On 2015-08-05 22:13, Chris Murphy wrote:
On Wed, Aug 5, 2015 at 6:45 PM, Paul Jones <paul@xxxxxxxxxxxxxxx> wrote:Would it be possible to store this type of critical information twice on each disk, at the beginning and end? I thought BTRFS already did that, but I might be thinking of some other filesystem. I've had my share of these types of oops! moments as well.That option is metadata profile raid1. To do an automatic -mconvert=raid1 when the user does 'btrfs device add' breaks any use case where you want to temporarily add a small device, maybe a USB stick, and now hundreds of MiBs possibly GiBs of metadata have to be copied over to this device without warning. It could be made smart, autoconvert to raid1 when the added device is at least 4x the size of metadata allocation, but then that makes it inconsistent. OK so it could be made interactive, but now that breaks scripts. So... where do you draw the line? Maybe this would work if the system chunk only was raid1? I don't know what the minimum necessary information is for such a case. Possibly it make more sense if 'btrfs device add' always does -dconvert=raid1 unless a --quick option is passed?
Perhaps we could print out a big noisy warning that could be silenced?
Attachment:
smime.p7s
Description: S/MIME Cryptographic Signature
