Vagrant: Multi-machine vagrantfile wrong configuration references when a machine is specified in vagrant up
Vagrant version
Vagrant 2.2.5
Host operating system
Macos Mojave/Ubuntu 18.04
Guest operating system
ubuntu 18.04/ubuntu 16.04
Vagrantfile
VAGRANTFILE_API_VERSION = '2'
ROOT_DIRECTORY = File.dirname(File.realpath(__FILE__))
Vagrant.require_version ">= 2.1.1"
Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
config.trigger.before :up, :provision do |trigger|
trigger.name = "upgrade or install librarian-puppet libs"
trigger.info = "upgrade or install librarian-puppet libs"
require 'fileutils'
FileUtils.mkdir_p './environments/production/modules'
trigger.on_error = :halt
trigger.run = {path: "bin/librarian-puppet.sh"}
end
config.vm.provision :shell, path: 'bin/bootstrap.sh'
config.vm.synced_folder './environments/production/site/easywelfare/templates/',
'/tmp/vagrant-puppet/templates'
config.vm.box_check_update = false
config.cache.scope = :box
config.cache.auto_detect = true
config.vm.boot_timeout = 600
{ :'test' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-test-1.public.#{ew_domain}",
:ip => '192.168.33.10',
:memory => '2048',
:cpus => 2,
:disk_size => '80GB',
:autostart => false,
:primary => false
},
:'test1' => {
:os => 'bento/ubuntu-18.04',
:hostname => "development-test-2.public.#{ew_domain}",
:ip => '192.168.33.11',
:memory => '2048',
:cpus => 2,
:autostart => false,
:primary => false
},
:'ws' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-ws-1.public.#{ew_domain}",
:ip => '192.168.33.12',
:memory => '2048',
:cpus => 1,
:autostart => false,
:primary => false
},
:'all' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-all-1.public.#{ew_domain}",
:ip => '192.168.33.13',
:memory => '2048',
:cpus => 2,
:autostart => false,
:primary => false
},
:'mail' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-mail-1.public.#{ew_domain}",
:ip => '192.168.33.14',
:memory => '1024',
:cpus => 3,
:autostart => false,
:primary => false
},
:'monitoring' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-monitoring-1.development.aws.#{ew_domain}",
:ip => '192.168.33.16',
:memory => '2048',
:cpus => 2,
:autostart => false,
:primary => false
},
:'puppetserver' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-puppetserver-1.development.aws.#{ew_domain}",
:ip => '192.168.33.17',
:memory => '4192',
:cpus => 2,
:autostart => false,
:primary => false
},
:'development' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-all-vm.public.#{ew_domain}",
:ip => '192.168.33.18',
:memory => '2048',
:cpus => 2,
:disk_size => '80GB',
:autostart => false,
:primary => false
},
:'development_test' => {
:os => 'bento/ubuntu-18.04',
:hostname => "development-all-vm.public.#{ew_domain}",
:ip => '192.168.33.19',
:memory => '2048',
:cpus => 2,
:autostart => false,
:primary => false
},
:'development_test1' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-all-vm.public.#{ew_domain}",
:ip => '192.168.33.20',
:memory => '3192',
:cpus => 2,
:autostart => false,
:primary => false
},
:'cicd' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-cicd-vm.public.#{ew_domain}",
:ip => '192.168.33.19',
:memory => '2048',
:cpus => 2,
:autostart => false,
:primary => false
},
:'openvpn' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-openvpn-vm.public.#{ew_domain}",
:ip => '192.168.33.19',
:memory => '2048',
:cpus => 2,
:autostart => false,
:primary => false
},
:'intranet' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-intranet-1.public.#{ew_domain}",
:ip => '192.168.33.20',
:memory => '2048',
:cpus => 1,
:autostart => false,
:primary => false
},
:'kubemaster' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-kubemaster-1.public.#{ew_domain}",
:ip => '192.168.33.20',
:memory => '2048',
:cpus => 1,
:autostart => false,
:primary => false
},
:'kubenode' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-kubenode-1.public.#{ew_domain}",
:ip => '192.168.33.21',
:memory => '2048',
:cpus => 1,
:autostart => false,
:primary => false
},
:'kubenode2' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-kubenode-2.public.#{ew_domain}",
:ip => '192.168.33.22',
:memory => '2048',
:cpus => 1,
:autostart => false,
:primary => false
},
:'kibana' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-kibana-1.public.#{ew_domain}",
:ip => '192.168.33.21',
:memory => '2048',
:cpus => 1,
:autostart => false,
:primary => false
},
:'aptcacher' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-aptcacher-1.development.aws.#{ew_domain}",
:ip => '192.168.33.24',
:memory => '1024',
:cpus => 1,
:autostart => false,
:primary => false
}
}.each do |name, configuration|
config.vm.define name ,
primary: configuration[:primary],
autostart: configuration[:autostart] do |instance|
