>>> On 11/23/2010 at 04:48 AM, Florian Bilek <[email protected]> wrote: 
> Nevertheless I suspect a problem in chccwdev
> or in the driver as it should not corrupt the content of the disk when it
> goes ffline.

The chccwdev process does not guarantee that all buffered I/Os are written to 
disk first.  I had a customer open a problem with this just recently.  The only 
guarantee we could arrive at was to use the "oflag=sync" parameter on the dd 
command.  Someone else commented that this kills performance, but not so much 
if you use a blocksize in the megabyte(s) range, bs=1M, for example.


Mark Post

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
----------------------------------------------------------------------
For more information on Linux on System z, visit
http://wiki.linuxvm.org/

Reply via email to