We've had some issues with logs not going back far enough, so I've found and modified the default log rotation schedule.

You can find the Jadu specific settings in /etc/logrotate.d/jadu

the default settings look like this -

        size 200M

        missingok

        rotate 10

        compress

        delaycompress

        notifempty

        create 620 jadu apache

        sharedscripts

        postrotate

                /usr/sbin/apachectl graceful

        endscript

These seem fine, unless you have a really busy site, or time of year, and you can imagine 200M will fill up pretty quickly, and rotate through.  You may only be keeping a few days worth of logs.

I've changed the jadu config file to the following, which adheres to our desired log policy a bit better.

        weekly

        missingok

        rotate 16

        dateext

        compress

        delaycompress

        notifempty

        create 620 jadu apache

        sharedscripts

        postrotate

                /usr/sbin/apachectl graceful

        endscript

Most notable is I'm keeping them by weeks now instead of by size.  The default version is nice because you know how much size to dedicate for logging.  With no size limit, you're going to eat some space if you're not prepared.

You'll also notice I'm keeping around 16 weeks worth of data.  This may seem excessive, but its quite helpful if you have to go back and look at something that happened 2 months ago... if, say, you had to investigate some suspicious behavior.  I'd hate to tell the security team, sorry, I have no idea because I only keep 1G of logs.

You can read a nice article about logrotator here if you're really interested in a custom configuration.