Using APC and Subversion Checkouts

I am in the process of moving my sites from Slicehost to Linode, as most are aware Slicehost is being phased out for Rackspace Cloud. Anyway, I upgraded my VPS at Linode last night to Debian “Squeeze” and installed APC, which created a problem with the WordPress installations I had there. Basically, I couldn’t access the admin. Through the power of Twitter and the invaluable help of Mark Jaquith, I learned that by default APC doesn’t play nice with files that are checked out via Subversion. I think the root of my problem was a stray apc config for overriding include once, but certainly the subversion issue could have been at play too.

Anyway, a little Googling on the subject, I discovered there’s a configuration option for APC to use ctime instead of the default mtime. Adding apc.stat_ctime=1 is all that is required.

While I’m on the subject of migrating from Slicehost to Linode, Andrew RIley pointed out that someone had written a ruby script to migrate the DNS information for you.

Addendum To be clear, I install WordPress using Subversion and the appropriate tag. My Habari installs are also all Subversion checkouts from trunk.

Migrating Servers

What an experience this has been. I hope to write a comprehensive post at some point about the experience, but suffice to say, it’s been quite the crash course on Apache, Debian, and DNS.

It all started with some curiosity, in which I purchased a Slice, and began poking around with the many tutorials they offer.  Ultimately, Owen provided me with his LAMPME script that did the basic setup for a debain, apache2, PHP 5, MySQL 5 set up, using mod_vhost (please don’t quote me on any of this 😀 ), which sorta got me going.  I got a bit sidetracked, and confused with setting up phpMyAdmin (I’m not that comfortable with CLI yet), and let the whole thing just sit, unused for a month or so.  However, due to the persistent problems with MediaTemple grid-servers, I vowed to stop shelling out money to them for the hassles.  If my sites go down, I want to be the one to blame, and no one else.  Besides, it’s high time that I learn more about this aspect of web development.

So, armed with help from Skippy, and some new found confidence, I got everything set up so as to have a basic understanding of how/where things go, and started moving domains.  I’m just about done, with this being one of the last.  If anyone notices anything wonky, assuming I have visitors, please let me know.

Also which is quite cool so far, is that I’m going to deploy the multi-site functionality of Habari to minimize the number of actual installs I have on the server, as I run everything off of head, this means less outdated installs.  Right now I’m looking at having just 2 actual installations, with everything else running off of those.  So far it’s worked out fine, though it took a little playing to get paths for things like images and files sorted out.  More on that later.  Certainly something that needs more attention, especially as silos mature.

Thanks to everyone who’ve helped in the process.