Datum: Wed, 06 Jun 2012 08:54:00 -0500, Robert Nichols
> That is dangerously misleading.
> A) The test only verifies that the files that existed at TIME can be
> restored to their state at that time. Such a restoral might or
> might not make use of all the intermediate increments between the
> current mirror and the given TIME.
> B) There is no test that a restore to any other time can be performed
> without error. The only checksum that is verified is the one for
> the given TIME. The only mirror_metadata file (which is where the
> checksums are stored) that is read is the one for the given TIME.
> The mirror_metadata files for other times could be corrupted or
> missing, and no error will be reported.
Thank you very much for clarification, I do see that I highly
misintepreted the function and overestimated the verification process
rdiff-backup offers. I also thought that rdiff-backup always needs all
intermediate increments to restore a file, thus thinking if I can
verify at oldest state, I also verify that nothing in between is
corrupt. I guess I was wrong then.
That said, I understand the only way to actually verify all increments
is to subsequently call verify-at TIME for all given TIMES you have
increments for. Is that correct? It does not really sound
thought-through and I guess it is a very time consuming process, even
on small to midsized repositories given the typical amount of
increments is likely to be 30 days or more.