[linux-lvm] LVM/Devicemapper kernel level breakage?

[Sven Eschenberg]

Follow up:

This behavior occurs with kernel 2.6.34.1 - with 2.6.33.5 I do not see
this irrational behavior. Though the flush threads pop up in top
regularly, they 'only' eat up 1-2% of CPU for the system. Should this go
to LKML?

Regards

-Sven


On Sun, 2010-07-18 at 21:18 +0200, Sven Eschenberg wrote:
> Hi list, I am having difficulties with one of my boxes.
> 
> Mainly this:
> 
>   PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
>  2262 root      20   0     0    0    0 R   25  0.0 836:51.10 flush-253:2
>  2363 root      20   0     0    0    0 R   25  0.0 838:28.16 flush-253:6
>  2364 root      20   0     0    0    0 R   25  0.0 916:14.92 flush-253:4
>  2365 root      20   0     0    0    0 R   25  0.0 855:02.50 flush-253:1
>  2362 root      20   0     0    0    0 R   23  0.0 916:14.92 flush-253:5
>  3090 root      20   0     0    0    0 R   23  0.0 853:27.24 flush-253:3
> 
> I was wondering what these flush processes are, I could determine they are
> kernel based threads (as far as I can see).
> Since they are called  flush, I'd asume they have to do with flushing
> buffers or whatsoever, so I came to the conclusion, that the numbers might
> be major:minor pairs.
> 
> Using lvs, I get:
> 
>   LV          KMaj KMin
>   Root        253  1
>   home        253  3
>   var         253  2
>   usr         253  5
>   usr.portage 253  4
>   www         253  6
> 
> And I certainly do not believe that there should be flush proceses hogging
> the CPU 24/7, and al those threads belong to logical volumes, or devie
> mapper devices (there are more, with a little les cpu time created by
> kpartx).
> 
> Is this a known bug? Any recommendation what to do?
> 
> Regards
> 
> -Sven
> 
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/