if (configuration[:os].include?('16.04') || configuration[:os].include?('18.04'))
instance.cache.synced_folder_opts = {
owner: "_apt",
group: "_apt",
mount_options: ["dmode=777", "fmode=666"]
}
end
# change disk sizing if disk_size is declared
if configuration.has_key?(:disk_size)
puts " ip is #{configuration[:ip]}"
puts " hostname is #{configuration[:ip]}"
puts " name is #{name}"
puts "disk size required is #{configuration[:disk_size]}"
config.disksize.size = configuration[:disk_size]
end
instance.vm.box = configuration[:os]
instance.vm.hostname = configuration[:hostname]
instance.vm.network 'private_network', ip: configuration[:ip]
if vagrantconfig
if vagrantconfig[name.to_s]
vagrantconfig[name.to_s].each do |settings|
if settings['public_network'] and settings['ipaddress']
instance.vm.network "public_network", ip: settings['ipaddress']
elsif settings ['public_network']
instance.vm.network "public_network"
end
end
end
end
# VirtualBox
instance.vm.provider 'virtualbox' do |vb|
# Boot in headless mode
vb.gui = false
# VM customization
vb.cpus = configuration[:cpus]
vb.customize ['modifyvm', :id, '--memory', configuration[:memory]]
vb.customize ['modifyvm', :id, '--natnet1', '192.168.199/24']
if vagrantconfig
if vagrantconfig[name.to_s]
vagrantconfig[name.to_s].each do |settings|
if settings['ram']
vb.customize ['modifyvm', :id, '--memory', settings['ram']]
end
if settings['cpus']
vb.cpus = settings['cpus']
end
end
end
end
end
if configuration.has_key?(:disk_size)
instance.vm.provision :shell, path: 'bin/resize_lvm.sh'
end
instance.vm.provision 'puppet' do |puppet|
puppet.environment = 'production'
puppet.environment_path = 'environments'
puppet.hiera_config_path = 'etc/hiera-vagrant.yaml'
puppet.module_path = ['environments/production/site',
'environments/production/modules',
'environments/production/custom_modules']
puppet.options = ['--verbose',
# '--debug',
# '--trace',
'--show_diff',
'--write-catalog-summary',
'--disable_warnings=deprecations',
'--pluginsync',
'--graph']
end
project_dir = '/opt/projects/Easywelfare'
os_router= CommandExecutor.new(personal_files_dir,configuration[:ip],project_dir)
config.trigger.after :provision,:up do |trigger|
trigger.only_on = "development"
trigger.ignore = :halt
trigger.name = "mount projects dir"
trigger.info = "mount projects dir"
retries = 0
max_retries = 10
begin
port_open?(configuration[:ip], 22)
rescue Exception => e
if retries <= max_retries
retries += 1
sleep 2 ** retries
retry
else
raise "Timeout: #{e.message}"
end
end
if ! os_router.already_mounted?
trigger.run = {inline: os_router.mount}
end
end
config.trigger.before :halt do |trigger|
trigger.only_on = "development"
trigger.name = "umount projects dir"
trigger.info = "umount projects dir"
if os_router.already_mounted?
trigger.run= {inline: os_router.umount}
end
end
if File.exists?('customize-after.sh')
instance.vm.provision 'shell', path: 'customize-after.sh'
end
end
end
end
def port_open?(ip, port, seconds=1)
Timeout::timeout(seconds) do
begin
TCPSocket.new(ip, port).close
true
rescue Errno::ECONNREFUSED, Errno::EHOSTUNREACH, SocketError
false
end
end
rescue Timeout::Error
false
end
Debug Output
https://gist.github.com/lzecca78/c2019a65990da676285c6ba3a85e96c2#file-vagrant_debug_output
Expected behavior
When i hit vagrant up development i expect that the references to the development hash are right and only that hash is used
Actual behavior
For some reasons, the provision takes also the first element of the list (in the case above the hash with key test), passing then wrong values to the functions above (i.e: it set the ip address to 192.168.33.10 instead of 192.168.33.18. Seems that it loops over the list before start to provision. This wasn't happening with the previous version of vagrant (2.2.4)
Steps to reproduce
- vagrant up development
- see the log in the stdout
lzecca78
All 13 comments
Hi @lzecca78 - Do you have a minimal vagrantfile that reproduces this behavior? The one you shared is quite large, and there are a number of things that could go wrong or make a mistake with on the user side considering you are iterating over a ruby hash to generate Vagrant configs. It might also help you to debug if you print the name and configuration hash so you can see if it's properly using the right values.
briancain
on 9 Jul 2019
@briancain sorry, you are right. I made a slim one, basically, the behaviour described in the issue happens no matter what is the complexity of the Vagrantfile. Seem that something was changed in the iteration logic when a multi-vm are declared in the Vagrantfile. I added some puts with the access to the hash values in order to underline better the wrong behaviour. See how this output change from vagrant 2.2.4 to 2.2.5.
