It’s getting to the point where I need to stop doing server generation for my RSS feed whenever it is requested. Mainly because RSS seems to be eating a majority of my bandwidth and doing the server side generation doesn’t give an etag or a last modified date, this makes conditional get broken when people try to fetch my feed. Fortunately, Apache can handle this stuff if I do a static file.
I’m getting sick of hacking up PyBlosxom, so I decided to do this using some mod_rewrite voodoo and a cron job. The first thing that you’ll need is a cron job to periodically request the feed. This is how I have my entry:
0 * * * * cd /home/patrick/Documents/www ; wget http://patrick.wagstrom.net/weblog/index.rss --user-agent="Cacher" > /dev/null ; if [ -f index.rss.1 ]; then mv index.rss.1 index.rss; fi
This fetches the feed once an hour and saves a copy in my root directory for my web server. You’ll notice that I tell WGet to use a special user agent that I call “Cacher” to fetch the document. Next we need some mod_rewrite foo to change it all around.
:::apache
RewriteCond %{HTTP_USER_AGENT} !Cacher
RewriteRule ^/weblog/index.rss /index.rss [L]
This looks at the user agent of the request and sees if it is not “Cacher”, if that is true, then it redirects silently to the cached file. Now, this may create some small delay in geting my feed updated, but it should also reduce the overall bandwidth consumption of my RSS feeds. Sigh, just another sign I need to finish Priblog.