Sunday, March 30, 2008

bye apache, hello lighty

One of the things that I'd wanted to clean up was jettisoning Apache as my web server. In 2003, it was the cat's pajamas and the only real choice if you were interested hosting virtual sites. In 2008, it's an awesome and truly flexible piece of software that is way overkill for what I need: virtual name based hosting is no longer a big deal, and some of the web frameworks I'm looking at fit the CGI script model better than "runs as an embedded module server" model.

All I really need is a basic traffic cop to point people to the right resources for what they're looking for. Apache can do this, but it's kind of like using a hammer to swat a fly. :)

And that's where lighttpd comes.

It's basically a lightweight web server that can't do everything Apache can, but it's got most of the good stuff. URL rewrites and redirects, ssl support, authentication modules, etc. The config file reads more like a perl script than an laundry list of XMLish options, and it just seems easier to Get Stuff Done with it.

Right now, I've got it setup with two simple rules... if you ask for a url that ends with "" and starts with "/blog", you get punted over to this site. Everything else takes you to a temporary "we're br0kened" page via rewrite regardless of the sitename you used.

serverbeach handles the recovery

The hosting company for the server where I had the drive failure was Server Beach. This is the 2nd time I've had a drive flame out with them in nearly 5 years that I've used them. The last time was New Year's Eve 2004.

On the whole, the process around getting something back online was pretty impressive. The previous failure took about 24 hours to resolve to a point where I was back online. This time, the ticked was answered within 2 hours of me submitting it on a Saturday night around 10pm, and a new server was provisioned in around 4 hours. The only hitch is that I was assigned a new IP address, which wasn't really a huge problem since I was already having to recover from the old IP not responding to stuff like SMTP, DNS, and HTTP requests.

The end result is that in less than 10 hours after I reported the problem, I had another server up and was able to stand up minor placeholder services before the 12 hour mark.

As painful as having to start over from scratch is and as much of a annoyance as migrating old data back into position is going to be, I have to say that the prospect of being able to start over fresh is kind of handy. There were some things that about the old system (organizationally) that were suffering from neglect and could have stood to been cleaned up.

hello world


Turns out the hard drive in the server has failed. Tossing the good ole IDE buffer I/O errors, which means the data's just gone, man.


The good news is: all the important stuff was backed up.

Bad news is: the blog and all the stuff I had up for download (ie, pics, movies, game mods, etc) wasn't considered important.

Basically gonna start over from scratch here. :)

I'm also taking a new approach to how I'm going to handle stuff infrastructure-wise. I'm currently looking into going more distributed with stuff in the future. Email using google apps, use blogger to host the blog, put everything into everydns, and maybe check out the changes to EC2 for other services I need or want to play around with and let S3 worry about saving the data.

And all this against a backdrop of getting a new batcave, buying a new (to me at least) car, and thoughts turning once again towards pilots licenses, college degrees, and electronics hacking.