> > [email protected]:~/LargeFiles > zpool import -o readonly testpool
> > internal error: Read-only file system
> > Abort (core dumped)
> > [email protected]:~/LargeFiles >
> Interesting, I've just filed 6569720 for this behaviour - thanks for
> spotting this! Regardless of whether ZFS supports this, we definitely
> shouldn't core dump.
I just remember those weirdo RAID5 controllers we had way back in a
different company. It was a pretty expensive thing, with hardware XOR,
battery, cache and all that. Going by that, it wasn't a toy. Yet, when a
disk failed, it dropped into readonly mode.
I guess if someone were to have a concoction like this, ZFS would break
in various ways. That is if someone insists on using hardware RAID.
Can't remember the brand, though.
> > Does ZFS still write to the zpool when mounted with readonly?
> Yep it does - it needs to mark the pool as being in use and no longer
> exported. I think this is done by changing the pool state nvpair in
> the vdev_label (if I'm not mistaken - section 1.3.3 in the on-disk
> format document covers this.)
Solaris doesn't by chance ship with a tool to mount devices as readonly,
but writes modifications into a (temporary) file? Kind of like various
zfs-discuss mailing list