The output of dracut -f -L 5 has some parts with (pseudo) random order. That gives annoying noise when debugging and investigating differences. The following patch makes the output stable for me: --- module-setup.sh 2012-10-15 12:49:16.700733551 +0200 +++ /usr/lib/dracut/modules.d/90kernel-modules/module-setup.sh 2012-10-15 12:50:18.330593970 +0200 @@ -43,7 +43,7 @@ # install unix socket support hostonly='' instmods unix instmods "=drivers/pcmcia" =ide "=drivers/usb/storage" - find_kernel_modules | block_module_filter | instmods + find_kernel_modules | block_module_filter | sort | instmods # if not on hostonly mode, install all known filesystems, # if the required list is not set via the filesystems variable if ! [[ $hostonly ]]; then Please consider applying a patch like this. dracut-018-105.git20120927.fc17.noarch
sorry, it's different now in upstream... why don't you use "lsinitrd" to inspect the contents of the initramfs?
I do also use lsinitrd, but monitoring changes in the build output is convenient for getting an easy answer to exactly _why_ the content in the initramfs changed and what consequences it has to change the dracut configuration or the installed package base. Ok, I will wait and see how it works with the next upstream version.