Anyone who’s ever tried to write a nontrivial application on Google App Engine has encountered at least seven* design decisions that have led to serious head-scratching moments. One of those happened to me about a month ago, while integrating Chef into our course at The Data Incubator. Our goal was to allow for one-click spinning up (on DigitalOcean’s cloud) and monitoring of our Fellows’ course machines, already under Chef management.
* No basis in fact - there are probably more than seven. It should be noted that the Google Cloud Platform is going to greatly improve this situation by allowing you to deploy Docker containers - woohoo!
##A First Look
Chef servers have an HTTP API. Seems like it’d be an easy integration, right? While GAE doesn’t let you do many things (including making SMTP connections), one thing you, thankfully, can do with relative ease is make HTTP requests (although everyone’s favorite python HTTP library, requests, is a total nightmare - but that’s for another blogpost). This was going to be a quick job - we’d spend a couple days coding, write some tests, and have one-click deployment, right? right?. As you probably guessed, that timeline was anything but right.
Problem #1 : PyCrypto and PyCrypto only.
Why, do you ask, is Chef’s API requiring the use of libcrypto? Well, let’s examine that Chef server HTTP API doc again. You’ll notice you have to send very specific requests, with very specificially encoded headers, to the Chef Server. In particular, you need to use a Chef Client private key to sign the request (using an RSA signature protocol - private key signing) and send that signature in the headers.
To circumvent this without getting unnecessarily deep into RSA implementation specifics, we looked at the Chef docs and found some simple commands to sign the headers using openssl. We found that PyCrypto’s
RSA.decrypt* function would appropriately generate our message signature, provided we byte-padded the input string beforehand (in accordance with a particular set of standards).
*A brief aside: we didn’t just stumble upon
RSA.decrypt as the function to use… we had a bit of prior knowledge. RSA private key signatures work essentially by having the signer raise a message (or, more accurately, the message’s hash)
m to the
d power (where
d is the private key), allowing anyone with the public key
e to verify that
m^ed == m (mod n of course). RSA encryption also raises a message
m' to the
d power when the recipient wants to read the message using their private key, where
m' == m ^ e. Anyway, enough math and protocols!
Here’s what that byte-padding looks like (props to this post for the snippet, replicated in our own codebase):
With this out of the way, we thought we were in the clear, until…
dev_appserver.py doesn’t respect ignoring SSL verification.
Our Chef server lives on a DigitalOcean box, and the SSL certificate is self-signed (we dont need client-facing trustworthiness on this one). No problem: we’d just pass
verify=False as a flag into our HTTP request code. Unfortunately, this recently became a problem in the GAE development environment.
We still haven’t found a great solution to this, so we just ensure that locally, we have our self-signed certificate added into OpenSSL a la the following post. If you have a better solution, please let me know!
Solution: An Open Source Library!
We open-sourced the code we use to hopefully save other devs this headache. It uses version
requests (the last version that can be used on GAE), and the latest PyCrypto (2.6.1 as of this writing). It works swimmingly, though only on the actual GAE environment if you don’t trust your Chef Server’s SSL certificate.
Check it out on github - you can
pip install chef_gae as well. Pull requests always welcome!