VAGRANTFILE_API_VERSION = '2'
ROOT_DIRECTORY = File.dirname(File.realpath(__FILE__))
Vagrant.require_version ">= 2.1.1"
Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
config.cache.scope = :box
config.cache.auto_detect = true
config.vm.boot_timeout = 600
{ :'test' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-test-1.public.#{ew_domain}",
:ip => '192.168.33.10',
:memory => '2048',
:cpus => 2,
:disk_size => '80GB',
:autostart => false,
:primary => false
},
:'test1' => {
:os => 'bento/ubuntu-18.04',
:hostname => "development-test-2.public.#{ew_domain}",
:ip => '192.168.33.11',
:memory => '2048',
:cpus => 2,
:autostart => false,
:primary => false
},
:'ws' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-ws-1.public.#{ew_domain}",
:ip => '192.168.33.12',
:memory => '2048',
:cpus => 1,
:autostart => false,
:primary => false
},
:'puppetserver' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-puppetserver-1.development.aws.#{ew_domain}",
:ip => '192.168.33.17',
:memory => '4192',
:cpus => 2,
:autostart => false,
:primary => false
},
:'development' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-all-vm.public.#{ew_domain}",
:ip => '192.168.33.18',
:memory => '2048',
:cpus => 2,
:disk_size => '80GB',
:autostart => false,
:primary => false
},
:'unit' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-unit-1.public.#{ew_domain}",
:ip => '192.168.33.19',
:memory => '1024',
:cpus => 1,
:autostart => false,
:primary => false
},
:'development_test' => {
:os => 'bento/ubuntu-18.04',
:hostname => "development-all-vm.public.#{ew_domain}",
:ip => '192.168.33.19',
:memory => '2048',
:cpus => 2,
:autostart => false,
:primary => false
},
:'development_test1' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-all-vm.public.#{ew_domain}",
:ip => '192.168.33.20',
:memory => '3192',
:cpus => 2,
:autostart => false,
:primary => false
},
:'cicd' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-cicd-vm.public.#{ew_domain}",
:ip => '192.168.33.19',
:memory => '2048',
:cpus => 2,
:autostart => false,
:primary => false
},
:'openvpn' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-openvpn-vm.public.#{ew_domain}",
:ip => '192.168.33.19',
:memory => '2048',
:cpus => 2,
:autostart => false,
:primary => false
},
:'kubemaster' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-kubemaster-1.public.#{ew_domain}",
:ip => '192.168.33.20',
:memory => '2048',
:cpus => 1,
:autostart => false,
:primary => false
},
:'kubenode' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-kubenode-1.public.#{ew_domain}",
:ip => '192.168.33.21',
:memory => '2048',
:cpus => 1,
:autostart => false,
:primary => false
},
:'kubenode2' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-kubenode-2.public.#{ew_domain}",
:ip => '192.168.33.22',
:memory => '2048',
:cpus => 1,
:autostart => false,
:primary => false
},
:'kibana' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-kibana-1.public.#{ew_domain}",
:ip => '192.168.33.21',
:memory => '2048',
:cpus => 1,
:autostart => false,
:primary => false
},
:'gitlab' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-gitlab-1.development.aws.#{ew_domain}",
:ip => '192.168.33.25',
:memory => '2048',
:cpus => 2,
:autostart => false,
:primary => false
},
:'redmine' => {
:os => 'bento/ubuntu-16.04',
:hostname => "development-redmine-1.development.aws.#{ew_domain}",
:ip => '192.168.33.26',
:memory => '2048',
:cpus => 2,
:autostart => false,
:primary => false
}
}.each do |name, configuration|
config.vm.define name ,
primary: configuration[:primary],
autostart: configuration[:autostart] do |instance|
if (configuration[:os].include?('16.04') || configuration[:os].include?('18.04'))
instance.cache.synced_folder_opts = {
owner: "_apt",
group: "_apt",
mount_options: ["dmode=777", "fmode=666"]
}
end
# change disk sizing if disk_size is declared
puts "ip is #{configuration[:ip]}"
puts "hostname is #{configuration[:hostname]}"
puts "name is #{name}"
if configuration.has_key?(:disk_size)
puts "disk size required is #{configuration[:disk_size]}"
config.disksize.size = configuration[:disk_size]
end
instance.vm.box = configuration[:os]
instance.vm.hostname = configuration[:hostname]
instance.vm.network 'private_network', ip: configuration[:ip]
# VirtualBox
instance.vm.provider 'virtualbox' do |vb|
# Boot in headless mode
vb.gui = false
# VM customization
vb.cpus = configuration[:cpus]
vb.customize ['modifyvm', :id, '--memory', configuration[:memory]]
vb.customize ['modifyvm', :id, '--natnet1', '192.168.199/24']
end
instance.vm.provision 'puppet' do |puppet|
puppet.environment = 'production'
puppet.environment_path = 'environments'
puppet.hiera_config_path = 'etc/hiera-vagrant.yaml'
puppet.module_path = ['environments/production/site',
'environments/production/modules',
'environments/production/custom_modules']
puppet.options = ['--verbose',
# '--debug',
# '--trace',
'--show_diff',
'--write-catalog-summary',
'--disable_warnings=deprecations',
'--pluginsync',
'--graph']
end
end
end
end
We seem to be seeing on 2.2.5 as well. I'm also on macOS Mojave.
