Of course! However, this probably isn't the right forum to answer it. I try to keep my work and blog separate (even though I fail on that goal sometimes), so I'd be happy to discuss this with you via e-mail. Drop me a line: major dot hayden at rackspace dot com.

I'm a customer and I don't know the answer =p

PS: Why do you use captcha? You lose what you type if you don't write words right. (As I did!) Aksimet is a really good spam stopper

It mounts great on my Ubuntu desktop but my Macbook Pro (10.6.2) is having a heck of a problem. I just can't seem to easily mount it to then work with in Textmate or TextEdit.

I have a mountpoint locally (empty) and think it must be something I'm not doing/missing.

It confused me a little at first, too. It's an on-demand pull. If someone sends a request for http://cdn.rackerhacker.com/style.css, the CDN provider will pull the latest copy from rackerhacker.com the first time. Subsequent accesses will just pull from the CDN only until the data expires.

If you update something on your site and you need to CDN to pick up the change, you can flush the file from the CDN on their website.

Push CDN's require you to upload your content to them when you make changes, and that requires some extra work that I'd rather not do.

Little confused by the pull vs. push thing. If the CDN pulls, doesn't that mean that it's using a polling interval to grab content and therefore is not-real time. Alternatively, wouldn't push mean that the site would push changed files to the CDN on demand?

My Golden rule for odd errors on a server I'm unfamiliar with:
Check DNS.

It's mind boggling just how often DNS is the problem

Those are great points! Good change management will definitely help future troubleshooting efforts by an administrator that is unfamiliar with the configuration.

I've worked at two companies so far where systems administrators deployed the code written by the developers and this really helps to keep administrators in the loop. Also, it requires that the developers write code well enough that it can be deployed by a person who is not familiar with the contents of the application.

However, if you’re not the primary administrator for the server, you may not always know what has changed recently or you may not be aware of changes in the server’s environment.

Monitoring
I've found that good monitoring and a few system notes can go a long way in this situation. For example, monitoring slave replication, NFS mounts or other inter-connected services can often give you a good point to start.

False Correlation
Be cautious against false correlation of events.

For example, after a kernel update, a client complained his web based forms were no longer functioning. He was sure that the reboot had caused the problem. We checked what may have happened and found no system updates to impact the service.

We then dug into the ftp logs to find that his offshore dev team had uploaded files just prior to our scheduled reboot. They introduced code changes that caused the problem.

Don't fall into a trap of false cause-effect scenarios. Coincidence can explain a large number of events, especially when changes are not well documented by all parties.