara.net",
  :box     => "ludodba/ol7.3-base",
  :nodes   => 1,
  :cpu     => 1,
  :mem     => 2048,
  :publan  => IPAddr.new("192.168.56.0/24"),
  :publan_start => 131
  },
  {
  :prefix  => "postgres",
  :domain  => "ludovicocaldara.net",
  :box     => "ludodba/ubu1604",
  :nodes   => 1,
  :cpu     => 1,
  :mem     => 2048,
  :publan  => IPAddr.new("192.168.56.0/24"),
  :publan_start => 141
  }
]

###############################
# CUSTOM CONFIGURATION END
###############################

######################################################
# Extending Class IPAddr to add the CIDR to the lan
class IPAddr
  def to_cidr_s
    if @addr
      mask = @mask_addr.to_s(2).count('1')
      "#{to_s}/#{mask}"
    else
      nil
    end
  end
end # extend class IPAddr

########
# MAIN #
########

Vagrant.configure(2) do |config|
  config.ssh.username = "root"  	# my boxes are password based for simplicity
  config.ssh.password = "vagrant"
  config.vm.graceful_halt_timeout = 360	# in case you install grid infra... do not force shutdown after a few seconds

  if File.directory?(sw_path)
    # our shared folder for oracle 12c installation files (uid 54320 is grid, uid 54321 is oracle)
    config.vm.synced_folder sw_path, "/media/sw", :mount_options => ["dmode=775","fmode=775","uid=54322","gid=54328"]
  end

  # looping through each cluster
  (0..(clusters.length-1)).each do |cluid|

    # assign variable clu to current cluster, for convenience
    clu = clusters[cluid]
      
    # looping through each node in the cluster
    (1..(clu[:nodes])).each do |nid|

      # let's start from the last node (see RAC Attack automation for the reason) :-)
      nid = clu[:nodes]+1-nid
      config.vm.define vm_name = "#{clu[:prefix]}%02d" % nid do |cnf|
	  
		# set the right box for the VM
		cnf.vm.box = clu[:box]
		if (clu[:box_version]) then
			cnf.vm.box_version = clu[:box_version]
		end #if
		
		# the new vm name
        vm_name = "#{clu[:prefix]}%02d" % nid
        fqdn = "#{vm_name}.#{clu[:domain]}"
        cnf.vm.hostname = "#{fqdn}"

		# incrementing public ip for the cluster
        pubip = clu[:publan].|(clu[:publan_start]+nid-1).to_s

        cnf.vm.provider :virtualbox do |vb|
          #vb.linked_clone = true  # in case you want thin provisioning. read the vagrant doc before setting it
          vb.name = vm_name
          vb.gui = false
          vb.customize ["modifyvm", :id, "--memory", clu[:mem]]
          vb.customize ["modifyvm", :id, "--cpus",   clu[:cpu]]
          vb.customize ["modifyvm", :id, "--groups", "/#{lab_name}/#{clu[:prefix]}"]
        end #config.vm.provider
		
        # Configuring virtualbox network for #{pubip}
        cnf.vm.network :private_network, ip: pubip

      end #config.vm.define
    end #loop nodes
  end  #loop clusters
end #Vagrant.configure

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

## -*- mode: ruby -*-

## vi: set ft=ruby :

require 'ipaddr'

###############################

# CUSTOM CONFIGURATION START

###############################

# lab_name is the name of the lab where all the files will be organized.

lab_name = "lab_bigdata"

# here is where you download your software, so it will be available to the VMs.

sw_path = "C:\\Users\\ludov\\Downloads\\Software"

# cluster(s) definition

clusters = [

{

:prefix => "hadoop", # prefix: VMs will be named prefix01, prefix02, etc

:domain => "ludovicocaldara.net", # domain name

:box => "ludodba/ol7.3-base", # base box, either "ludodba/ol7.3-base" or "ludodba/ubu1604"

:nodes => 3, # number of nodes for this cluster

:cpu => 1,

:mem => 2048,

:publan => IPAddr.new("192.168.56.0/24"), # public lan for the cluster

:publan_start => 121 # starting IP, each VM will increment it by one

{

:prefix => "kafka", # eventually, continue with another cluster!

