Skip to contentSkip to conversation

sunit & swidth on RAID-1?

DISCUSSION mike dentifrice 4 participants 5 messages Feb 11, 2009 1 message not in the archive

lore
Share & tools
lore
#0 lore raw

Hello,

I would like to know if it makes any sense to specify sunit & swidth parameters as mount options to XFS filesystems sitting on a Linux software RAID-1 array, as advertised here [1] or there [2].

I'm asking, because it seems that chunk_size does not matter in a RAID-1 context, nor do stripe units; or do they?

Any tip would be appreciated,

[1] http://www.socalsysadmin.com/2008/10/15/optimizing-xfs-on-raid-arrays/
[2] http://feedblog.org/2008/06/18/howto-configure-raid-strides-in-xfs/

Signature · 6 lines
mike dentifrice <[email protected]>

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#1 lore raw
Quoted 6 lines · mike dentifrice

Hello,

I would like to know if it makes any sense to specify sunit & swidth parameters as mount options to XFS filesystems sitting on a Linux software RAID-1 array, as advertised here [1] or there [2].

Quoted 15 lines

I'm asking, because it seems that chunk_size does not matter in a RAID-1 context, nor do stripe units; or do they?

Any tip would be appreciated,

[1] http://www.socalsysadmin.com/2008/10/15/optimizing-xfs-on-raid-arrays/
[2] http://feedblog.org/2008/06/18/howto-configure-raid-strides-in-xfs/

Signature · 6 lines
mike dentifrice <[email protected]>

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

mkfs.xfs will auto-optimize for software-raid arrays for the most part, with HW raid on a 3ware controller, tweaking mkfs.xfs does not help very much. with SW raid, at least for raid1 - I use the defaults.

Justin.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

A message that is not in the archive (sent off-list or not yet received): 20090211165316.GA4430@dentifrice

#3 lore raw
Quoted 10 lines · mike dentifrice

Justin Piszcz a dit:

Quoted 4 lines

mkfs.xfs will auto-optimize for software-raid arrays for the most part, with HW raid on a 3ware controller, tweaking mkfs.xfs does not help very much. with SW raid, at least for raid1 - I use the defaults.

Thanks for your reply. However, I wonder about the defaults:

xfs_info /dev/md1 | grep -E '(sunit|swidth)'

Quoted 12 lines

Outputs:

sunit=0 swidth=0 blks, unwritten=1
sectsz=512 sunit=0 blks

What's the consequence of 0 values in this case?

Greetings,

Signature · 1 line
mike dentifrice <[email protected]>
# xfs_info /dev/md2 | grep -E '(sunit|swidth)'
         =                       sunit=0      swidth=0 blks
         =                       sectsz=512   sunit=0 blks, lazy-count=0

Mine is similar, I guess you would need to benchmark to find out if you were worried about it. I ran multiple copies from a RAID1 (dual raptors) and was able to get the performance of both of the drives combine, so I am not really worried about this setting.

Justin.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#4 lore raw
In reply to a message not in the archive
Quoted 4 lines · On Mittwoch 11 Februar 2009 mike dentifrice wrote:

        sunit=0      swidth=0 blks, unwritten=1
        sectsz=512   sunit=0 blks

What's the consequence of 0 values in this case?

XFS then knows that 512 byte (aka sector) alignment is sufficient, as RAID-1 will only write 512b to each disk and does not need full stripe unit writes like RAID-4/5/6 need. Therefore performance is better with the values you have.

mfg zmi

Signature · 10 lines
// Michael Monnerie, Ing.BSc    -----      http://it-management.at
// Tel: 0660 / 415 65 31                      .network.your.ideas.
// PGP Key:         "curl -s http://zmi.at/zmi.asc | gpg --import"
// Fingerprint: AC19 F9D5 36ED CD8A EF38  500E CE14 91F7 1C12 09B4
// Keyserver: wwwkeys.eu.pgp.net                  Key-ID: 1C1209B4

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#5 lore raw
In reply to mike dentifrice
Quoted 8 lines · mike dentifrice

Hello,

I would like to know if it makes any sense to specify sunit & swidth parameters as mount options to XFS filesystems sitting on a Linux software RAID-1 array, as advertised here [1] or there [2].

I'm asking, because it seems that chunk_size does not matter in a RAID-1 context, nor do stripe units; or do they?

I don't think it does matter. And for what it's worth, mkfs.xfs queries an MD raid, but ignores raid1/10/0:

                /*
                 * Ignore levels we don't want aligned (e.g. linear)
                 * and deduct disk(s) from stripe width on RAID4/5/6
                 */
                switch (md.level) {
                case 6:
                        md.raid_disks--;
                        /* fallthrough */
                case 5:
                case 4:
                        md.raid_disks--;
                        /* fallthrough */
                case 1:
                case 0:
                case 10:
                        break;
                default:
                        return 0;

In general, mkfs.xfs on md should just "do the right thing"

-Eric

Quoted 5 lines

Any tip would be appreciated,

[1] http://www.socalsysadmin.com/2008/10/15/optimizing-xfs-on-raid-arrays/
[2] http://feedblog.org/2008/06/18/howto-configure-raid-strides-in-xfs/

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs