vagrant/website/www/source/blog/2014-02-24-feature-preview-vagrant-1-5-share.html.markdown at d6b7f3a855092292fef6eb48e07f65f6afb9a545

9.5 KiB

Raw Blame History

page_title	title	author	author_url
Vagrant 1.5 Feature Preview: Vagrant Share	Feature Preview: Vagrant Share	Jack Pearkes	https://github.com/pearkes

A primary goal of Vagrant is not only to provide easy-to-use development environments, but also to make it easy to share and collaborate on these environments.

With Vagrant 1.5, we're introducing a feature that will allow you to share your running Vagrant environment with anyone, on any network connected to the internet. We're calling this feature 'Vagrant Share.'

This feature lets you share a link to your web server to a teammate across the country, or just across the office. It'll feel like they're accessing a normal website, but actually they'll be talking directly to your running Vagrant environment. They'll be able to see any changes you make, as you make them, in real time.

With Vagrant Share, others can not only access your web server, they can access your Vagrant environment like it was any other machine on a local network. They can have access to any and every port.

Read on for a demo and more details.

Demo

Before we get into details about Vagrant share, let's show a few demos. You may need to go fullscreen to read the text.

Sharing an HTTP server:

Sharing SSH access:

Sharing a static IP with Vagrant Connect:

The feature we call "Vagrant Share" introduces two new Vagrant commands: vagrant share and vagrant connect.

The share command is used to share a running Vagrant environment, and the connect command compliments it by accessing any shared environment. Note that if you're just sharing HTTP access, the accessing party does not need Vagrant installed. This is covered later.

We'll cover the details of each command next.

By default, Vagrant Share shares HTTP access to your Vagrant environment to anyone in the world. The URL that it creates is publicly accessible and doesn't require Vagrant to be installed to access -- just a web browser.

$ vagrant share
==> default: Local HTTP port: 5000
    default: Local HTTPS port: disabled
==> default: Your Vagrant Share is running!
==> default: URL: http://frosty-weasel-0857.vagrantshare.com
...

Once the share is created, a relatively obscure URL is outputted. This URL will route directly to your Vagrant environment; it doesn't matter if you or accessing party is behing a firewall or NAT.

Currently, HTTP access is restricted through obscure URLs. We'll be adding more ACLs and audit logs for this in the future.

SSH Access

While sharing your local webserver is a powerful collaboration tool, Vagrant Share doesn't stop there. With just a single flag, Vagrant Share can allow anyone to easily SSH into your Vagrant environment.

Perhaps you're having issues where your app isn't running properly or you just want to pair program. Now, with just one flag, anyone you want can SSH into your Vagrant environment from anywhere in the world.

SSH access isn't shared by default. To enable sharing SSH, you must add the --ssh flag to vagrant share:

$ vagrant share --ssh
==> default: SSH Port: 22
==> default: Generating new SSH key...
    default: Please enter a password to encrypt the key:
    default: Repeat the password to confirm:
    default: Inserting generated SSH key into machine...
==> default: Checking authentication and authorization...
==> default: Creating Vagrant Share session...
    default: Share will be at: awful-squirrel-9454
==> default: Your Vagrant Share is running!
...

When the --ssh flag is provided, Vagrant generates a brand new SSH keypair for SSH access. The public key portion is automatically inserted into the Vagrant environment. The private key portion is uploaded to the server managing the Vagrant Share connections. The password used to encrypt the private key is not uploaded anywhere, however, meaning we couldn't access your VM if we wanted to. It is an extra layer of security.

Once SSH access is shared, the person wanting to access your Vagrant environment uses vagrant connect to SSH in:

$ vagrant connect --ssh awful-squirrel-9454
Loading share 'awful-squirrel-9454'...
Password for the private key:
Executing SSH...