Our simplified Vagrantfile:
Vagrant.configure("2") do |config|
config.vm.box = "ubuntu/trusty64"
config.vm.define "base" do |base|
end
config.vm.define "gui" do |gui|
config.vm.provider "virtualbox" do |vb|
vb.gui = true
end
end
end
On Vagrant 2.2.4, running vagrant up base would get you a normal headless VM. Now, on 2.2.5, vagrant up base opens up the VirtualBox GUI window.
chrisko
on 10 Jul 2019
@chrisko - that's because inside your "gui" guest you've defined a top scope config option using config. If you just wanted it to apply to the gui guest you would say
config.vm.define "gui" do |gui|
gui.vm.provider "virtualbox" do |vb|
vb.gui = true
end
end
So @lzecca78 - I wrote my own minimal vagrantfile...this is essentially what you are doing, right? It works as expected for me:
{:test1 => {:ip => "192.168.33.99"},
:anothertest => {:ip => "192.168.33.100"}
}.each do |name, conf|
config.vm.define name do |a|
a.vm.box = "bento/ubuntu-18.04"
a.vm.network "private_network", ip: conf[:ip]
a.vm.provision :shell, inline:<<-SHELL
ip addr
SHELL
end
end
Both guests get the right name to ip address as defined in the hash above...
Bringing machine 'test1' up with 'vmware_desktop' provider...
Bringing machine 'anothertest' up with 'vmware_desktop' provider...
==> test1: Cloning VMware VM: 'bento/ubuntu-18.04'. This can take some time...
==> test1: Checking if box 'bento/ubuntu-18.04' version '201906.18.0' is up to date...
==> test1: Verifying vmnet devices are healthy...
==> test1: Preparing network adapters...
==> test1: Starting the VMware VM...
==> test1: Waiting for the VM to receive an address...
==> test1: Forwarding ports...
test1: -- 22 => 2222
==> test1: Waiting for machine to boot. This may take a few minutes...
test1: SSH address: 127.0.0.1:2222
test1: SSH username: vagrant
test1: SSH auth method: private key
test1:
test1: Vagrant insecure key detected. Vagrant will automatically replace
test1: this with a newly generated keypair for better security.
test1:
test1: Inserting generated public key within guest...
test1: Removing insecure key from the guest if it's present...
test1: Key inserted! Disconnecting and reconnecting using new SSH key...
==> test1: Machine booted and ready!
==> test1: Configuring network adapters within the VM...
==> test1: Waiting for HGFS to become available...
==> test1: Enabling and configuring shared folders...
test1: -- /home/brian/code/vagrant-sandbox: /vagrant
==> test1: Running provisioner: shell...
test1: Running: inline script
test1: 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
test1: link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
test1: inet 127.0.0.1/8 scope host lo
test1: valid_lft forever preferred_lft forever
test1: inet6 ::1/128 scope host
test1: valid_lft forever preferred_lft forever
test1: 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
test1: link/ether 00:0c:29:ba:21:c0 brd ff:ff:ff:ff:ff:ff
test1: inet 172.16.183.216/24 brd 172.16.183.255 scope global dynamic eth0
test1: valid_lft 1793sec preferred_lft 1793sec
test1: inet6 fe80::20c:29ff:feba:21c0/64 scope link
test1: valid_lft forever preferred_lft forever
test1: 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
test1: link/ether 00:0c:29:ba:21:ca brd ff:ff:ff:ff:ff:ff
test1: inet 192.168.33.99/24 brd 192.168.33.255 scope global eth1
test1: valid_lft forever preferred_lft forever
test1: inet6 fe80::20c:29ff:feba:21ca/64 scope link
test1: valid_lft forever preferred_lft forever
==> anothertest: Cloning VMware VM: 'bento/ubuntu-18.04'. This can take some time...
==> anothertest: Checking if box 'bento/ubuntu-18.04' version '201906.18.0' is up to date...
==> anothertest: Verifying vmnet devices are healthy...
==> anothertest: Preparing network adapters...
==> anothertest: Fixed port collision for 22 => 2222. Now on port 2200.
==> anothertest: Starting the VMware VM...
==> anothertest: Waiting for the VM to receive an address...
