On Wed, 26 Apr 2006, Erich Schubert wrote: > A recent real-world benchmark suggested that XFS is the top choice for
Well, just make sure / is not XFS. 1. It is not fsck-friendly, you need early userspace or specially deployed boot machinery (like a service partition) to xfs_repair it 2. It does not journal or order data, and just about *nothing* fsync()s stuff in / 3. / is almost always rw, i.e. you *really* want a sturdy filesystem there, and XFS is engineered for speed at the cost of reliability, not the other way around. 4. Someone might have this strange idea they can corrupt the filesystem installing a boot loader inside it and things should still work... and XFS won't let you get away with that like ext2/ext3 does. > filesystem performance; reiserfs didn't do very well in that study, > nor did ext3. > http://www.debian-administration.org/articles/388 That article is quite clueless in parts, and its choice of a workload is... strange at best. See http://lwn.net/Articles/181343/ for some responses to it. > The interesting number was the mount time though: > ReiserFS took 5 to 15 times longer to mount the FS (2.3 secs) when > compared to other FS (Ext3 = 0.2, JFS = 0.2, XFS = 0.5), and also 2 > times longer to umount the FS (0.4 sec) Yes, reiser is very slow to u?mount... If that's a deciding factor for you, switch to ext3. -- "One disk to rule them all, One disk to find them. One disk to bring them all and in the darkness grind them. In the Land of Redmond where the shadows lie." -- The Silicon Valley Tarot Henrique Holschuh _______________________________________________ initscripts-ng-devel mailing list [email protected] http://lists.alioth.debian.org/mailman/listinfo/initscripts-ng-devel
