On Mon, 2003-12-15 at 14:46, Gustavo Rincon wrote:
> I created a filesystem using mkfs.xfs on a 2.7Terabyte md devices (Two 1.5
> Terabytes LUN defined in a 3ware 8000 raid controller) and the output was:
> meta-data=/dev/md0 isize=256 agcount=32, agsize=22888728
> = sectsz=512
> data = bsize=4096 blocks=732439296, imaxpct=25
> = sunit=8 swidth=16 blks, unwritten=1
> naming =version 2 bsize=4096
> log =internal log bsize=4096 blocks=32768, version=1
> = sectsz=512 sunit=0 blks
> realtime =none extsz=65536 blocks=0, rtextents=0
I think you may have run into a problem with large AGs (Allocation
Groups). Your AGs above are very large, which is a feature recently
added to XFS.
Can you try re-making the filesystems with 4G AGs? This should happen
automatically with xfsprogs < 2.6.0, or you can use the
-d agsize=4g option. Either way, you should see about 700 for the
"agcount" value, and your (agsize * bsize) should be less than 4G (in
bytes) in the mkfs output. Or put another way, agsize should come out
to 1048576 in mkfs output (if I got all my math right).
If that makes the problem go away, we probably know where to look.
Eric Sandeen [C]XFS for Linux http://oss.sgi.com/projects/xfs
sandeen@xxxxxxx SGI, Inc. 651-683-3102