As the owner of a server using an LSI Logic RAID card, I'm no stranger to disk corruption. It's given me a perfect (if unintentional) testbed for seeing how "restore -C" can be used to identify which files have had their data destroyed by this cheap hardware. The existing behavior doesn't call out any difference between a file whose data and metadata (e.g. mtime) differ vs. the dump and a file that only has its data different (a likely victim of cheap RAID hardware or other forms of data corruption.)
This patch prints a warning to stderr if the file and the dump differ but the mtime of the file and the dump match. When running with "-v" it also adds the atime/mtime/ctime to the "checking" line and if the file/dump data AND mtime differ prints a message saying that the change looks intentional.
Log in to post a comment.