If you tried to view this site last week and couldn’t, then I apologise. Up until the weekend, this web server was frequently unreachable, and I’ve lost track of the number of times that I had to restart Apache, and/or the whole server. Thankfully, things seem to be better now.
I’m still not sure what caused it; whether I got hit by an overzealous web crawler, or a package update last weekend broke something, but either way the virtual machine that the server runs on was constantly running out of memory. Even when I tried to reign in Apache’s multiple processes, pages wouldn’t load.
Eventually on Thursday night I downloaded all of my data off the server onto my Mac, and installed a fresh virtual machine image. The previous image has been in use since I switched to Bytemark as a host a few years ago, and has seen an upgrade from Debian Lenny to Squeeze, and from Bytemark’s old vhost manager to their new Symbiosis manager. This fresh image meant that I didn’t have any pre-upgrade baggage, and a clean set of configuration files.
Even that didn’t initially solve the problem initially. It was only after I removed the W3 Total Cache plugin from WordPress that things got better. I’m not sure if it was something to do with how I configured the plugin, but, ironically, not havingW3 Total Cache running resulted in a bigger performance boost. Crucially, there were no out of memory errors.
Rebuilding the site has meant that some things aren’t working as yet. I haven’t got around to re-enabling SpamAssassin so I’m getting quite a bit more junk in my inbox at present (a lot of it in Portuguese for some reason…) and there are some other missing WordPress plugins that I haven’t reinstalled. I’m trying out WP Fast Cache as a more basic caching solution for the most frequently-visited pages, and it seems to work okay.