Tim Preston writes:
> Greg Oster wibbled one day
> 
> > If swap partitions are sucessfully removed then RAIDframe will happily 
> > unconfigure the RAID set and mark it as 'clean'.  RAIDframe doesn't care wh
> at 
> > sort of data is on the RAID set, and doesn't have any mechanisms for treati
> ng
> > swap as 'special'.
> 
> On a, slightly, similar note. 
> 
> IS there any work in progress to make the kernel able to dump core to a
> swap device on a RAID set? At the moment this doesn't work, at least not
> for me =)

Kernel dumps typically use a slightly more "braindead" way of talking to the 
disks (e.g. polled IO) and there isn't such an equivalent in the RAIDframe 
driver...  In particular, RAIDframe won't really know what has caused the 
crash, and only knows how to talk to the underlying devices via the 
VOP_STRATEGY() stuff.  And if it's the driver for an underlying device (or, 
heaven forbid, part of RAIDframe itself!) that has experienced the problem, 
then the dump a) at best wouldn't work or b) could mess everything up in a 
big hurry.  I'd like to think that dumping to a separate, unused partition is
the safest strategy if you need/want to keep the crash dumps... the last thing 
you want is to be attempting to use bits of a driver that have crashed, and 
may have left things in some sort of wonky state...

Later...

Greg Oster