On Thu, Sep 29, 2005 at 11:07:31PM +0100, Rui Paulo wrote:
> On 2005.09.29 15:10:12 -0400, Ryan Cresawn wrote:
> | Rui,
> | 
> | Yes I am having problems. 
> 
> Sorry, my bad. Should be fixed now.
> 
> 		-- Rui Paulo

Rui,

Thanks for fixing that problem.  The pages build correctly now.

It appears to me that an important consideration with respect to the
dumpdev paragraph in 16.3.8 has been overlooked.  In the example, the
author defines a dumpdev partition of 2097152 sectors.  Earlier the
number of bytes per sector is shown to be 512.  If this is correct, we
can calculate the size of the dumpdev partition, /dev/wd1b, to be
1073741824 bytes or ~1 GB.

$ dc
2097152
512
*p
1073741824

What bothers me is that I can't find in this chapter how the author
chose the size for this partition.  It's clear in swapctl(8) that a
dumpdev should be no less than the real memory in a computer.

-D      The -D option requires that a dumpdev also be in the argument
        list.  The kernel dump device is set to dumpdev.  This changed
        is made via the swapctl(2) system call.  The dump device is
        used when the system crashes to write a current snapshot of
        real memory, to be saved later with savecore(8) at system
        reboot, and analyzed to determine the problem.

Since resizing partitions after the RAID is created is not possible,
should it be suggested that users carefully think about the total
amount of real memory their computer has and can support and then set
their dumpdev partion(s) to an appropriate value?

If I overlooked this documentation elsewhere then kindly point me to
it.

Ryan