Messages in this thread | | | Date | Mon, 27 Apr 1998 08:28:46 +1000 | From | Richard Gooch <> | Subject | Re: Lots of SCSI-disks, how?! |
| |
Alan Cox writes: > > > So now we have > > > mount /devfs/xxyyzz /usr > > > which uses a dynamic (16 bit) major/minor number. > > > > BAD idea. Think about nfs. You do not want to change all major/minor > > numbers just because one scsi disk is in flames and so it failed to > > boot :-). > > NFSv3 addresses a lot of these issues, up to 32bit dev_t anyway
I should hope so. Relying on persistence of device numbers across a server reboot is pretty fragile. What happens if a SCSI controller dies in a puff of smoke? On any sensible server you'll have multiple SCSI buses anyway, so in an emergency you just reallocate discs across the remaining busses. That is likely to change device numbers. The solution is to use an automounter on clients and force anyone with open NFS filesystems to reboot.
Regards,
Richard....
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |