On Thu, 31 Aug 2006, Mark Kirkwood wrote:
Mark Kirkwood wrote:
Ow Mun Heng wrote:
I've already updated it to the latest based on the suspend2 version.
$uname -r
2.6.17-suspend2-r4
$eix xfsprogs
Available versions: 2.7.3 2.7.11 2.8.10
Installed: 2.8.10
If not mistaken, the issue, (or barriers if not mistaken) was introduced
in the 2.6.17 kernel series.
the 2.6.16 series wasn't affected. (I could be wrong, I don't have net
access so, I can't verify)
Right - as it happens I'm doing an update today, so will let you know if
I see any write performance change.
FWIW, I've updated to 2.6.17 and I don't see any change in performance at all
(215Mb/s reads and 100Mb/s writes).
Now I'm on the standard source tree:
$ uname -r
2.6.17-gentoo-r7
$ eix xfsprogs
Available versions: 2.7.3 2.7.11 ~2.8.10
Installed: 2.7.11
which may be a factor.
The other thing I notice is that my filesystems are all under 50%, whereas
your troublesome one was at 80%or so:
$ df -m
Filesystem 1M-blocks Used Available Use% Mounted on
/dev/md/2 529 134 395 26% /
/dev/md/0 129 10 120 8% /boot
/dev/md/3 3911 32 3880 1% /tmp
/dev/md/4 3911 175 3737 5% /var
/dev/md/5 19537 3008 16530 16% /usr
/dev/md/6 19537 2668 16870 14% /home
/dev/md/7 104841 25682 79160 25% /data0
I might try writing a few big files to fill one of 'em up and see if it makes
any difference!
Cheers
Mark
--
gentoo-user@gentoo.org mailing list
You can also check your fragmentation, xfs_db -c frag /dev/..
and defrag it with xfs_fsir ..
--
gentoo-user@gentoo.org mailing list