:domain => "ludovicocaldara.net",

:box => "ludodba/ol7.3-base",

:nodes => 1,

:cpu => 1,

:mem => 2048,

:publan => IPAddr.new("192.168.56.0/24"),

:publan_start => 131

{

:prefix => "postgres",

:domain => "ludovicocaldara.net",

:box => "ludodba/ubu1604",

:nodes => 1,

:cpu => 1,

:mem => 2048,

:publan => IPAddr.new("192.168.56.0/24"),

:publan_start => 141

}

]

###############################

# CUSTOM CONFIGURATION END

###############################

######################################################

# Extending Class IPAddr to add the CIDR to the lan

class IPAddr

def to_cidr_s

if @addr

mask = @mask_addr.to_s(2).count('1')

"#{to_s}/#{mask}"

else

nil

end

end # extend class IPAddr

########

# MAIN #

########

Vagrant.configure(2) do |config|

config.ssh.username = "root" # my boxes are password based for simplicity

config.ssh.password = "vagrant"

config.vm.graceful_halt_timeout = 360 # in case you install grid infra... do not force shutdown after a few seconds

if File.directory?(sw_path)

# our shared folder for oracle 12c installation files (uid 54320 is grid, uid 54321 is oracle)

config.vm.synced_folder sw_path, "/media/sw", :mount_options => ["dmode=775","fmode=775","uid=54322","gid=54328"]

end

# looping through each cluster

(0..(clusters.length-1)).each do |cluid|

# assign variable clu to current cluster, for convenience

clu = clusters[cluid]

# looping through each node in the cluster

(1..(clu[:nodes])).each do |nid|

# let's start from the last node (see RAC Attack automation for the reason) :-)

nid = clu[:nodes]+1-nid

config.vm.define vm_name = "#{clu[:prefix]}%02d" % nid do |cnf|

# set the right box for the VM

cnf.vm.box = clu[:box]

if (clu[:box_version]) then

cnf.vm.box_version = clu[:box_version]

end #if

# the new vm name

vm_name = "#{clu[:prefix]}%02d" % nid

fqdn = "#{vm_name}.#{clu[:domain]}"

cnf.vm.hostname = "#{fqdn}"

# incrementing public ip for the cluster

pubip = clu[:publan].|(clu[:publan_start]+nid-1).to_s

cnf.vm.provider :virtualbox do |vb|

#vb.linked_clone = true # in case you want thin provisioning. read the vagrant doc before setting it

vb.name = vm_name

vb.gui = false

vb.customize ["modifyvm", :id, "--memory", clu[:mem]]

vb.customize ["modifyvm", :id, "--cpus", clu[:cpu]]

vb.customize ["modifyvm", :id, "--groups", "/#{lab_name}/#{clu[:prefix]}"]

end #config.vm.provider

# Configuring virtualbox network for #{pubip}

cnf.vm.network :private_network, ip: pubip

end #config.vm.define

end #loop nodes

end #loop clusters

end #Vagrant.configure

The nice thing, (beside speeding up the creation and basic configuration) is the organization of the directories. The configuration at the beginning of the script will result in 5 virtual machines:

your VM directory
        |- lab_bigdata 
                |- hadoop
                        |- hadoop01  (ol7)
                        |- hadoop02  (ol7)
                        |- hadoop03  (ol7)
                |- kafka
                        |- kafka01   (ol7)
                |- postgres
                        |- postgres01  (ubuntu 16.04)

your VM directory

|- lab_bigdata

|- hadoop

|- hadoop01 (ol7)

|- hadoop02 (ol7)

|- hadoop03 (ol7)

|- kafka

|- kafka01 (ol7)

|- postgres

|- postgres01 (ubuntu 16.04)

It is based, in part (but modified and simplified a lot), from the RAC Attack automation scripts by Alvaro Miranda.

I have a more complex version that automates all the tasks for a full multi-cluster RAC environment, but if this is your requirement, I would rather check oravirt scripts on github (https://github.com/oravirt) . They are much more powerful and complete (and complex…) than my Vagrantfile. 🙂

Cheers