I want puppet to look for hiera.yaml in /etc but it's looking for it in /etc/puppet. I put a line into puppet.conf:
hiera_config = /etc/hiera.yaml
But still gives me the hiera.yaml update warning when I run the script.
I'm running the script from Vagrant 1.2.2. Using puppet 3.2.2
I'm running Centos 6.4 in a vm.
I found that the puppet provisioner in vagrant now support hiera_config_path which does exactly what is desired.
config.vm.provision :puppet do |puppet|
# path on host machine to hiera.yaml
puppet.hiera_config_path = '/Users/me/vms/hiera/hiera.yaml'
# This sets the relative path for hiera data directories
puppet.working_directory = '/Users/me/vms/hiera'
end
This is documented in Vagrant: Up and Running but I didn't find it until I started looking into the vagrant source to implement this feature myself.
Hmmm... On Vagrant 1.2.2 and Puppet 3.2.3, I am able to set hiera_config in puppet.conf without problems. I would double-check that you are editing /etc/puppet.conf on the Vagrant vm, not on the host machine, and that the hiera_config line is the [main] block, not just in the [master] block.
If both of those conditions are true and it is still not working, you might try explicitly setting hiera_config in your Vagrantfile:
config.vm.provision :puppet do |puppet|
...
puppet.options = '--hiera_config=/etc/hiera.yaml'
end
Good luck!
Puppet provisioning runs as root user, not vagrant, so that's why it doesn't take notice of your puppet.conf in /vagrant.
If you run puppet config print inside the vm from user vagrant and root you see ALL puppet config settings per user and compare.
Related
I am new to Vagrant but good in Docker.
In Vagrant I am aware of the fact that
config.vm.provision :shell,path: "bootstrap.sh", run: 'always'
in the Vagrantfile will provision vagrant box while doing vagrant up. With this, the vagrant box interactive console appears after the intended provisioning is done.
But I need to configure in such a way that, first the control goes in to vagrant box console and then the intended script is up and running. Because my requirement is to run a script automatically post vagrant up and not to run a bootstrapped script.
In analogy with Docker, my question can be seen as
what is the Vagrant equivalent for CMD in Dockerfile ?
You can look at vagrant triggers. You can run dedicated script/command after each specific vagrant command (up, destroy ...)
For example
Vagrant.configure("2") do |config|
# Your existing Vagrant configuration
...
# start apache on the guest after the guest starts
config.trigger.after :up do |trigger|
trigger.run_remote = {inline: "service apache2 start"}
end
end
I have a production server architecture that is quite complex that involves several types and tiers of servers including: Web, Varnish, HAProxy, ELK, Database, Sensu, etc. The production architecture is provisioned with puppet masters and the puppet code for the masters is provisioned with Git through many, submodules.
I want to connect these Git repos to VM's locally to further develop pieces of the puppet architecture. However each Vagrant VM requires a complete self contained copy of all of the Git repos in order to stand up just one VM. This seems like a very inefficient use of drive space for local development on what will be potentially a dozen different types of servers.
Is there a way to point all of the Vagrant VM's VagrantFiles to a common local folder outside of the vagrant directory such that each Vagrant instance can still read the folder and provision the server?
Edited --
Based on comment from #Treminio, here is the portion of my VagrantFile showing the attempt to declare an absolute path from host computers root:
config.vm.provision "puppet" do |puppet|
puppet.manifest_file = "init.pp"
puppet.manifests_path = "/Users/jdugger/vm/puppet/manifests"
puppet.module_path = "/Users/jdugger/vm/puppet/modules"
puppet.hiera_config_path = "/Users/jdugger/vm/puppet/hieradata"
end
... and the error response by Vagrant:
==> default: Configuring and enabling network interfaces...
==> default: Mounting shared folders...
default: /vagrant => /Users/jdugger/vm/pupt/vagrant
default: /vagrant_data/scripts => /Users/jdugger/scripts
default: /tmp/vagrant-puppet/modules-169f1d27ef31a534405e2e9fcde2eedf => /Users/jdugger/vm/puppet/modules
default: /tmp/vagrant-puppet/manifests-be5a69bfb646cf9329b8921f221ffab8 => /Users/jdugger/vm/puppet/manifests
==> default: Running provisioner: puppet...
