On 2017/1/17 上午11:10, Stan Hoeppner wrote:

On 01/16/2017 08:35 PM, Jake Yao wrote:

quoted

I have a raid5 array on 4 NVMe drives, and the performance on the
array is only marginally better than a single drive. Unlike a similar
raid5 array on 4 SAS SSD or HDD,  the performance on array is 3x
better than a single drive, which is expected.

It looks like when the single kernel thread associated with the raid
device running at 100%, the array performance hit its peak. This can
happen easily for fast devices like NVMe.

The md raid personalities are limited to a single kernel write thread. 
Work is in progress to alleviate this bottleneck by using multiple write
threads.  When it will hit mainline I don't know.

If you want 8 writing threads, and your md raid5 device is /dev/md0, you
may have a try with,
	echo 8 > /sys/block/md0/md/group_thread_cnt

quoted

This can reproduced by creating a raid5 with 4 ramdisks as well, and
comparing performance on the array and one ramdisk. Sometimes the
performance on the array is worse than a single ramdisk.

The kernel version is 4.9.0-rc3 and mdadm is release 3.4, no write
journal is configured.

Is this a known issue?

It was, but you are on 4.9 kernel, group_thread_cnt should work for you.

Coly