Fedora 11, Anaconda and dmraid

I had an issue when first installing F11 that I failed to blog about at the time unfortunately. However, this is a long running issue with Anaconda that was exacerbated in F11 - so thought it was worth noting my fix.

The issue occured while installing Fedora with Anaconda (the standard Fedora installer). When I arrived at the disk partitioning stage, Anaconda did not list either of my sata drives available on the system.  Apparently what is happening is that Anaconda does not recognize any drives that have raid meta data written to it. This is true even if after you have raid turned off in bios. As long as software raid was at some point activated, meta information is written to the disks and Anaconda will reject them as available disks.

The normal fix for this issue (pre and post F11) was to start Anaconda with the nodmraid option:

linux nodmraid

The behaviour is described in this bug report: https://bugzilla.redhat.com/show_bug.cgi?id=499733

Since this fix was not available in F11 (as described above), I had to find another work around.  Ultimately, the work around is to actually fix the problem by removing the meta data altogether.

**Standard Disclaimer**
Obviously this will kill any raid you have setup - so be careful.  In fact, don't do this.  Why are you following my advice, I wouldn't.

Unfortunately, I don't have the _exact_ notes from when I went through this; however the basic steps are to ensure raid is disabled in bios and systematically remove the meta data on each raid disk.

To list raid disks, as root run: dmraid -r

To remove meta data on all disks, as root run: dmraid -rE

To remove meta data on a specific disk, as root run: dmraid -rE /dev/name_of_disk

If I recall correctly, you may be able to do this from within Anaconda by hitting Ctrl-Alt-F2 to get a terminal.

man dmraid is your friend.

However, unfortunately there seems to be another related bug, where dmraid will not erase invalid meta data. Check https://bugzilla.redhat.com/show_bug.cgi?id=517761 for more info. Basically, the fix is to just run Anaconda with the 'nodmraid' directive.