Skytap Provider for Vagrant (Beta)

The Skytap Vagrant provider is a Vagrant plugin for creating, provisioning, and controlling VMs on the Skytap cloud computing platform. It allows you to:

  • Create multi-VM environments using source VMs from one or more Skytap templates
  • SSH into the instances
  • Customize hardware settings via the Vagrantfile
  • Sync folders between your local machine and Skytap VMs via NFS

NOTE: This plugin requires Vagrant 1.2+ and Ruby 2.0 or greater.

Concepts

Skytap environments map neatly onto Vagrant multi-machine environments. An environment contains one or more VMs, and may also contain networks for the VMs to connect to. Environments may be snapshotted as templates, which can then be used to create new environments. The Skytap public template library is a collection of templates containing a variety of pre-configured VMs.

Before You Begin

Before you begin, make sure you have:

  • Ruby 2.0 or higher installed on your local machine
  • The latest version of Vagrant installed on your local machine (available from https://www.vagrantup.com/)
  • A Skytap username and API token from the "My Account" page

  • A Skytap VPN in the region where you'll be creating environments; a NAT-enabled VPN is recommended.

    To check if a VPN is available, navigate to a Skytap environment in the region and open the network settings. If the VPN section is visible in the network settings, a VPN is available. If you do not have a Skytap VPN, work with your Skytap administrator to create one. For instructions, see Creating a VPN Connection to an External Network.

Installing the Skytap Provider and Starting Your First Environment

  1. Ensure that your local machine is on one of your Skytap VPN's remote subnets.
  2. To install the provider, type the following at the command line: vagrant plugin install vagrant-skytap
  3. Create a new directory.

  4. Create a file called Vagrantfile (with no file extension) containing the following. This Vagrantfile describes a Skytap environment containing a single VM, using the source VM indicated by the vm_url setting (a generic Ubuntu 14.04 server in the US-West region) and upgrading it to 2 CPUs.

    Vagrant.configure(2) do |config|
  config.vm.box = "skytap/empty"

  config.vm.provider :skytap do |skytap, override|
    skytap.username = "<username>"
    skytap.api_token = "<api_token>"
  end

  config.vm.define "web" do |server|
    server.vm.provider :skytap do |box|
      box.vm_url = "https://cloud.skytap.com/vms/3157858"
      box.cpus = 2
    end
  end
end

  5. Update the username and api_token settings and save the file. If you don't want to store your username and API token in the Vagrantfile, you can set them in the environment variables VAGRANT_SKYTAP_USERNAME and VAGRANT_SKYTAP_API_TOKEN.

  6. Navigate to the directory containing the Vagrantfile and enter the following at the command line: vagrant up --provider skytap Vagrant will create a new Skytap environment containing the VM,.

  7. When prompted by Vagrant, select the VPN for the region you want to connect to.

  8. Choose "skytap" as the user login for the VM.

  9. Wait for vagrant up to complete, then do vagrant ssh to verify that you can access the new VM.

Multi-machine Example

The following defines two VMs in a single environment. Both are based on the same Ubuntu template as above, but have different hardware settings. Since the source VM in the public library template is connected to a network, both of the VMs in the new environment will be connected to a single network.

config.vm.define "web" do |server|
  server.vm.provider :skytap do |box|
    box.vm_url = "https://cloud.skytap.com/vms/3157858"
    box.cpus = 2
    box.cpuspersocket = 1
    box.ram = 1024
  end
  server.vm.synced_folder "~/web_files", "/synced", type: :nfs
end

config.vm.define "db" do |server|
  server.vm.provider :skytap do |box|
    box.vm_url = "https://cloud.skytap.com/vms/3157858"
    box.cpus = 8
    box.cpuspersocket = 4
    box.ram = 8192
  end
  server.vm.synced_folder "~/db_files", "/synced", type: :nfs
end

Supported Commands

For the most part these behave identically to the builtin Vagrant commands.

