On Wed, 2009-01-14 at 03:20 +0100, Max Laier wrote: > On Wednesday 14 January 2009 02:10:55 Jeremie Le Hen wrote: > > Dear Kostik, > > > > On Tue, Jan 13, 2009 at 11:23:55PM +0200, Kostik Belousov wrote: > > > What are the supposed advantages of this approach over the dumpdev > > > variable in rc.conf ? I see that having whole partition usage > > > configuration in fstab is natural, so the idea of the patch is > > > probably right. > > > > This is exactly the point. fstab(5) will describe every partition > > usage. > > > > > One the other hand, is it possible to enchance > > > this to allow specification of the swap partition that is also > > > a dump partition, in fstab ? > > > > Well, actually I think it makes more sense to automatically add "dp" > > device as swap. You will find an updated patch attached where swapon(8) > > has been modified to implement this behaviour. Is it better now? > > I don't agree here. The point of having a dedicated dump device could be to > not overwrite the state of the swap when dumping core. In addition, a dump > device could double as a place for suspend to disk if/when we implement this - > in this scenario it would also be required that the dump device does not hold > valuable swap data. I second this: I have some machines which are SCSI throughout, but have a single slow IDE disk reserved as a dump device (as dump isn't happy with the SCSI controller). I wouldn't want it to be treated automatically as a swap device though, as the IDE disk is significantly slower, and not mirrored. GavinReceived on Wed Jan 14 2009 - 13:18:43 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:40 UTC