Re: Getting high cache_bypass_misses in my setup

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 2019/10/23 3:12 上午, Eric Wheeler wrote:
> On Tue, 15 Oct 2019, Coly Li wrote:
>> On 2019/10/12 10:23 下午, Sergey Kolesnikov wrote:
>>> Hello everyone.
>>>
>>> I'm trying to get my bcache setup running, but having almost all my
>>> traffic bypassing the cache.
>>> Here are some stats that I have:
>>>
>>>
>>> root@midnight:~# cat
>>> /sys/fs/bcache/9820f407-457a-46e3-abc0-f2214d39b64c/bdev0/stats_total/bypassed
>>> 2.8G
>>> root@midnight:~# cat
>>> /sys/fs/bcache/9820f407-457a-46e3-abc0-f2214d39b64c/bdev0/stats_total/cache_bypass_misses
>>> 247956
>>> root@midnight:~# cat
>>> /sys/fs/bcache/9820f407-457a-46e3-abc0-f2214d39b64c/bdev0/stats_total/cache_bypass_hits
>>> 5597
>>> root@midnight:~# cat
>>> /sys/fs/bcache/9820f407-457a-46e3-abc0-f2214d39b64c/bdev0/stats_total/cache_hits
>>> 233
>>> root@midnight:~# cat
>>> /sys/fs/bcache/9820f407-457a-46e3-abc0-f2214d39b64c/bdev0/stats_total/cache_misses
>>> 243
>>>
>>> And now for my machine setup.
>>> Running ubuntu 18.04 LTS with 5.0.0-31-lowlatency kernel.
>>> Cache device is a partition on NVMe PCI-e SSD with 4k logical and
>>> physical sector size.
>>> Backing device is LVM logical volume on a 3-drive MD RAID-0 with 64K
>>> stripe size, so it's optimal IO is 192K.
>>> I have aligned backing-dev data offset with
>>> make-bcache -B -o 15360 --writeback /dev/vm-vg/lvcachedvm-bdev
>>>
>>> I have tried all recommendations for routing traffic to SSD:
>>>
>>> echo 0 > /sys/fs/bcache/9820f407-457a-46e3-abc0-f2214d39b64c/congested_read_threshold_us
>>> echo 0 > /sys/fs/bcache/9820f407-457a-46e3-abc0-f2214d39b64c/congested_write_threshold_us
>>> echo 0 > /sys/fs/bcache/9820f407-457a-46e3-abc0-f2214d39b64c/bdev0/sequential_cutoff
>>>
>>> But I still get almost all traffic going to cache_bypass_misse. BTW,
>>> what does this stat mean? I don't get it from the in-kernel manual
>>>
>>> Any help?..
>>
>> I have no much idea. The 4Kn SSD is totally new to me. Last time I saw
>> Eric Wheeler reported 4Kn hard diver didn't work well as backing device,
>> and I don't find an exact reason up to now. I am not able to say 4Kn is
>> not supported or not, before I have such device to test...
> 
> We pulled the 4Kn SSD configuration, it wasn't stable back in v4.1.  Not 
> sure if the problem has been fixed, but I don't think so.  
> 
> Here is the original thread:
> 
> https://www.spinics.net/lists/linux-bcache/msg05971.html
Yes, this is the problem I wanted to say. Kent suggested me to look into
the extent code, but I didn't find anything suspicious. Also I tried to
buy a 4Kn SSD, but it seemed not for consumer market and I could not
find it from Taobao (www.taobao.com).

I keep this problem in my mind always, just no progress ....

-- 

Coly Li



[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Index of Archives]     [Linux ARM Kernel]     [Linux Filesystem Development]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Security]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [ECOS]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux