Antonio Diaz Diaz wrote: > For example, if it were possible to make filesystem-specific tools > create a list of blocks to be rescued, then ddrescuelog could be > modified to create a domain-logfile from the list and ddrescue could > rescue the produced domain in a single call.
It was easier for me to just tap existing tools. For NTFS there is ntfscluster command which can be used to create a list of blocks to rescue, but it needs the inode number of the file, so we need enough meta-data to do ntfsls and find the number. I will probably write a script generating ddrescue calls from ntfscluster output some day in the future, as the strace method is slow / inefficient for multi-GB files. But the strace method will still be useful for meta-data recovery. >> I've also been toying with another idea - an NBD server which copies >> missing sectors with ddrescue. > > I don't get it. Ddrescue does not write anything to its output in case > of a read error (missing sector?) in its input. I meant "missing" as "not yet copied". The NBD server would serve read requests from the destination drive, but first invoke ddrescue to copy any sectors which haven't been copied yet. So you could mount / browse already recovered data, and automatically trigger recovery for anything that's missing. Regards _______________________________________________ Bug-ddrescue mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bug-ddrescue
