On Fri, 16 May 2008, Jonathan wrote: > To confirm the RAID1 matters because RAID 10 is RAID 0 (stripe) over > RAID 1 (mirrors)? > > I don't mind having to revert to BIOS for rebuild in the event of a > hd fault, but I wouldn't mind understanding a little more about the > lack of error handling. > > Do you mean if one disk in the underlying mirror set fails that > dmraid will 1) ignore and not notify > 2) fall over in a big heap > 3) notify, keep running but not rebuild > 4) not handle inconsistencies without reversion to BIOS? Read doc/dmraid_design.txt - dmraid can't update the RAID metadata to record a disk failure. I am not sure what happens in the event of a disk failure - up to devmapper I guess. I would say that if you had a degraded array and rebooted the BIOS wouldn't know it's degraded and DM wouldn't either, so you might fail to boot, boot but get corrupted data, etc etc.. -- Daniel O'Connor software and network engineer for Genesis Software - http://www.gsoft.com.au "The nice thing about standards is that there are so many of them to choose from." -- Andrew Tanenbaum GPG Fingerprint - 5596 B766 97C0 0E94 4347 295E E593 DC20 7B3F CE8C
Description: This is a digitally signed message part.