On Thu, Jan 28, 2010 at 7:02 PM, Dushan Tcholich <dusanc@xxxxxxxxx> wrote: > Hello > > When discussing R4 ( > http://forums.gentoo.org/viewtopic-p-6151767.html#6151767 )some users > mentioned slow fsync and this way to reproduce: > > Get a directory full of directories and small files by extracting > portage to test dir > http://mirror.bih.net.ba/gentoo/snapshots/portage-20100127.tar.bz2 > > ls -R /mnt/media/test/ | wc -l > 177073 > > Do a: > time chown -R portage:portage /mnt/media/test/ > > Then : > time sync > > real 40m9.350s > user 0m0.000s > sys 0m4.884s > > Throughput was around 80kb/s on a WD50000AAKS, no cryptocompress, > mounted with noatime, kernel 2.6.30 > > Is this normal, Do you have a spare partition on your drive? If yes, then try to write its random blocks and take a look at the throughput: it should be the same. > and why people tell me that ext4 and btrfs do this in > "several seconds" > Does the modification hits the platter for them If they pack inode items together in sequential blocks, then why not? However in this case those file systems will be slower on other tests. -- To unsubscribe from this list: send the line "unsubscribe reiserfs-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
