Hi, On Fri, Feb 05, 2016 at 01:57:40AM +0800, Eric Lan wrote: > Every checkpoint represent a complete filesystem at a given time. When > a system corruption occurs, is it possible to mount a specified > checkpoint (when the system is still clean) as read-write to > "time-shift" to the state when the bad thing still not happened. Of > course the checkpoints newer than the "time-shift" point will be > discarded. I would also very much like to see an easy method to make the fs "forget" the checkpoints from a certain point onwards. I guess it would be enough to satisfy all the features from this mail, Eric would like to see. Mode of operation would be: 1 Notice a problem with the data on a system (e.g. update went wrong, all libraries are broken. We need to revert) 2 Verify on an earlier checkpoint, that everything is fine 3 Revert to that checkpoint, discarding all checkpoints later than that. 4 profit. I guess though that this is not as trivial as it seems. Kind regards, /jbn -- Joel Brunenberg - Troisdorf PGP: 0xAD25981C
Attachment:
signature.asc
Description: PGP signature
