[cobalt-users] Carrie's A_Webalizer

[Step Firth <step@xxxxxxxxxxxxxxx>]

Hi.

> Original Message
> From: "Carrie Bartkowiak" <ravencarrie@xxxxxxxx>
> 
> You might also try renaming webalizer.pl so that it runs *before*
> logrotate does - awebalizer.pl for example, so that it can grab the
> latest stuff, but without the Incremental value set to "yes", this
> will still cause holes once logrotate gets around to gunzipping the
> log files.
> 
> CarrieB

This Webalizer 1.3 log holes problem is very annoying, and I have a very
crude fix I use, more on that later.

Carrie suggested renaming Webalizer AWebalizer so it runs BEFORE LogRotate
has the chance to blitz (archive) the web.log file. Has anyone tried this?
Does it work?

I rejoiced at such a simple work around, but having thought about it a bit
more, I am no longer convinced, or at the very least confused.

Currently, in the daily logrotate, it splits the log file and distributes it
amongst the sites, and if each individual web.log file is too big, archives
it and resets it. Then Webalizer runs but has no data to report on.

The AWebalizer scenario...
AWebalizer runs, and makes stats from yesterday's web.log file, for
yesterday's data. Log rotate runs, and blitzes/archives web.log.
Next day AWebalizer has no data to work on, therefore you have a whole in
your logs.

Am I wrong in this (I hope so).


What I do, is have a very crude fix.
In the cron.hourly I have a script that does the following

if hour = 7am (ie. after webalizer has run)
   for each site
      create new web.log
   }
}

Therefore the web.log file never gets big enough to be archived/blitzed by
logrotate. This works unless the site has too much traffic per day for the
logrotate file to allow in web.log


I would welcome any other thoughts or solutions on this.


Step