Re: Odd issue with mkfs.xfs under Lenny/amd64

[Robert Goley <ragoley@rdasys.com>]

I am replying to this because I found the answer.  I hope this little 
bit of information helps others confused by the xfs documentation.  
while sunit and swidth have to be set with the block size value in 512 
bytes (sunit) and as a multiple of that 512 bytes (swidth), The values 
returned by mkfs.xfs and xfs_info are listed as multiples of the block 
size.  In most cases I believe this will be 4096 but is subject to 
change based on options.  So the values I set were indeed being used but 
just displayed poorly.  For my sunit value of 512 the (64*4096) is 
correct and my swidth value of 2560 (320*4096) is correct also.  Hope 
this helps someone else....

Robert

Robert Goley wrote:
> I have not used xfs in a couple years but want to for a new server.  I 
> have followed all the xfs documentation I could find.  I believe my 
> commands are correct but am not seeing the results I expected from the 
> documentation.  The issues occur with the sunit and swidth values.  I 
> have a hardware raid5 array with 6 disks and a 256k strip size.  I 
> have created a LVM volume using the -I option with 256k for the 
> value.  I am creating the xfs filesystem with the following command:
>
> "mkfs.xfs -f -b size=4096 -l size=128m -d 
> sunit=512,swidth=2560,agsize=3g -L test /dev/raidvg/test"
>
> It displays this info...
> meta-data=/dev/raidvg/test       isize=256    agcount=34, 
> agsize=786432 blks
>         =                       sectsz=512   attr=2
> data     =                       bsize=4096   blocks=26214400, imaxpct=25
>         =                       sunit=64     swidth=320 blks
> naming   =version 2              bsize=4096 log      =internal 
> log           bsize=4096   blocks=32768, version=2
>         =                       sectsz=512   sunit=64 blks, lazy-count=0
> realtime =none                   extsz=4096   blocks=0, rtextents=0
>
> I tried mounting with with the command recommended below as well as 
> without options.
> mount -o sunit=512,swidth=2560,noatime,nodiratime,logbufs=8 
> /dev/raidvg/test /test
>
> Here are the results from xfs_info:
> meta-data=/dev/mapper/raidvg-test isize=256    agcount=34, 
> agsize=786432 blks
>         =                       sectsz=512   attr=2
> data     =                       bsize=4096   blocks=26214400, imaxpct=25
>         =                       sunit=64     swidth=320 blks
> naming   =version 2              bsize=4096 log      
> =internal               bsize=4096   blocks=32768, version=2
>         =                       sectsz=512   sunit=64 blks, lazy-count=0
> realtime =none                   extsz=4096   blocks=0, rtextents=0
>
>
>
> As you can see, these values do not match the settings I told it to 
> use.  If I use sunit=4096 and swidth=20480 values instead, It 
> correctly shows the values I wanted in mkfs.xfs and xfs_info as long 
> as I do not specify different values in the mount options.  Using 
> "mount /dev/raidvg/test /test" or "mount -o 
> sunit=4096,swidth=20480,noatime,nodiratime,logbufs=8 /dev/raidvg/test 
> /test" it will show the values I expected to see for my setup.  Here 
> is an xfs_info of those mounts.
>
> meta-data=/dev/mapper/raidvg-test isize=256    agcount=34, 
> agsize=786432 blks
>         =                       sectsz=512   attr=2
> data     =                       bsize=4096   blocks=26214400, imaxpct=25
>         =                       sunit=512    swidth=2560 blks
> naming   =version 2              bsize=4096 log      
> =internal               bsize=4096   blocks=32768, version=2
>         =                       sectsz=512   sunit=64 blks, lazy-count=0
> realtime =none                   extsz=4096   blocks=0, rtextents=0
>
>
>
> Can anyone explain these results?  It does not match up with anything 
> I have found on the web. Is this a bug?
>
> Robert