Really LVM, really...

So we're doing some work on a server, and like always, the work must be done in the wee hours of the AM. So I'm working away creating LVM2 volumes and everything is going well. I'm even following what seems to be "best practices" by not using the raw device (dev/sdb) but instead creating a partition first then using that for LVM. On one of the servers I created a 600GB partition on (/dev/sdb1) which is basically the entire "disk" (its really just a fibre volume that can be expanded later but whatever). I create the LVM side making a LV of only 500GB to give me some wiggle room in the future... when the users say "oh no, the drive is full" I can tell them that if I work really hard and they pay me lots of OT I can maybe find another 100GBs in there somewhere... like that episode of ST:NG when Scotty is beamed aboard instead of some expendable crewman, long story short, Geordi tells the Cap. that Problem_A will take x hours to fix, Scotty asks "how long will it really take to fix" and is amazed that Geordi didn't pad the time (x * 3 or x * 5 etc.)...

Like I said, one server worked fine with a 600GB disk, the other server however, is getting an upgrade from 1.8TB to 3TB. Well, as you know, you can't use fdisk to make a 3TB partition, you must use parted and the partition has to be a GPT. Well, when you make the PV everything seems fine, then you make a VG and do a pvdisplay and everything goes FUBAR!

You see, we're forced to us an OS version that is a little old... okay, REALLY old. We are on Fedora Core6, the current version (as of this writing) is 13, and they don't use the "Core" anymore! So there is this little "bug" in LVM that has been fixed since 2007ish (yeah, it's mid 2010) that is causing me some troubles...

A pvdisplay shows that only 1024 GB are usable and 8647 EB are not usable... I know it's late but it actually says EB (exabytes). To put this in prospective, it is postulated that the whole world (including iPods, Digital Cameras, laptops, Google, Microsoft, IBM, Amazon, etc.) in total is using less than 1000EB. So there is over eight times the world's total digital storage not currently usable on my server! I'll admit I was a little shocked it was not usable ;)

After some googling I found a bug report (https://bugzilla.redhat.com/show_bug.cgi?id=365741) over at RHLE saying they will fix it in future builds. Well the bad news is, to use a newer build (one where that problem is fixed) I will need to completely rebuild both systems... now where am I going to get that kind of user downtime?! The good news is the bug report says it should just be an error in reporting and there is a work around (the work around didn't seem to work) so I decided to move forward and finish building out the volume. I'm writing this as I watch 700GB of data get copied over to the new volume... wee.