The `puppet` binary appears not to be in the PATH of the guest. This
could be because the PATH is not properly setup or perhaps Puppet is not
installed on this guest. Puppet provisioning can not continue without
Puppet properly installed.
This response may not be because of the puppet path - not sure - It appears that puppet is not on the guest box (isn't it supposed to be running from Vagrant?). No version comes up when with:
[vagrant#localhost ~]$ puppet --version
Response is:
-bash: puppet: command not found
Update ---
#Treminio is correct. I have been able to provision with puppet manifests and modules external to the vagrant/ directory. The Path problem appears to be because Puppet is not installed on the guest VM. To resolve this I added a shell script found here:
http://garylarizza.com/blog/2013/02/01/repeatable-puppet-development-with-vagrant/
This was added just before the puppet provisioning declaration. Just as a note there doesn't seem to be a lot of advanced examples that demonstrate the external file capability or that you need to install puppet outside of the puppet provisioner.
However each Vagrant VM requires a complete self contained copy of all of the Git repos in order to stand up just one VM.
This is incorrect.
What you want to change is the puppet.module_path value.
It can be any location on your host's disk, and Vagrant will automatically mount it inside your VM.
I'm trying to provision a Vagrant instance with Chef by connecting to a Chef server.
My Vagrantfile looks similar to this:
Vagrant.configure("2") do |config|
config.vm.box = "opentable/win-2012r2-standard-amd64-nocm"
config.vm.provider = "virtualbox"
config.vm.communicator = "winrm"
config.vm.guest = :windows
config.omnibus.chef_version = :latest
config.berkshelf.enabled = true
config.vm.provision :chef_client do |chef|
chef.chef_server_url = "https://chef.website.com"
chef.validation_key_path = "C:/chef/validation.pem"
chef.run_list = ["recipe[cookbook::default]"]
end
end
According to this documentation, this Vagrantfile should contain all the fields I need to connect to the Chef server and pull down the relevant cookbooks and databags.
However, I am running into the following error when running the Chef client provisioner:
The following berks command failed to execute:
C:\opscode\chefdk\embedded\bin/berks.BAT upload --berksfile C:/path/to/cookbook/Berksfile --no-freeze --force
The stdout and stderr are shown below:
stdout: E, [2015-08-10T17:02:57.654352 #8488] ERROR -- : Ridley::Errors::ClientKeyNotFoundOrInvalid: client key is not found or invalid or not found at: 'C:/chef/client.pem'
Indeed, the client.pem is not there.
While the error itself is pretty self explanatory, I don't understand why I need to specify a client.pem on the initial Chef client run.
Is there a way for Vagrant to create this itself? Better yet, can I make it so that it doesn't need a client identifier at all?
This is only a Vagrant instance, so I don't need to keep this node on the Chef server. Based on the Chef client provisioner documentation I don't see why I need a client.pem file, as the intructions make no mention of this.
Thanks for your help!
Try disabling Berkshelf
config.berkshelf.enabled = false
Bootstrapping a new Chef client against a chef server does not require Berkshelf. Berkshelf is a tool used on Chef workstations to manage cookbooks on the chef server.
PS
The Vagrant-berkshelf plugin at one stage was deprecated in favour of Test kitchen. It appears to be back again, but in the meantime I've grown to appreciate the benefits of Test Driven Infrastructure!
I have a vagrantfile using a box on top of virtualbox with a provision script.
Now I am trying to use packer to output a box already after provision.
However I cannot find a builder to use the ".box" file I already have. What am I doing wrong?
