OK, here is the scoop. We have one appliance that gets a custom partitioning via kickstart and a bit of Python love. Once the partition is complete, we install the base packages and then our appliance package. During the installation of the appliance package, it reads in the size of the multiple partitions it has available and their sizes. This all works like a charm. Side note, our Kickstart scripts are being run through Anaconda from CentOS.

As it stands, when the partitions are created, there is 5% by default that is utilized with every partition for super user access. This way here, it saves you from running out of space and being unable to access the drive. This is great on directories such as /boot, /var, /, /home, and etc. But when we partition our 750GB drives, we want a large growing space that doesn’t need this 5% reserved blocks percentage. Typically when you use mke2fs to create the partitions, outside of Kickstart and Anaconda, you would pass the -m flag with 0 (zero) as the variable. This would get rid of any reserved space, therefor allowing you to utilize the entire space. With the default 5% on 4x750GB hard drives, we lose a total of 150GB of space. That is a lot of space to lose, especially when our appliances main duty is storage.

I know we could add a %post section to our Kickstart scripts, call tune2fs -m 0 /partition/location, and then reinstall our appliance package so it can read the new drive partitioning, but is there any other way to do this? Someone said to use mke2fs.conf for this, however Kickstart and Anaconda documentation is far from good when it comes to tricky configurations, and it seems nobody else in the world is doing this. Does anyone know how to go about using the /etc/mke2fs.conf with Kickstart so I can have the drives partitioned with the -m 0 flag from the get go?

/me wishes we used Debian/Ubuntu with FAI!