[dm-devel] buffer/cache: Underlying device vs. mapper device

[Arno Wagner]

On Sat, Dec 18, 2004 at 07:23:56PM +0000, Alasdair G Kergon wrote:
> On Fri, Dec 17, 2004 at 10:17:15PM +0100, Arno Wagner wrote:
> > The data read from /dev/hdX does not reflect the write to /dev/mapper/xx 
> > or only after a lot of data ( > main memory) has been written. Waiting
> > does not help. (After 1 hour still no change.)
>  
> I've seen similar issues in 2.6 for a long time with loop devices 
> I use for testing needing to be flushed explicitly.
> (blockdev --flushbufs)
> 
> I've not managed to find time to investigate yet.
> 

O.K., so it is a problem and a known one. I can wait and use
"blockdev --flushbufs" in the meantime. Thanks for the hint.

I assume writes to the mapped device actually reach the physical
device, since otherwise people should have seen problems with
filesystems on mapper devices.

Arno
-- 
Arno Wagner, Dipl. Inform., CISSP --- CSG, ETH Zurich, wagner at tik.ee.ethz.ch 
GnuPG:  ID: 1E25338F  FP: 0C30 5782 9D93 F785 E79C  0296 797F 6B50 1E25 338F
----
Cuddly UI's are the manifestation of wishful thinking. -- Dylan Evans