In case such tags conflict with one another, the usual order could apply..
Michel.
-- Michel van der Laan - michel@nijenrode.nl http://www.nijenrode.nl/~michel In your mail from 12-4-1998 you write: > > On 12-Apr-98 Michel wrote: > > > > Not having this problem now, but having run into similar annoying > > problems when adding/removing partitions and harddrives in the past, > > thus screwing up the fstab alltogether, perhaps it's an idea to have an > > option that can force a specific order of devices - making newly added > > devices always show up end the end of the /dev/hdS|/dev/sdS list... > > Perhaps a pri/sec/mas/slv & scsiID <> /dev/-order mapping file in /etc that > > > performs a remount of devices once it's accessable, or a tag in the > > partitiontable(s) somewhere ? > > > > The problem is worse because the change is in host adapter numbering so > the schemes of host/id/lun won't work anymore. > > scsi0 <---> scsi1 > > Suddenly an old partition in a backup disk drive attached to the second > adapter (scsi1) becomes root filesystem !!! > > We normally use two scsi hosts adapters in each server and normally > they are same brand (to avoid the use of different drivers and because > scsi host adpt. have a certain tendency to misbehave when they find other > host adpt. of a different brand...). So this issue is going to happen in > several computers. Besides that Linux-2.0 _and_ NT4 (I know, I know...) do > behave as expected...There have been changes in device numbering when going > from 2.0 to 2.1 (I recall the lp[01] issue) but this one is too drastic and > fundamental and, IMHO, against what is normal when you _consciously specify > in BIOS_ the order that you like or prefer. It would be the same that if > suddenly scsi-id-0 became /dev/sdb and scsi-id-1 were /dev/sda against the > conciously specified jumper position and breaking with the schematics of any > other O.S. > > I hope that default behavior become standard-2.0, and 'reverse' option remain >s > for unusual or particular configurations, etc. > > Otherwise the new PCI schematics seem to be a _real improvement_ and is nice >to > see that Linux keeps on going with the '>leading< edge of technology' > > -- > Manuel J. Galan > manolow@step.es- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu