On Wed, Mar 24, 2010 at 10:47:27AM -0700, Russ Price wrote: > > On Tue, March 23, 2010 12:00, Ray Van Dolson wrote: > > ZFS recognizes disks based on various ZFS special > > blocks written to them. > > It also keeps a cache file on where things have been > > lately. If you > > export a ZFS pool, swap the physical drives around, > > and import it, > > everything should be fine. If you don't export > > first, you may have to > > give it a bit of help. And there are pathological > > cases where for example > > you don't have a link in the /dev/dsk directory which > > can cause a default > > import to not find all the pieces of a pool. > > Indeed. Before I wised up and bought an HBA for my RAIDZ2 array > instead of using randomly-assorted SATA controllers, I tried > rearranging some disks without exporting the pool first. I almost had > a heart attack when the system came up reporting "corrupted data" on > the drives that had been switched. As it turned out, I just needed to > export and re-import the pool, and it was fine after that. Needless > to say, when the HBA went in, I made sure to export the pool FIRST.
In my limited testing (with an HBA based system), I've been able to move drives around without exporting first... but sounds like good practice just to export anyways to be on the safe side. :) Thanks, Ray _______________________________________________ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
