On Sat, 2006-03-11 at 16:15 -0800, dean gaudet wrote: > On Sat, 11 Mar 2006, Ming Zhang wrote: > > > On Sat, 2006-03-11 at 06:53 -0500, Paul M. wrote: > > > Since its raid5 you would be fine just pulling the disk out and > > > letting the raid driver rebuild the array. If you have a hot spare > > > > yes, rebuilding is the simplest way. but rebuild will need to read all > > other disks and write to the new disk. when serving some io at same > > time, the rebuilding speed is not much, > > > > but if i do a dd clone and plug it back. the total traffic is copy one > > disk which can be done very fast as a fully sequential workload. with > > that bitmap feature, the rsync work after plugging back is minor. > > > > so the one disk fail window is pretty small here. right? > > you're planning to do this while the array is online? that's not safe... > unless it's a read-only array... what i plan to do is to pull out the disk (which is ok now but going to die), so raid5 will degrade with 1 disk fail and no spare disk here, then do ddresue to a new disk which will have same uuid and everything, then put it back, then bitmap will shine here right? so raid5 is still online while that disk is not part of raid5 now. and no diskio on it at all. so do not think i need an atomic operation here. > > if you've got a bitmap then one thing you *could* do is stop the array > temporarily, and copy the bitmap first, then restart the array... then > copy the rest of the disk minus the bitmap. > > you basically need an atomic copy of the bitmap from before you start the > ddrescue... and you need to use that copy of the bitmap when you > reassemble the array with the new disk. > this raid5 over raid1 way sounds interesting. worthy trying. > or you could stop the raid5, and make a raid1 (legacy style, without raid > superblock) of the dying disk and the new disk... then reassemble the > raid5 using the raid1 for the one component... then restart the raid5. > > regardless of which method you use you're going to need to take the array > offline at least once to reassemble it with the duplicated disk in place > of the dying disk... > > i think i'd be tempted to do the raid1 method ... because that one > requires you go offline at most once -- after the raid1 syncs you can fail > out the dying drive and leave the raid1 around "degraded" until some > future system maintenance event where you can reassemble without it. (a > reboot would automagically make it disappear too -- because it wouldn't > have a raid1 superblock anyhow). > > -dean - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
