[rhelv6-list] RHEL6.2 XFS brutal performence with lots of files
Jussi Silvennoinen
jussi_rhel6 at silvennoinen.net
Mon Apr 15 08:50:35 UTC 2013
> avg-cpu: %user %nice %system %iowait %steal %idle
> 11.12 0.03 2.70 3.60 0.00 82.56
>
> Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
> md127 134.36 10336.87 11381.45 19674692141 21662893316
Do use iostat -x to see more details which will give a better indication
how busy the disks are.
> Running hdparm on one of the software raid5 drives reports decent numbers.
>
> /dev/sdb:
> Timing cached reads: 12882 MB in 2.00 seconds = 6448.13 MB/sec
> Timing buffered disk reads: 396 MB in 3.06 seconds = 129.39 MB/sec
>
> running some crude dd tests shows reasonable numbers, I think.
>
> # dd bs=1M count=1280 if=/dev/zero of=test conv=fdatasync
> 1342177280 bytes (1.3 GB) copied, 29.389 s, 45.7 MB/s
Yes crude and not very useful.
> I have other similiar filesystems on ext4 with similiar hardware and
> millions of small files as well. I don't see such sluggishness with small
> files and directories there. I guess I picked XFS for this filesystem
> initially because of its fast fsck times.
Are those other systems also employing software raid? In my
experience, swraid is painfully slow with random writes. And your workload
in this use case is exactly that.
> # grep md127 /proc/mounts
> /dev/md127 /mesonet xfs
> rw,noatime,attr2,delaylog,sunit=1024,swidth=4096,noquota 0 0
inode64 is not used, I suspect it would have helped alot. Enabling it
afterwards will not help for data which is already on the disk but it will
help with new files.
--
Jussi
More information about the rhelv6-list
mailing list