> What would the effect of this be for changing/adding disks on a
> metadevice? In my experience, systems large enough for it to be an
> issue have a disk in the RAID fail more often then a controller, or
> you need to add more capacity to the system. Using uuid's wouldn't
> help at all, since they are based on the partitions, not on the disk
> itself.
(this is not an issue for hardware-RAID, so you must be talking
software-RAID) The latest RAID driver already does 'autodetection', which
means on partition check we read the RAID superblock which contains
various info about the array. So if a disk say was /dev/sdb11 before
reboot, and /dev/sdb06 after reboot, this is not a problem for the RAID
driver, we have internal persistant numbering. (this numbering persists
even if you transfer the RAID array from one system to another one)
conclusion: every 'meta' driver has to provide persistant
self-identification anyway, especially when dealing with disk/controller
failures.
-- mingo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html