On Wed, Jul 1, 2020, at 9:43 PM, Neal Gompa wrote: > On Wed, Jul 1, 2020 at 9:27 PM James Cassell > <fedoraproject@xxxxxxxxxxxxx> wrote: > > > > > > On Wed, Jul 1, 2020, at 9:03 PM, Przemek Klosowski via devel wrote: > > > On 7/1/20 3:50 PM, Josef Bacik wrote: > > > > This sounds like a "wtf, why are you doing this btrfs?" sort of thing, > > > > but this is just the reality of using checksums. It's a checksum, not > > > > ECC. > > > > > > Yes, exactly---why isn't it ECC? Wouldn't it work better, especially in > > > the context of faulty hardware? > > > > > > I do realize it would require changing the on-disk format, and maybe > > > slow the critical path... > > > > > > > Or maybe make all metadata raid 1, even on single disk set up? > > > > Not that isn't interesting, but what would be the mirror target on a > single disk setup? > The idea is that the second copy of metadata on the same disk might be readable in case the first copy has a checksum error, in case of fault hardware. I haven't tried it, but I'd gladly give up a little space for more robustness, especially if btrfs is sensitive to metadata corruption by the hardware. If btrfs demands a separate device for raid1 metadata, I wonder if a small 1G partition could be dedicated for purely mirrored metadata use. V/r, James Cassell _______________________________________________ devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@xxxxxxxxxxxxxxxxxxxxxxx