==> anothertest: Forwarding ports...
anothertest: -- 22 => 2200
==> anothertest: Waiting for machine to boot. This may take a few minutes...
anothertest: SSH address: 127.0.0.1:2200
anothertest: SSH username: vagrant
anothertest: SSH auth method: private key
anothertest:
anothertest: Vagrant insecure key detected. Vagrant will automatically replace
anothertest: this with a newly generated keypair for better security.
anothertest:
anothertest: Inserting generated public key within guest...
anothertest: Removing insecure key from the guest if it's present...
anothertest: Key inserted! Disconnecting and reconnecting using new SSH key...
==> anothertest: Machine booted and ready!
==> anothertest: Configuring network adapters within the VM...
==> anothertest: Waiting for HGFS to become available...
==> anothertest: Enabling and configuring shared folders...
anothertest: -- /home/brian/code/vagrant-sandbox: /vagrant
==> anothertest: Running provisioner: shell...
anothertest: Running: inline script
anothertest: 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
anothertest: link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
anothertest: inet 127.0.0.1/8 scope host lo
anothertest: valid_lft forever preferred_lft forever
anothertest: inet6 ::1/128 scope host
anothertest: valid_lft forever preferred_lft forever
anothertest: 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
anothertest: link/ether 00:0c:29:07:fa:28 brd ff:ff:ff:ff:ff:ff
anothertest: inet 172.16.183.217/24 brd 172.16.183.255 scope global dynamic eth0
anothertest: valid_lft 1794sec preferred_lft 1794sec
anothertest: inet6 fe80::20c:29ff:fe07:fa28/64 scope link
anothertest: valid_lft forever preferred_lft forever
anothertest: 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
anothertest: link/ether 00:0c:29:07:fa:32 brd ff:ff:ff:ff:ff:ff
anothertest: inet 192.168.33.100/24 brd 192.168.33.255 scope global eth1
anothertest: valid_lft forever preferred_lft forever
anothertest: inet6 fe80::20c:29ff:fe07:fa32/64 scope link
anothertest: valid_lft forever preferred_lft forever
@briancain i try to go deeper: when you define an hash of hash of vms, each vm with custom parameters and you choose to start just only one of them (your choice must not match with the first element of the hash), the behaviour we are having upgrading from Vagrant 2.2.4 to 2.2.5 is that vagrant is using (_wrongly_) the first vm-hash to identifies parameter, instead of using the hash values from the key selected from the command vagrant up <vm-name> . The gist i paste in the issue is the debug log of the command vagrant up development. If you give a deep look inside that huge file, it should print only the hash values that match the label development but if you look it firstly print the following info:
ip is 192.168.33.10
hostname is 192.168.33.10
name is test
(ref: https://gist.github.com/lzecca78/c2019a65990da676285c6ba3a85e96c2#file-vagrant_debug_output-L156-L158)
only after some raws (and waiting pretty much time) below it prints the right information: https://gist.github.com/lzecca78/c2019a65990da676285c6ba3a85e96c2#file-vagrant_debug_output-L301-L304
but with Vagrant 2.2.4 and previous releases as well, this problem didn't exist at all.
The consequences of this, is not only the stdout that change twice, but is the following:
- before start the provision vagrant try twice to make the ssh connection (first the wrong first access hash, second to the right one)
- when i use the
configuration[:ip](that should use thedevelopmentip address) in the trigger that takes this parameter to do thenfs commandat the end of the provision, hangs because is using the one taken from the first wrong one : https://gist.github.com/lzecca78/c2019a65990da676285c6ba3a85e96c2#file-vagrant_debug_output-L1529-L1533 and this is why i end tocrl-cbecause it hangs forever, because of is using an ip related to a vm (in my case, the ip of the vmtest) that wasn't been created at all.
I can assure that the same identical Vagrantfile, never gave us this kind of behaviours with all the previous releases of vagrant so far.
Is pretty crystal clear that some behaviour has changed in the latest release, i didn't change this Vagrantfile since the release 2.1.1, the only change i did was to upgrade the vagrant release to 2.2.5. For this reason i opened this issue, because there is a problem that with the previous releases never happened.
I give you all the support you need to gather more informations , if needed, to go deeper in this problem, because i am 100% sure that something was changed and the behaviour changed with it.
lzecca78
on 11 Jul 2019
Hi @lzecca78 - Sorry, I think you might be confusing yourself with the puts statements. Those will get evaluated no matter what, even if you invoke a single guest. When Vagrant parses a Vagrantfile, it evaluates every line, similar to a ruby script. A Vagrantfile is essentially just a special DSL ruby program that Vagrant uses. So a ruby construct like puts works just like a ruby program, where as the DSL constructs for Vagrant like config options store state internally, since they aren't native ruby methods but something Vagrant defines.
Additionally, Vagrant doesn't do anything special when it comes to these ruby constructs. When you iterate over a hash, that's ruby itself, not Vagrant. This is the main reason why I think there might be something wrong with your Vagrantfile. Now, we did recently upgrade ruby to a new version, so perhaps in an older version of ruby, whatever you were doing worked, but now with the newer version of ruby, it works differently? That seems surprising, though.
