Re: Getting high cache_bypass_misses in my setup

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

 



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

--
Eric Wheeler



> 
> -- 
> 
> 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