|
From: Rene R. <re...@gr...> - 2004-04-25 22:41:27
|
On Sun, 2004-04-25 at 20:59, Joe Zacky wrote: > Rene Rask wrote: > > > > >X should be the number of backup runs. > >Example: Backups run on monday,tuesday,and wednesday. > >I have chosen to keep 20 incremental backups. 20/3 is 6.6 times one week > >(7 days) which gives 46.6 days to keep backups and we round the number > >up to 47 to be safe. > > > >20/3 is the number of weeks > >times 7 is the day count. > > > >I hope this makes sense. > > > > > It makes mathematical sense, but it's confusing and not intuitive. I > would never guess that by looking at the configuration screen. > Additionally, if someone changes the days of their backups, say from 3 > to 5, that shortens the retention period even though backups have only > been running 3 times per week instead of 5. If we use this method, we > should calculate the retention date and show that on the screen so > people would know what it is. A message like: "Backups will be retained for 45 days with the current setting" should do the trick. I would expect incremental to behave as I described. I guess my way of thinking about the backup is a bit different from yours, so it is good to explain this to people in an easy to understand way. > >It seems like I'm wrong here. Guess we could use the find method. But > >I'd still prefer the db method. That is better if we want to do more > >advanced things in the future. > > > > > The database method is safer and more accurate because the modification > date in the incremental archive could be manually changed, say if > someone moved the directory or restored it from a tape backup. I'll play > with the database. The 'find' method was just so easy... I suggest you take a look at the code where I use the db for searching. You basically just make a search for files older than X. Just like you can do with the current setup. It shouldn't be hard since its basically done already :) Cheers |