Re: request baset device mapper in Linux

[Lukas Hejtmanek <xhejtman@ics.muni.cz>]

On Sat, Jul 23, 2011 at 04:28:06PM +0900, Jun'ichi Nomura wrote:
> How fast is the single disk performance?

about 120MB/s if I run dd on a single disk only. 

> Could you check /proc/interrupts and /proc/softirqs and
> see how they are distributed among CPUs?

I don't see anything strange except function call interrupts hit single CPU
(that ksoftirqd). 

> As for the memory usage, what happens if you add 'iflag=direct' to dd?

then ksoftirqd does not eat significant amount of CPU.


> Also, is it possible for you to try the attached patch?
> I would like to know whether it changes the phenomenon you see.
> This patch should reduce the number of calls to recalc segments.
> If it is the root cause, the patch should fix your case.
> The patch is generated for 3.0 but should be easily applicable to
> other version of request-based dm.

the patch fixed the problem. I didn't test whether it broke something else or
not, though.

> As Kiyoshi suggested, it is important to know whether this
> problem occurs with the latest kernel.
> So if you could try 3.0, it would be very helpful.

so, do I need to check recent kernel as the patch works for me?

-- 
Lukáš Hejtmánek