Vagrant Command Skytap Action
vagrant destroy [<vm_name>, <vm_name>] Delete an environment or VM(s)
vagrant global-status Show the status of all Vagrant-managed VMs on the host machine; this includes VMs from other Vagrant providers
vagrant halt [<vm_name>, <vm_name>] Shut down an environment or VM(s). Any VMs which do not shut down gracefully will be powered off.
vagrant halt [<vm_name>, <vm_name>] --force Power off an environment or VM(s) without performing a graceful shutdown
vagrant help Display the standard help information
vagrant reload [<vm_name>, <vm_name>] Shut down and then run an environment or VM(s); this is equivalent to vagrant halt followed by vagrant up.
vagrant resume [<vm_name>, <vm_name>] Runs one or more suspended VM(s)
vagrant ssh [<vm_name>] Begin an SSH session with a VM
vagrant ssh-config [<vm_name>, <vm_name>] Generate an OpenSSH configuration file based on the VM settings
vagrant status [<vm_name>, <vm_name>] Show the runstate of one or more VM(s)
vagrant suspend [<vm_name>, <vm_name>] Suspend an environment or VM(s)
vagrant up [<vm_name>, <vm_name>] Run an environment or VM(s), creating them from settings in the Vagrantfile if they do not already exist.

Notes:

  • When the first VM is created, a Skytap environment will be created; when all VMs are deleted, the containing environment will also be deleted.
  • The timeout for graceful shutdown is currently set to 5 minutes.
  • Changes to hardware settings of an existing VM will take effect when the VM is being powered on; that is, when doing vagrant reload, or vagrant up when the machine is halted.

Additional Supported Actions

Sync Local Folders with the VM's Folders using NFS

The Skytap Vagrant provider supports Vagrant's built-in NFS sharing facility. In the following example, a local directory ~/web_files will be visible on the VM at the path /synced.

config.vm.define "web" do |server|
  server.vm.provider :skytap do |box|
    box.vm_url = "https://cloud.skytap.com/vms/3157858"
    # ...
  end
  server.vm.synced_folder "~/web_files", "/synced", type: :nfs
end

For more information, see https://docs.vagrantup.com/v2/synced-folders.

Skytap-specific Vagrantfile Settings

Setting Required? Description
vm_url yes The URL of the source VM to use when creating a new VM.
cpus no Number of CPUs (more specifically, the number of virtual cores).
cpuspersocket no Number of virtual cores per processor.
ram no RAM (megabytes).
guestos no The VMware guest OS for the virtual machine.
vpn_url no The URL of the Skytap VPN to use when connecting to the VM.

Notes:

  • Source VMs must come from a template, not an environment, and they must be saved in the powered off state.
  • Multi-machine environments may use source VMs from multiple templates, from your customer account and/or the public template library, as long as all are in the same region. Your user account must have permissions to see the templates containing the source VMs.
  • cpus must be evenly divisible by cpuspersocket. E.g., two quad-core processors have a total of 8 virtual cores, so the cpus value would be 8. (Most VMs in the public template library are single-core.)
  • The guestos setting is distinct from from Vagrant's config.vm.guest setting.
  • If you do not know the URL for the VPN you wish to use, contact your Skytap administrator.

Login Credentials

In addition to setting username and password in the Vagrantfile with config.ssh.username and config.ssh.password, the Skytap Vagrant provider also supports VM Credentials stored with the Skytap VM. Credentials are a free-form field; if formatted as "username / password", the Skytap provider will parse the credentials and present them to the user when the VM is first created.

NOTE: Regardless of how the login is obtained, it will be stored in cleartext in the environment's data directory (.vagrant).

Troubleshooting and Known Issues

To enable logging while troubleshooting, see https://docs.vagrantup.com/v2/other/debugging.html. When reporting issues with the Skytap Vagrant provider, please include the output when using VAGRANT_LOG=debug . NOTE: make sure to edit out your API token before sending or posting the log output!

Known issues

  • Vagrant must be able to connect to the new VM over the selected Skytap VPN.
  • The source VM must have an SSH service configured to run on startup, or (for Windows VMs) be configured for WinRM access. For more information about WinRM configuration, see https://docs.vagrantup.com/v2/boxes/base.html, under "Windows Boxes".
  • At this time, WinRM credentials stored in Skytap VMs will be ignored. The username and password for WinRM connections must be stored in the Vagrantfile (config.winrm.username and config.winrm.password).
  • Running, reloading, or destroying a Skytap VM can result in "stale NFS file handle" errors on other providers' VMs. This is a known issue when using multiple providers on the same host machine. The workaround is to use vagrant reload on the affected VM to refresh that VM's NFS mount(s).
  • At this time, vagrant share is not supported.
  • High-level networking features (including port forwarding and private networks) are currently unsupported.
  • Although several Skytap public library VMs include credentials for the root login, its use is not recommended.
  • Once a VM has been created, it is not possible to change its connection method (i.e., connect using a different VPN). This is the case whether the VPN is specified in the Vagrantfile with the vpn_url setting, or selected interactively.