public inbox for linux-bcache@vger.kernel.org
 help / color / mirror / Atom feed
* Flushing dirty data after bcache detach
       [not found] ` <423285490.335846.1377074174264.JavaMail.root-qD/LZSXtZz5FYGZjsVlUvw@public.gmane.org>
@ 2013-08-21  8:48   ` Georg Schönberger
       [not found]     ` <1057573722.337424.1377074937216.JavaMail.root-qD/LZSXtZz5FYGZjsVlUvw@public.gmane.org>
  0 siblings, 1 reply; 2+ messages in thread
From: Georg Schönberger @ 2013-08-21  8:48 UTC (permalink / raw)
  To: linux-bcache-u79uwXL29TY76Z2rM5mHXA

Hi bcache team,

we are currently developing performance test scripts with Fio for bcache.
At the end of a test we want to detach and destroy the cache. Therefore we did:
    echo 1 > /sys/block/$dev/bcache/detach
    echo 1 > /sys/block/$dev/bcache/stop

    echo 1 > /sys/fs/bcache/*/stop
    echo 1 > /sys/fs/bcache/*/unregister

With a simple randwrite test with we had about 2GB dirty data in the cache. After detaching the dirty data flush rate was about 900KB/s from SSD to HDD.
Is there a way to accelerate flushing of dirty data?
What is the correct way to detach/destroy a writeback caching device?

* Kernel:  3.11.0-2-generic #5-Ubuntu SMP Mon Aug 12 16:09:51 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux
* Kernel module info:
filename:       /lib/modules/3.11.0-2-generic/kernel/drivers/md/bcache/bcache.ko
author:         Kent Overstreet <kent.overstreet-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
license:        GPL
license:        GPL
author:         Kent Overstreet <koverstreet-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>
srcversion:     EE441BF41AAB31C3E44F5D2
depends:        
intree:         Y
vermagic:       3.11.0-2-generic SMP mod_unload modversions 
* iostat while flushing after detach:
Device:            tps    kB_read/s    kB_wrtn/s    kB_read    kB_wrtn
sda               0.00         0.00         0.00          0          0
sdc             229.50       912.00         6.00       1824         12
sdb             228.00         0.00       912.00          0       1824
bcache0           0.00         0.00         0.00          0          0

Thanks a lot for your help in advance, Georg.
--
: Georg Schönberger
: Research & Development Executive
: Thomas-Krenn.AG | The server-experts
: http://www.thomas-krenn.com | http://www.thomas-krenn.com/wiki

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: Flushing dirty data after bcache detach
       [not found]     ` <1057573722.337424.1377074937216.JavaMail.root-qD/LZSXtZz5FYGZjsVlUvw@public.gmane.org>
@ 2013-08-22  6:07       ` Kent Overstreet
  0 siblings, 0 replies; 2+ messages in thread
From: Kent Overstreet @ 2013-08-22  6:07 UTC (permalink / raw)
  To: Georg Schönberger; +Cc: linux-bcache-u79uwXL29TY76Z2rM5mHXA

On Wed, Aug 21, 2013 at 10:48:57AM +0200, Georg Schönberger wrote:
> Hi bcache team,
> 
> we are currently developing performance test scripts with Fio for bcache.
> At the end of a test we want to detach and destroy the cache. Therefore we did:
>     echo 1 > /sys/block/$dev/bcache/detach
>     echo 1 > /sys/block/$dev/bcache/stop
> 
>     echo 1 > /sys/fs/bcache/*/stop
>     echo 1 > /sys/fs/bcache/*/unregister
> 
> With a simple randwrite test with we had about 2GB dirty data in the cache. After detaching the dirty data flush rate was about 900KB/s from SSD to HDD.
> Is there a way to accelerate flushing of dirty data?
> What is the correct way to detach/destroy a writeback caching device?

The stop is unnecessary for the cache set (that'll shut down the cache
set without detaching anything or flushing dirty data).

900KB/sec sounds like a bug though - I'll try and take a look at it
tomorrow.

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2013-08-22  6:07 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
     [not found] <423285490.335846.1377074174264.JavaMail.root@thomas-krenn.com>
     [not found] ` <423285490.335846.1377074174264.JavaMail.root-qD/LZSXtZz5FYGZjsVlUvw@public.gmane.org>
2013-08-21  8:48   ` Flushing dirty data after bcache detach Georg Schönberger
     [not found]     ` <1057573722.337424.1377074937216.JavaMail.root-qD/LZSXtZz5FYGZjsVlUvw@public.gmane.org>
2013-08-22  6:07       ` Kent Overstreet

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox