James Foris wrote:
Nathan Scott wrote:
On Wed, Aug 11, 2004 at 03:39:44PM -0500, James Foris wrote:
I have been using XFS in both IRIX and Linux for many years now, and
overall I am
quite happy with it - recommend it to everyone, run it on my home
systems, etc.
But, recently I ran into something that I need some help in
understanding/explaining.
The title says it all, really, but the details follow below.
Hi there,
Before I forget completely, from talking to Christoph a few
days back he suggested that the problems you were seeing may
have been resolved by some fixes that have gone into the tree
in the last week or two (they certainly touch this area) - if
you could retry your tests and let us know whether you still
see this degredation on Linus' current -bk tree, that would
be much appreciated!
More correctly, it happened between 2.6.5 and 2.6.5-bk1
From my test logs:
-----------------------
(first entry of each set is normal, second has O_DIRECT flag added)
Linux petpc1 2.6.5 #1 SMP Sun Sep 12 17:16:05 CDT 2004 i686 i686 i386 GNU/Linux
ext2
performance numbers include fsync : 6.690 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 7.978 sec; 64.176 MB/s 13.63 %
full
performance numbers include fsync : 0.188 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 8.118 sec; 63.067 MB/s 13.63 %
full
ext3
performance numbers include fsync : 6.541 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 9.015 sec; 56.794 MB/s 14.48 %
full
performance numbers include fsync : 0.003 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 7.815 sec; 65.513 MB/s 14.48 %
full
xfs
performance numbers include fsync : 6.539 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 7.740 sec; 66.147 MB/s 13.44 %
full
performance numbers include fsync : 0.000 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 7.278 sec; 70.353 MB/s 13.44 %
full
--------------------------------------------------------------------------------------------------^^
jfs
performance numbers include fsync : 6.517 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 8.016 sec; 63.871 MB/s 13.47 %
full
performance numbers include fsync : 0.000 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 7.455 sec; 68.677 MB/s 13.47 %
full
reiserfs
performance numbers include fsync : 7.412 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 8.798 sec; 58.195 MB/s 14.26 %
full
performance numbers include fsync : 0.004 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 50.301 sec; 10.179 MB/s 14.26
% full
Linux petpc1 2.6.5-bk1 #2 SMP Mon Sep 13 11:05:42 CDT 2004 i686 i686 i386
GNU/Linux
ext2
performance numbers include fsync : 6.658 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 7.937 sec; 64.512 MB/s 13.63 %
full
performance numbers include fsync : 0.151 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 8.142 sec; 62.885 MB/s 13.63 %
full
ext3
performance numbers include fsync : 6.293 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 8.660 sec; 59.124 MB/s 14.48 %
full
performance numbers include fsync : 0.003 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 7.849 sec; 65.232 MB/s 14.48 %
full
xfs
performance numbers include fsync : 6.565 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 7.782 sec; 65.793 MB/s 13.44 %
full
performance numbers include fsync : 0.000 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 8.545 sec; 59.917 MB/s 13.44 %
full
--------------------------------------------------------------------------------------------------^^
jfs
performance numbers include fsync : 6.494 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 8.016 sec; 63.873 MB/s 13.47 %
full
performance numbers include fsync : 0.000 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 7.396 sec; 69.227 MB/s 13.47 %
full
reiserfs
performance numbers include fsync : 7.151 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 8.594 sec; 59.579 MB/s 14.26 %
full
performance numbers include fsync : 0.004 sec
wrote 512.000 MB in 16 writes, 32.000 MB/write, 88.133 sec; 5.809 MB/s 14.26 %
full
So..... something in the 2.6.5-bk1 patchset caused the change. Any suggestions
where
to begin looking (other than fs/direct_io.x) ?
Thanks,
Jim Foris
Thanks again,
Jim Foris
thanks.
|