I just got a solution to this tiny little problem (convert a vagrant .box file to .ova for use by packer):
Create a vm using the .box file as a base. I use this Vagrantfile, with box opscode-centos-7.0:
$provisioning_script = <<PROVISIONING_SCRIPT
adduser packer
echo "packer" | passwd packer --stdin
echo "packer ALL=(ALL:ALL) NOPASSWD: ALL" > /etc/sudoers.d/packer
PROVISIONING_SCRIPT
Vagrant.configure(2) do |config|
config.vm.box = "opscode-centos-7.0"
config.ssh.insert_key = false
config.vm.provider "virtualbox" do |v|
v.name = "packer-base"
end
config.vm.provision :shell, inline: $provisioning_script
end
run vagrant up
run vagrant halt
run vboxmanage export --ovf20 -o packer-base.ova packer-base
run vagrant destroy
This also creates the packer user with a default password so that packer can easily connect to the instance to do stuff. Also note the insert_key parameter that will prevent replacing the vagrant default insecure key with a secure one and allow subsequent vagrant setups to properly connect via SSH to the new images (after packer is done).
Packer out-of-the-box doesn't support using Vagrant boxes as input (yet).
But there is a custom plugin, see this comment.
If you want to build a vagrant box that runs with provider virtualbox, have a look here.
However, it takes an iso or ovf as input, not a vagrant box.
Have a look at these templates to get you started using the virtualbox builder with packer.
Make sure your run the post-processor to convert the virtualbox vm into a vagrant box.
Hi i was trying to understand what exactly is done when running
vagrant up
my reason for that is that in my case we need to install a lot of utilities.
i.e Version control tools, build tools, ide, etc...
which takes a lot of time.
so actually i wanted a 'box' with all those tools.
After i have clean environment and got all tools, i would like to make CI for our product.
If i will reinstall all utilities it should take a lot of time. so what i am actually need is just installing and testing our product.
How should i handle that ??
create my own box? does the command reinstall all utilities when we make CI ??
what i actually need are 2 processes :
1.installing utilities for my vm. (once a month)
2.test our product (each commit\push to version control)
how can i achieve that ?
For the first time, vagrant up will create a new VM for you, pulling the box image if needed, and it will provision it with what you configured in the Vagrantfile. In the provision configuration, you can tell Chef or Puppet to install all the utilities and tools that you need.
When you suspend or halt the VM, the next time you do a vagrant up it will only bring that VM back up. It will not install or try to provision it again.
You can force it with vagrant up --provision or just vagrant provision.
This usually works well in a development environment.
In a CI environment, it may not be possible to have the VM already provision, forcing you to run the provisioning step every time. You can achieve what you need packaging your own box with the tools already installed, essentially creating a golden or base image.
Just be extra careful so that the CI environment don't differ for what you have in production.
All depend on the setting in Vagrantfile
you have modules folder to put all puppet modules, manifests folder with site.pp and Vagrantfile as below under same place.
Give you a sample of Vagrantfile I used mostly.
VAGRANTFILE_API_VERSION = "2"
Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
config.vm.box = "precise64"
config.vm.box_url = "https://cloud-images.ubuntu.com/vagrant/precise/current/precise-server-cloudimg-amd64-vagrant-disk1.box "
config.vm.provision :puppet do |puppet|
puppet.module_path = "modules"
puppet.manifests_path = "manifests"
puppet.manifest_file = "site.pp"
end
config.vm.define :www do |config|
config.vm.host_name = "www.example.com"
config.vm.network :private_network, ip: "192.168.1.2"
end
config.vm.define :db1 do |config|
config.vm.host_name = "db1.example.com"
config.vm.network :private_network, ip: "192.168.1.4"
end
end
So after run vagrant up www or vagrant up db1 it will start the box with puppet apply directly. You can understand this way as puppet masterless
Vagrant just make the box up running from linux image defined in config.vm.box_url as a simple fresh linux box, and mount your local folder to new box's /vagrant folder, then hand over to puppet. How the server to be provisioned, will be depended on site.pp (define applications on each node) and puppet modules. The command is similar as
puppet apply --modulepath /vagrant/module /vagrant/manifests/site.pp
So if your puppet modules are fine, your new box will automatically have all utilities and products installed. Then you run vagrant ssh www or vagrant ssh db1, you can login it and start working.
You can put your local folder with moduels, manifests folders and Vagrantfile to version control (such as git). So developers can clone the git repository to their own computer easily.