New Linux Cluster Goes Into Production

Networking, Microsoft, Linux Add comments

So many months ago, I wrote “Web Server Migration: IIS/Win2K -> Apache2/RHEL4″ where I mentioned that I got the “ok” to begin working towards migrating our existing web server, a single Win2K box running IIS5 over to a pair of RHEL4 boxes running Apache2 set up in a failover configuration (using heartbeat). The new machines are now in production and the whole project is about 75% complete.

You may remember me bitching about the “embedded SATA RAID” these HP DL320’s come with in “RHEL4 and DL320 Embedded SATA RAID”. Well, it’s a big piece of shit, avoid it at all costs. Seriously.

I spent hours upon hours trying to figure out how to get the HP “drivers” to play well with it, having to revert back to installing RHEL4 Update 1 (after Update 3 had been released). Then, of course, once I upgraded the kernel, it broke. I finally said the hell with trying to use the hardware RAID and switched to just using Linux’s built-in software RAID and it’s working perfectly. If you run into this as well, just skip the bullshit and go for the software RAID.

So this is, officially, I suppose, the follow-up to “High-Availability Failover w/ Apache and Red Hat Enterprise Linux”. The new Linux “cluster” is up and running and is in production. A number of our web sites have been moved to this cluster — all of them, actually, except for the main one. The primary site depends so much on FrontPage Server Extensions that it’s not even funny. There are forms galore on that site, so everything had to be converted to use something else — in this case, a custom in-house application for creating web-based forms. Fortunately, that was a job for our webmaster and not me (oh, thanks again for the new book, Daniel).

This past Monday, new firewall policies were pushed that opened up 80/TCP and 443/TCP to the cluster and the DNS records were changed that evening. With only a four hour TTL value in DNS, it didn’t take long for all the incoming requests to start hitting the Linux cluster instead of the old Win2K box. Unfortunately, that server is also housing a few applications based on .NET, so we can’t completely get rid of the old server. The good news is, however, that it will soon be served up by Apache2 on Red Hat Enterprise Linux 4, and the FrontPage crap will be gone!

I’m gonna give it a week or so with the current sites to make sure we don’t run into any issues, then we’ll move our main web site over as well. The other sites are “specialty” sites, as I call them, and they don’t really get much traffic anyways (probably <100 hits/day altogether), while our “primary” site is averaging about 1.5 million/month. And since I have l33t Perl and PHP skills, I get to do some cool shit with our new site (dynamic content — finally! — and RSS feeds and podcasts, oh my!).

Should be cool…

(On a side note, I should mention that when I started working here three years and a few days ago, there was exactly one Linux box on our whole campus. It was being used by the guy who had the position that I now have. In the time that I’ve been working here, we now have a number of applications and services that RELY on Linux: our wireless networks, our web cache/proxy server, numerous in-house applications, scripts that automate our jobs and make things easier, employee directories, and a number of other things. $boss was kinda leery at first when I wanted to start moving “production” apps to Linux, but I think he’s gotten over that initially “nervousness” quite well!). Linux 1, Microsoft 0. =)

Leave a Reply

WP Theme & Icons by N.Design Studio
Entries RSS Comments RSS Login


Copyright © 2007 Jeremy L. Gaddis.
25 monkeys, 0.905 seconds.