On Wed, Dec 23, 2015 at 3:15 PM, Donald Pearson <donaldwhpearson@xxxxxxxxx> wrote: > On Wed, Dec 23, 2015 at 12:20 PM, Goffredo Baroncelli >> Ducan talked about a N-way mirroring, where each disks contains a copy of the same data. Nobody talked about N-way mirroring where N is less than the number of the available disks. >> > > Well that was certainly implied as the unimplemented solution to > dropping half the drives that the OP tested. N-way mirroring where N > = the number of drives is just Raid1 on crack and not the Raid10 > use-case that the OP is asking about. How does the OP's use case normally get implemented? For separate controllers, this would need to be software raid10, but you'd need a way to specify the drive pairings. How does mdadm create -l raid10 enable that? Or to make absolutely certain, do you put them all in a container and then first create -l raid1, and then second create -l raid0? In any case, what you get is drive level granularity for mirroring. A drive has an exact (excluding layout options, but still data exact) copy. That's not true with Btrfs where the granularity is the data chunk (1+GiB). A given drive's chunks will definitely have copies on multiple drives rather than on a single drive. And those multiple drives will variably be on both sides of a controller or drive make/model division. One of the major differences of Btrfs with all profiles is that it deals with different sized devices elegantly. That's because of the chunk level granularity. So I think that having mirrors of drives rather than chunks means that we have to have exact size drive pairings. -- Chris Murphy -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
