Thank you to Mark. These instructions were exactly
right. I believe that the answer is turning out to be
that once I reached the max count for the lowest-level
runs ("daily" for me, rather than "hourly"), so that
deleted and replaced files in the earliest snapshots
have started rolling off, the total space started
slightly shrinking, and now it's roughly holding
steady -- exactly as I would expect.
Also, I slightly suspect that df doesn't exactly
report filespace consumption the way that, say, du
does. I didn't use du just to save the time of
shlogging through the root tree. I get the impression
that df reports a figure mixed up with inodes and
things that a data geek like me doesn't have a good
grip on.
On Sun, 11 Oct 2009, Mark Foster wrote:
mark> Bump the verbosity and the files backed up
mark> should appear in the rsnapshot.log
