Regulars will recall the thread back in March-April about the /etc/oldmaps files and the way that ONTAP 6.x uses them (or is supposed to) to make sure files moved into a tree being dumped get included in incremental dumps even if their ctime/mtime's haven't been updated. And they may even remember that my last contribution (2 April) was
So either I didn't understand after all, or this splendid new facility has bugs in it ... :-(
I pursued this with NetApp, and Stephen Manley did some great work on it, and the problem I was having is now understood. It's burt 45289 if you want to add it to your BugWatch list. It arises if you use "dump" on a path which is a whole volume or qtree, but also use the "l" option so that only part of it is actually being dumped. If it wasn't for the "l" option, nothing could be moved into the area being dumped without it being given a new ctime, and so in that case ignoring the /etc/oldmaps files would be a correct optimisation (and that's what it does), but it becomes a false one with "l" present in the options.
Many thanks again to NetApp support for their help in tracking this down.
Chris Thompson University of Cambridge Computing Service, Email: cet1@ucs.cam.ac.uk New Museums Site, Cambridge CB2 3QG, Phone: +44 1223 334715 United Kingdom.