bad performance on touch/cp file on XFS system

[Zhang Qiang <zhangqiang.buaa@xxxxxxxxx>]

Dear XFS community & developers,

I am using CentOS 6.3 and xfs as base file system and use RAID5 as hardware storage.

Detail environment as follow:

 ÂOS: CentOS 6.3

 ÂKernel: kernel-2.6.32-279.el6.x86_64

 ÂXFS option info(df output): /dev/sdb1 on /data type xfs (rw,noatime,nodiratime,nobarrier)

Detail phenomenon:

 ÂÂ

  # df

  Filesystem      ÂSize ÂUsed Avail Use% Mounted on

  /dev/sda1       Â29G  17G  11G Â61% /

  /dev/sdb1       893G Â803G  91G Â90% /data

  /dev/sda4       2.2T Â1.6T Â564G Â75% /data1

 ÂÂ

  # time touch /data1/1111

  real  Â0m23.043s

  user  Â0m0.001s

  sys   0m0.349s

 ÂÂ

  # perf top

  Events: 6K cycles

  Â16.96% Â[xfs]           [k] xfs_inobt_get_rec

  Â11.95% Â[xfs]           [k] xfs_btree_increment

  Â11.16% Â[xfs]           [k] xfs_btree_get_rec

   7.39% Â[xfs]           [k] xfs_btree_get_block

   5.02% Â[xfs]           [k] xfs_dialloc

   4.87% Â[xfs]           [k] xfs_btree_rec_offset

   4.33% Â[xfs]           [k] xfs_btree_readahead

   4.13% Â[xfs]           [k] _xfs_buf_find

   4.05% Â[kernel]         Â[k] intel_idle

   2.89% Â[xfs]           [k] xfs_btree_rec_addr

   1.04% Â[kernel]         Â[k] kmem_cache_free

It seems that some xfs kernel function spend much time (xfs_inobt_get_rec, xfs_btree_increment, etc.)

I found a bug in bugzilla [1], is that is the same issue like this?

It's very greatly appreciated if you can give constructive suggestion about this issue, as It's really hard to reproduce from another system and it's not possible to do upgrade on that online machine.

[1] https://bugzilla.redhat.com/show_bug.cgi?format=multiple&id=813137

Thanks in advance

Qiang