Re: creating filesystem on huge disks in RHEL3

["Robert G. (Doc) Savage" <dsavage peaknet net>]

On Thu, 2005-04-07 at 17:37 -0400, Viktor Hornak wrote:
> Hi,
> 
> I was wondering if someone could advice on how to create an (ext3) 
> filesystem on a huge disk (by current measures). The disk is actually a 
> hardware RAID, so it presents itself as /dev/sdb and is about 1.4TB big. 
> I created one big partition (/dev/sdb1) with fdisk and was trying to 
> create an ext3 filesystem afterwards (simply running mkfs.ext3 
> /dev/sdb1). The problem is that after a while, the "Writing inode 
> tables: ..." slowed down to a crawl and now it's about 0.8% (!) through 
> with inode creation and has been running for about 4 hours. It looks 
> like this is not the way to go. Is ext3 not scaling well for such big 
> partitions? Should I use a different filesystem? Or should I specify 
> some switches for mkfs.ext3? Also, maybe this is not how one should go 
> about it - maybe I need to create several partitions and merge them 
> somehow (LVM?).
> Any help appreciated!

Victor,

My first thought was to advise patience. But looking back when I ran
ext2fs on a ~1TB RAID5 array with RHEL3/ESu2, the filesystem creation
process only took about 1/2 hour. That was quick compared to the 12+
hours BIOS partition creation process. I didn't use LVM back then,
although I wish now that I had.

Now I'm wondering if I can safely use parted under RHEL4 to resize that
ext3 filesystem and recover the ~70G that RHEL3 couldn't get its arms
around the first time. There seem to be extended attribute issues
standing in my way. See Bugzilla #152301.

-- Doc 
Robert G. (Doc) Savage, BSE(EE), CISSP, RHCE | Fairview Heights, IL
RHEL4/ES on Tyan S2468UGN w/3G, dual Athlon MP 2800+, 1T RAID5
"Perfection is the enemy of good enough."
                         -- Admiral of the Fleet Sergei G. Gorshkov