In this case for example, I took my same minimal Vagrantfile and printed puts conf[:ip]. Although I only invoked the second guest defined in the hash, both ips in the hash were printed. Also if you notice, my vm still gets the proper ip when it gets brought up:
brian@localghost:vagrant-sandbox % be vagrant up anothertest
192.168.33.99
192.168.33.100
Bringing machine 'anothertest' up with 'vmware_desktop' provider...
==> anothertest: Checking if box 'bento/ubuntu-18.04' version '201906.18.0' is up to date...
==> anothertest: Cloning VMware VM: 'bento/ubuntu-18.04'. This can take some time...
==> anothertest: Checking if box 'bento/ubuntu-18.04' version '201906.18.0' is up to date...
==> anothertest: Verifying vmnet devices are healthy...
==> anothertest: Preparing network adapters...
==> anothertest: Starting the VMware VM...
==> anothertest: Waiting for the VM to receive an address...
==> anothertest: Forwarding ports...
anothertest: -- 22 => 2222
==> anothertest: Waiting for machine to boot. This may take a few minutes...
anothertest: SSH address: 127.0.0.1:2222
anothertest: SSH username: vagrant
anothertest: SSH auth method: private key
anothertest: Warning: Connection reset. Retrying...
anothertest:
anothertest: Vagrant insecure key detected. Vagrant will automatically replace
anothertest: this with a newly generated keypair for better security.
anothertest:
anothertest: Inserting generated public key within guest...
anothertest: Removing insecure key from the guest if it's present...
anothertest: Key inserted! Disconnecting and reconnecting using new SSH key...
==> anothertest: Machine booted and ready!
==> anothertest: Configuring network adapters within the VM...
==> anothertest: Waiting for HGFS to become available...
==> anothertest: Enabling and configuring shared folders...
anothertest: -- /home/brian/code/vagrant-sandbox: /vagrant
==> anothertest: Running provisioner: shell...
anothertest: Running: inline script
anothertest: 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
anothertest: link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
anothertest: inet 127.0.0.1/8 scope host lo
anothertest: valid_lft forever preferred_lft forever
anothertest: inet6 ::1/128 scope host
anothertest: valid_lft forever preferred_lft forever
anothertest: 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
anothertest: link/ether 00:0c:29:2a:84:95 brd ff:ff:ff:ff:ff:ff
anothertest: inet 172.16.183.223/24 brd 172.16.183.255 scope global dynamic eth0
anothertest: valid_lft 1794sec preferred_lft 1794sec
anothertest: inet6 fe80::20c:29ff:fe2a:8495/64 scope link
anothertest: valid_lft forever preferred_lft forever
anothertest: 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
anothertest: link/ether 00:0c:29:2a:84:9f brd ff:ff:ff:ff:ff:ff
anothertest: inet 192.168.33.100/24 brd 192.168.33.255 scope global eth1
anothertest: valid_lft forever preferred_lft forever
anothertest: inet6 fe80::20c:29ff:fe2a:849f/64 scope link
anothertest: valid_lft forever preferred_lft forever
Please try my minimal Vagrantfile and see if it works for you. It should be similar to yours, but smaller, agreed? They both have hashes that define key values, where the key is the guest name and the value is a config option for that given guest.
If my example is not like yours, then lets work to get them the same and see if we can figure out what is going on.
briancain
on 11 Jul 2019
Also, I looked at your Vagrantfile again. Are you using the vagrant-triggers plugin, or the native triggers feature in Vagrant core? If it's core, then this statement will not work like you think it might per my explanation above:
config.trigger.before :up, :provision do |trigger|
trigger.name = "upgrade or install librarian-puppet libs"
trigger.info = "upgrade or install librarian-puppet libs"
require 'fileutils'
FileUtils.mkdir_p './environments/production/modules'
trigger.on_error = :halt
trigger.run = {path: "bin/librarian-puppet.sh"}
end
The FileUtils call will always run every time Vagrant parses your Vagrantfile, and this is true for the rest of your trigger blocks too like this one:
config.trigger.after :provision,:up do |trigger|
trigger.only_on = "development"
trigger.ignore = :halt
trigger.name = "mount projects dir"
trigger.info = "mount projects dir"
retries = 0
max_retries = 10
begin
port_open?(configuration[:ip], 22)
rescue Exception => e
if retries <= max_retries
retries += 1
sleep 2 ** retries
retry
else
raise "Timeout: #{e.message}"
end
end
if ! os_router.already_mounted?
trigger.run = {inline: os_router.mount}
end
end
Something like that should be inside a ruby block: https://www.vagrantup.com/docs/triggers/usage.html#ruby-option
briancain
on 11 Jul 2019
thanks again @briancain ! I am using the trigger feature present in Vagrant core, yes. So as far as i understood, the Vagrantfile have the expect behaviour almost incidentally. I guess that something was change whit the new ruby version and so the problem that was incidentally hidden before, now is emerged. Is the only explanation i can give to all this. I will try to modifiy the Vagrantfile according to the documentation. Do you have some tips to _convert_ the triggers i have in the _right way_ ? Going back to your previous answer, i always used the ruby hash for declaring multi-vm Vagrantfile, because was easier than declare each time a block with the specs. Probably my assumptions to the behaviour were always wrong so far ( i used this kind of setup in Vagrant since 5-6 years more or less) and now this _ unstable balance _ was broken by the last ruby update i guess.
