On Thu, 18 Apr 2002 at 1:38pm, Toralf Lund wrote > I want my data to be backed up once a week, and the dump process to run on > weekdays only, so I've specified > > dumpcycle 1 week > > and > > runspercycle 5
Looks good. > Also, I'm not interested in incremental backups right now, so I'm using > "skip-incr" in the dumptype definitions. Err, ok. But this leaves you open to the possibility of losing a week's worth of changes to your filesystems. Any particular reason you want to skip incrementals? > That's fine in many causes, but what happens if the size of some of the > filesystems that were dumped on Monday of the first week has been reduced? > I would then want some of the data that was originally scheduled for > Tuesday (or even the other days) to be considered as well, for maximum > utilisation of the tapes etc. Similarly, if something bad happened when > backing up one of Monday's file systems so that it had to be delayed to > Tuesday, I still want it to be included on the following Monday. > > I'm not absolutely sure, but it does not look like this will happen in my > configuration; it seems like directories that was backed up less than 1 > week ago will always be skipped regardless of the available dump space etc. > > So is there any way I can tell Amanda to automatically start dumps ahead > of schedule if output space permits it. Or is there anything else I can do > to get the desired effect? I guess I could try with a shorter dump cycle, > but that might mean that the same data would be backed up twice during the > same week, which I don't want. Amanda *will* promote dumps (aka do a level 0 ahead of schedule) if there's tape space. Note that amanda tries to use approximately the same amount of tape each run, not necessarily the maximum amount of tape. Amanda is also good at recovering from "bad" backups. I'm a little confused, I must admit. At one point you say that you want amanda to do dumps early if there's tape space, but then you say that you don't want the same data backed up twice in the same week. What exactly are you trying to accomplish? -- Joshua Baker-LePain Department of Biomedical Engineering Duke University
