Re: XFS Tunables for High Speed Linux SW RAID5 Systems?

[Eric Sandeen <sandeen@sandeen.net>]

Justin Piszcz wrote:
> Hi,
> 
> I was wondering if the XFS folks can recommend any optimizations for high 
> speed disk arrays using RAID5?
> 
> As well as if there is anything else I can do on the MD side?
> 
> fs.xfs.restrict_chown = 1
> fs.xfs.irix_sgid_inherit = 0
> fs.xfs.irix_symlink_mode = 0

OT but...

/me wonders if some of these could go away by now... :)

-Eric

> fs.xfs.panic_mask = 0
> fs.xfs.error_level = 3
> fs.xfs.xfssyncd_centisecs = 3000
> fs.xfs.inherit_sync = 1
> fs.xfs.inherit_nodump = 1
> fs.xfs.inherit_noatime = 1
> fs.xfs.xfsbufd_centisecs = 100
> fs.xfs.age_buffer_centisecs = 1500
> fs.xfs.inherit_nosymlinks = 0
> fs.xfs.rotorstep = 1
> fs.xfs.inherit_nodefrag = 1
> fs.xfs.stats_clear = 0
> 
> There is also vm/dirty tunable in /proc.
> 
> I was wondering what are some things to tune for speed?  I've already 
> tuned the MD layer but is there anything with XFS I can also tune?
> 
> echo "Setting read-ahead to 64MB for /dev/md3"
> blockdev --setra 65536 /dev/md3
> 
> echo "Setting stripe_cache_size to 16MB for /dev/md3"
> echo 16384 > /sys/block/md3/md/stripe_cache_size
> 
> (also set max_sectors_kb) to 128K (chunk size) and disable NCQ
> 
> Justin.
> 
>