If you think that the best practice for the a multi-vm Vagrantfile is the ones written here : https://www.vagrantup.com/docs/multi-machine/ , i will convert everything i have in that format, although is more verbose, i will do that only once.
lzecca78
on 11 Jul 2019
@lzecca78 - I think ultimately your hash structure should be just fine. I agree, considering you have so many guests that use the same config options but different values, the hash structure is probably the right way to go here, so I would keep that as is. I think what was actually going wrong here is your trigger definitions, which are relatively new to Vagrant. As I was showing above with my example, the hash iteration should work as expected. :pray:
Now onto the trigger issue....I am hoping it isn't too much work. Based on the two code blocks I extracted from your Vagrantfile earlier, you would simply need to do something like:
config.trigger.after :provision,:up do |trigger|
trigger.only_on = "development"
trigger.ignore = :halt
trigger.name = "mount projects dir"
trigger.info = "mount projects dir"
trigger.ruby do |env,machine|
retries = 0
max_retries = 10
begin
port_open?(configuration[:ip], 22)
rescue Exception => e
if retries <= max_retries
retries += 1
sleep 2 ** retries
retry
else
raise "Timeout: #{e.message}"
end
end
if ! os_router.already_mounted?
trigger.run = {inline: os_router.mount}
end
end
end
config.trigger.before :up, :provision do |trigger|
trigger.name = "upgrade or install librarian-puppet libs"
trigger.info = "upgrade or install librarian-puppet libs"
trigger.ruby do |env,machine|
require 'fileutils'
FileUtils.mkdir_p './environments/production/modules'
end
trigger.on_error = :halt
trigger.run = {path: "bin/librarian-puppet.sh"}
end
I haven't actually tested that for correctness with your Vagrantfile fyi, but the idea generally is, any trigger that you wish to run ruby code with, needs to exist inside this option:
trigger.ruby do |env,machine|
puts "Your awesome ruby code goes here!"
end
Now that puts statement and any other ruby code will only be executed when the trigger is defined to run, rather than when it gets parsed. Here's my example from earlier, but with a ruby trigger:
{:test1 => {:ip => "192.168.33.99"},
:anothertest => {:ip => "192.168.33.100"}
}.each do |name, conf|
config.vm.define name do |a|
a.vm.box = "bento/ubuntu-18.04"
a.vm.network "private_network", ip: conf[:ip]
a.vm.provision :shell, inline:<<-SHELL
ip addr
SHELL
a.trigger.after :up do |t|
t.ruby do |env,machine|
puts "the ip: #{conf[:ip]}"
end
end
end
end
And the result....
brian@localghost:vagrant-sandbox % be vagrant up anothertest ±[●][master]
Bringing machine 'anothertest' up with 'vmware_desktop' provider...
==> anothertest: Cloning VMware VM: 'bento/ubuntu-18.04'. This can take some time...
==> anothertest: Checking if box 'bento/ubuntu-18.04' version '201906.18.0' is up to date...
==> anothertest: Verifying vmnet devices are healthy...
==> anothertest: Preparing network adapters...
==> anothertest: Starting the VMware VM...
==> anothertest: Waiting for the VM to receive an address...
==> anothertest: Forwarding ports...
anothertest: -- 22 => 2222
==> anothertest: Waiting for machine to boot. This may take a few minutes...
anothertest: SSH address: 127.0.0.1:2222
anothertest: SSH username: vagrant
anothertest: SSH auth method: private key
anothertest:
anothertest: Vagrant insecure key detected. Vagrant will automatically replace
anothertest: this with a newly generated keypair for better security.
anothertest:
anothertest: Inserting generated public key within guest...
anothertest: Removing insecure key from the guest if it's present...
anothertest: Key inserted! Disconnecting and reconnecting using new SSH key...
==> anothertest: Machine booted and ready!
==> anothertest: Configuring network adapters within the VM...
==> anothertest: Waiting for HGFS to become available...
