halvm-web

This repository contains a simple (for now) web server for the HaLVM. By compiling this web server and combining it with a tar file containing your website, you can serve simple static websites from your own Xen server or Amazon's EC2. Eventually, we hope this web server expands to support dynamically-generated functions and other cool features. See the issue list for more information and interesting things to work on.

Quick Start

The easiest way to build halvm-web is to use the HaLVM Docker tools. In this case, perform the following steps:

1. git clone https://github.com/GaloisInc/halvm-web
2. docker run -v ${PWD}:/halvm halvm/extended-gmp halvm-cabal sandbox init
3. docker run -v ${PWD}:/halvm halvm/extended-gmp halvm-cabal install --package-db=/usr/lib64/HaLVM-2.1.1/package.conf.d

[sorry for the weird argument there. We're working on it.]

These three steps will create a .cabal-sandbox/bin/halvm-web binary on your local disk, which contains the halvm-web unikernel. Now its up to you to build your site! So go do that. Perhaps this is the time for you to try Hakyll, to double down on the Haskell nerdiness.

When you're done, create a new directory called site, and copy all your files into that directory. So, for example, your mainline index file should be site/index.html.

Now ask yourself: Where do I want to run this?

Option #1: I want to run this right here!

If you happen to be running on a host running Xen, you should be able to start up your web server using the handy run.sh provided with this repository. It'll prep your tar file for you, kill off any old web servers running, and start yours. You'll see it boot and find an IP address, which you can then play around with as you see fit.

Option #2: I want to run this IN THE CLOUD!

In this case, we have two steps to follow. The first is simple, the second very long. Also, I'm going to assume that you have the appropriate AWS credentials stored in AWS_ACCESS_KEY and AWS_SECRET_KEY, as per normal.

1. tar cvf site.tar site/
2. docker run -v ${PWD}:/halvm halvm/extended ec2-unikernel -o ${AWS_ACCESS_KEY} -w ${AWS_SECRET_KEY} .cabal-sandbox/bin/halvm-web site.tar

This will print a lot of status updates, but at the end of this process, it will give you an AMI reference. Go to your AWS console, and you can launch it as normal; just make sure to give it a network card, and make sure that the security group you assign it allows for access to port 80.

Option #3: I want to run this on another Xen machine!

That's cool ... but I'm afraid we haven't really optimized this case. But the core idea is the same. You'll first need to build the halvm-web binary and transfer it to your target machine. You can do this with Docker, as above. Then you'll need to build your site as a static, uncompressed tarball. Please make sure that your site lives under the site/ path! In other words, your site index should be site/index.html in the tarball. Once you've got this working, transfer it to the target machine, as well.

Finally, the file halvm-web.config should provide a good, basic structure for a Xen config file for your virtual machine. You'll need to open it up and update the paths appropriately. Once you're done, try sudo xl create halvm-web.config -c, and you should be good to go!

How does this thing work?

Halvm-web works by combining a web serving core (the halvm-web binary) with a "ramdisk" that contains the content. This way you can update your content without having to constantly rebuild the web server.