On Wed, Mar 19, 2003 at 02:39:26PM -0600, Steve Lord wrote:
>
> Also, pay attention to the mkfs output, the sunit and swidth lines,
> these control how data will get layed out. You want to make sure they
> line up with your device configuration. This may or may not happen
> automatically depending on your setup. Read the mkfs.xfs man page for
> how to control them yourself.
Ah. I'm just trying XFS for the first time and also have some tuning
questions!
Is the general recommendation to set sunit/su and swidth/sw when running
on a hardware RAID system? In my case, I have a bunch of volumes created
on an IBM FAStT700 SAN (rebranded LSI 4884, like the SGI SAN Server
1000). The RAID controller lets me adjust "segment size" from 8-256K,
with a default of 64KB on a RAID5 volume. The docs say "A segment is the
amount of data, in kilobytes, that the controller writes on a single
drive in a logical drive before writing data on the next drive." Am I
correct when I interpret this as the "strip unit size" defined in the
mkfs.xfs man page?
Assuming I'm on the right track then, since I have 512 byte data
blocks, the 64KB stripe takes up 128 blocks. I have 12 drives in the
stripe, so to create a XFS FS I'd run:
mkfs.xfs -d sunit=128,swidth=1536
or
mkfs.xfs -d su=64k,sw=768k
When I create a FS like this, I get:
# mkfs.xfs -f -i size=512 -d sunit=128,swidth=1536 /dev/sdb1
meta-data=/dev/sdb1 isize=512 agcount=188, agsize=1048576 blks
data = bsize=4096 blocks=197025168, imaxpct=25
= sunit=16 swidth=192 blks, unwritten=0
naming =version 2 bsize=4096
log =internal log bsize=4096 blocks=24064, version=1
= sunit=16 blks
realtime =none extsz=65536 blocks=0, rtextents=0
Am I going about this the right way, or have I gone wrong somewhere along
the way in my logic? :)
On a related note, I'm wondering what happens when I grow volumes. In this
particular volume, it's 752GB right now but what if I add 2 more drives
to the volume to grow it by another 120GB? Can I respecify the swidth value
somehow or should I simply not specify sunit/swidth on volumes I know will
be growing and leave them at the default 0 values?
Finally, anything I should be looking out for on 1.7TB volumes?
Regards,
-Mark
>
> Steve
>
> --
>
> Steve Lord voice: +1-651-683-3511
> Principal Engineer, Filesystem Software email: lord@xxxxxxx
>
--
Mark Mayo <mark@xxxxxxxxxx> http://www.vmunix.com/~mark
PGP Fingerprint: 32BA C076 D78C 109B 2558 25E5 0CCE C6C1 262E 28AF
Commitment, n.: Commitment can be illustrated by a breakfast of
ham and eggs. The chicken was involved, the pig was committed.
|