==> anothertest: Enabling and configuring shared folders...
anothertest: -- /home/brian/code/vagrant-sandbox: /vagrant
==> anothertest: Running provisioner: shell...
anothertest: Running: inline script
anothertest: 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
anothertest: link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
anothertest: inet 127.0.0.1/8 scope host lo
anothertest: valid_lft forever preferred_lft forever
anothertest: inet6 ::1/128 scope host
anothertest: valid_lft forever preferred_lft forever
anothertest: 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
anothertest: link/ether 00:0c:29:5b:32:42 brd ff:ff:ff:ff:ff:ff
anothertest: inet 172.16.183.224/24 brd 172.16.183.255 scope global dynamic eth0
anothertest: valid_lft 1795sec preferred_lft 1795sec
anothertest: inet6 fe80::20c:29ff:fe5b:3242/64 scope link
anothertest: valid_lft forever preferred_lft forever
anothertest: 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
anothertest: link/ether 00:0c:29:5b:32:4c brd ff:ff:ff:ff:ff:ff
anothertest: inet 192.168.33.100/24 brd 192.168.33.255 scope global eth1
anothertest: valid_lft forever preferred_lft forever
anothertest: inet6 fe80::20c:29ff:fe5b:324c/64 scope link
anothertest: valid_lft forever preferred_lft forever
==> anothertest: Running action triggers after up ...
==> anothertest: Running trigger...
the ip: 192.168.33.100
You'll see at the end there that the ruby code is running with the correct ip address from the hash.
briancain
on 11 Jul 2019
@briancain i know we are going pretty OOT, but i am trying to convert all my triggers in the right way you told me, but i am facing a really weird behaviour :
this is more ore less the trigger:
config.trigger.after :provision, :up do |trigger|
trigger.only_on = "development"
trigger.ignore = :halt
trigger.name = "mount projects dir"
trigger.ruby do |env,machine|
if name == 'development'
puts "name is #{name}"
project_dir = '/opt/projects/Easywelfare'
os_router= CommandExecutor.new(personal_files_dir,configuration[:ip],project_dir)
retries = 0
max_retries = 10
begin
port_open?(configuration[:ip], 22)
rescue Exception => e
if retries <= max_retries
retries += 1
sleep 2 ** retries
retry
else
raise "Timeout: #{e.message}"
end
end
if ! os_router.already_mounted?
puts "mounting nfs"
system(os_router.mount)
end
end
end
end
analyzing it, what i was expecting was that it was triggered at the end, only for vm that is called development (because of the trigger.only_on), but what happens is that the trigger is executed n times where n is the number of the hashes in the Vagrantfile:
=> development: Notice: Applied catalog in 20.54 seconds
==> development: Configuring cache buckets...
==> development: Running action triggers after up ...
==> development: Running trigger: mount projects dir...
ip is 192.168.33.10
name is test
==> development: Running trigger: mount projects dir...
ip is 192.168.33.11
name is test1
==> development: Running trigger: mount projects dir...
ip is 192.168.33.12
name is ws
==> development: Running trigger: mount projects dir...
ip is 192.168.33.14
name is mail
==> development: Running trigger: mount projects dir...
ip is 192.168.33.16
name is monitoring
==> development: Running trigger: mount projects dir...
ip is 192.168.33.17
name is puppetserver
==> development: Running trigger: mount projects dir...
ip is 192.168.33.18
name is development
==> development: Running trigger: mount projects dir...
ip is 192.168.33.19
name is unit
==> development: Running trigger: mount projects dir...
ip is 192.168.33.19
name is development_test
==> development: Running trigger: mount projects dir...
ip is 192.168.33.20
name is development_test1
==> development: Running trigger: mount projects dir...
I dunno why, but is the current behaviour. Is there something wrong?
lzecca78
on 12 Jul 2019
Where are you printing the ip is ___? I don't see that in the code you shared. Also if I remember right that trigger was inside a guest definition, right? Try defining it as a guest scoped trigger, rather than a global config trigger. So for any given trigger defined inside the guest, replace config with instance, since I believe that's what the local guest config option is called in your Vagrantfile.
briancain
on 12 Jul 2019
@briancain thank you a lot! that's was the main issue, once i replaced config with instance everything works as expected! Really thanks for the support even if wasn't related at all with the opened issue. I guess now is possible to close the issue, because basically the issue emerged with the 2.2.5 but was not related with this version, but with a misconfiguration of the Vagrantfile. Thanks a lot again!
lzecca78
on 13 Jul 2019
I'm going to lock this issue because it has been closed for _30 days_ ⏳. This helps our maintainers find and focus on the active issues.
If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.
![hashibot[bot] picture](https://avatars.githubusercontent.com/in/8332?v=4&s=40)
hashibot[bot]
on 28 Jan 2020
Related issues
StefanScherer
·
3Comments
bbaassssiiee
·
3Comments
lebogan
·
3Comments
mpontillo
·
3Comments
hesco
·
3Comments
Most helpful comment
@briancain thank you a lot! that's was the main issue, once i replaced
configwithinstanceeverything works as expected! Really thanks for the support even if wasn't related at all with the opened issue. I guess now is possible to close the issue, because basically the issue emerged with the 2.2.5 but was not related with this version, but with a misconfiguration of the Vagrantfile. Thanks a lot again!