Hi Nathan,
Here's the info you requested based on my default settings, which do not
include the use of the external journal:
sh-2.04# xfs_info.sh /hd/vol_mnt0/
meta-data=/hd/vol_mnt0 isize=1024 agcount=35, agsize=1048560
blks
data = bsize=4096 blocks=36397056, imaxpct=25
= sunit=16 swidth=32 blks, unwritten=0
naming =version 2 bsize=4096
log =internal bsize=4096 blocks=32768 version=2
= sunit=1 blks
realtime =none extsz=65536 blocks=0, rtextents=0
At this point, I should point out that I've just realized that my default
inode size is not 1k, but rather 256. I'm also specifying sunit and swidth,
and didn't realize that the default inode drops back down to a lower
default. I'm going to rerun the tests with 1k then 2k inode size. However,
you may still be interested in the performance impact described below.
Again, when I say "with acls", the inode size is 256.
Also, as far as performance goes, when acls are turned off, and the journal
is placed on an external device, the nbench throughput is 60MB/sec at 6
clients. It tapers off to 53MB/sec at 32 clients. However, when I turn
acls on, my peak is 40MB/sec at 4 clients and drops to 26MB/sec at 30
clients. If I rerun the last test, but with the internal journal, my peak
goes back up to 52MB/sec and the low is 37MB/sec.
I'll let you know how things go.
-Mike
-----Original Message-----
From: linux-xfs-bounce@xxxxxxxxxxx [mailto:linux-xfs-bounce@xxxxxxxxxxx] On
Behalf Of Nathan Scott
Sent: Monday, March 22, 2004 3:36 PM
To: Mike Young
Cc: 'Chris Wedgwood'; linux-xfs@xxxxxxxxxxx
Subject: Re: External Journal and ACL Support
On Mon, Mar 22, 2004 at 03:28:05PM -0800, Mike Young wrote:
> Hi Chris,
>
> That's an excellent question. My default is 1k right now. I'm going to
> give 2k a try.
Can you send your filesystem geometry (xfs_info output)?
And any performance numbers you have for each of the different
geometries you've tried would be good to see too.
The difference moving from the default 256 byte inode to any
larger inode size is where the big gains are seen, I wouldn't
expect too much improvement going from 1k to 2k. Though it
may help a bit for small directories with ACLs... (and small
directories in general).
cheers.
> -----Original Message-----
> From: linux-xfs-bounce@xxxxxxxxxxx [mailto:linux-xfs-bounce@xxxxxxxxxxx]
On
> Behalf Of Chris Wedgwood
> Sent: Monday, March 22, 2004 3:17 PM
> To: Mike Young
> Cc: linux-xfs@xxxxxxxxxxx
> Subject: Re: External Journal and ACL Support
>
> On Mon, Mar 22, 2004 at 02:31:16PM -0800, Mike Young wrote:
>
> > To set the journaling up for external placement, I'm using "mkfs.xfs
> > /dev/md1 -l logdev=/dev/md0,size=10000b". For ACLs, I'm just using
> > getfacl to store the acls in a file till I can reapply them to the
> > new directories using "setfacl -set-file".
>
> Do larger inodes help? I wonder is with ACLs you're not becoming
> seek-bound on metadata.
>
>
>
> --cw
>
>
>
>
>
--
Nathan
|