Yeah, sorry, my documentation is pretty horrid at the moment, but there is a script already in place to do exactly what you're asking, using analog. The post about it was here
http://xepher.net/forum/index.php?topic=437.msg6963#msg6963 but the gist of it is this.
Login via SSH and run the command "generate-web-stats" and then visit
http://username.xepher.net/webstats/ (using your actual username.) It'll take a minute or two for stats to run, and it only has access to the current log files (e.g. this month.) I'm afraid the analog output may not be too useful for tracking 404s though. The log files are publically readable though, and located in /var/log/apache2/ The format is a bit strange though, as I had to set them up to be parseable for the virtualhosts. All the data from a normal CLF is there though.
Oh, and if anyone is curious, overall server stats are at
http://xepher.net/stats/xepher-net/I'm open to new suggestions for how to handle statistics, as the current system (logging included) was set up many years ago, and all the analyzers I looked at were pretty lame back then. I see better things from AWstats these days, and I've been considering trying it out again, but I'm looking for something that can sort by virtual hosts if possible.