Welcome to Ubuntu 12.04.1 LTS

 * Documentation:  https://help.ubuntu.com/
Last login: Wed Feb 26 08:38:55 2014 from 192.168.148.1
vagrant@precise64:/vagrant$

The name of the share and the password used to encrypt the private key must be communicated to the other person manually, as a security measure.

Vagrant Connect

Vagrant share can share any TCP/UDP connection, and is not restricted to only a single port. When you run vagrant share, Vagrant will share the entire Vagrant environment.

When the person you are sharing with runs vagrant connect SHARE-NAME, Vagrant will give this person a static IP they can use to access the machine as if it were on the local network:

$ vagrant connect awful-squirrel-9454
==> connect: Connecting to: awful-squirrel-9454
==> connect: Starting a VM for a static connect IP.
    connect: The machine is booted and ready!
==> connect: Connect is running!
==> connect: SOCKS address: 127.0.0.1:62167
==> connect: Machine IP: 172.16.0.2
==> connect:
==> connect: Press Ctrl-C to stop connection.
...

Security Concerns

Sharing your Vagrant environment understandably raises a number of security issues.

With the launch of Vagrant 1.5, the primary security mechanism for Vagrant Share is security through obscurity along with an encryption key for SSH. Additionally, there are several configuration options made available to help control access and manage security:

--disable-http will not create a publicly accessible HTTP URL. When this is set, the only way to access the share is with vagrant connect.
--ssh-once will allow only one person to SSH into your shared environment. After the first SSH access, the keypair is physically deleted and SSH access won't be possible anymore.

In addition to these options, there are other features we've built to help:

Vagrant share uses end-to-end TLS connections. So even unencrypted TCP streams are encrypted through the various proxies and only unencrypted during the final local communication between the local proxy and the Vagrant environment.
SSH keys are encrypted by default, using a password that is not transmitted to our servers or across the network at all.
SSH is not shared by default, it must explicitly be shared with the --ssh flag.
A web interface we've built shows share history and will show basic access logs in the future.
Share sessions expire after a short time (currently 1 hour), but can also be expired manually by ctrl-c from the sharing machine or via the web interface.

Most importantly, you must understand that by running vagrant share, you are making your Vagrant environment accessible by anyone who knows the share name. When share is not running, it is not accessible.

And, after Vagrant 1.5 is released, we will be expanding the security of this feature by adding ACLs, so you're able to explicitly allow access to your share based on who is connecting.

For maximum security, we will allow you to run your own Vagrant Share server. We won't be launching this right with Vagrant 1.5, but it will be an option shortly after that.

Technical Details

We've been demoing Vagrant Share around the world over the past month or so. The response has been overwhelmingly positive, but the first reaction from everyone is always: "How does this work?" In this section, we'll briefly cover some technical details of the feature.

There are a lot of moving parts that make Vagrant Share work. Here is an overview of the primary components:

Local Proxy - This runs on the share host machine (not within the Vagrant environment). It connects to the remote proxy and proxies traffic to and from the Vagrant environment and the remote proxy. It is also responsible for registering new shares with the remote proxy.
Remote Proxy - This runs on a remote server on the internet. It creates shares and is connected to local proxies. It also handles all ACLs, security audit logs, SSH keys, and more.
Connect Proxy VM - When vagrant connect is called, Vagrant runs a very small proxy virtual machine (13 MB RAM-only!). This virtual machine exposes the static IP that the connecting person uses to access the share. Any traffic sent to this IP is routed to the remote proxy, which in turn routes down to the local proxy and the shared Vagrant environment.

The connection from the connect proxy to the remote proxy uses the standard SOCKS5 protocol. The connection between the remote proxy and the local proxy uses a modified variant to reduce the number of packets that must be sent for any given connection.

What's Next?

Vagrant Share will ship with Vagrant 1.5. To use it, you'll need an account in the yet to be announced web service.

At that time, we'll publish further details about share, connect and the account required to use them.

Next week, we'll cover another feature of Vagrant 1.5 — stay tuned.

9.5 KiB Raw Blame History