On Fri, 8 Oct 2004 19:28:54 -0700, Patrick May <patrick@xxxxxxxxxxxxxxx> wrote: > On Sat, Oct 09, 2004 at 12:01:17PM +1000, CB wrote: > > Can anyone suggest approaches for backing up a small home network (2 > > linux pcs and a linux/winxp dual-booting laptop). My criteria are: > > > > * automaticity. It just won't happen unless it does itself once set up. > > I'll probably have to trigger the laptop one manually, but even there > > I'd like user data to be regularly (or even constantly?) synced with > > backup if possible. > > > > * use of existing media: I can't buy anything new and expensive, so have > > to be able to back up either to cd's (not likely though: would need a > > stack) or an external usb hard drive. My machines only have a total of > > about 150GB of disk space. > > > > * ease of setup. Obviously I'll have to spend a bit of time to set it > > up, but I'd really like it not to take me days, and I'm no linux expert. > > > > Possible things I've glanced at include rsync (at least for user data), > > cpio, tar, backuppc (anyone used it? looks interesting), mondorescue. > > > > I'm interested in real personal experience: it seems to me that any of > > these tools *could* do what I need but I'd love to hear from someone who > > has really used them with success. > > I've been thinking about this as well for my subnet at home, currently nothing is getting backed up unless I or my fiancee remember to copy stuff to CD, USB media, whatever. We are pretty good about that though since we'd be dead w/o our business data if bad things happened (tm). My current thoughts are leaning towards the following: Setting up one machine that will act as both the file/print server and also provide authentication + home dirs for the users. Those home dirs in turn will be backed up to a dir then CD media using some combination of chron+rsynch+tar until such time as size dictates DVD or (more likely) tape. Alternately IDE hard drives could be rotated using removable bays, the larger DAT tapes are actually more expensive than HDDs. Either implementation would include offsite storage, probably weekly (daily if *really* important) if critical data is involved. I don't really care all that much about system data since using kickstart + my own notes + whatever data I backed up means it should only take an hour or two to rebuild or replace a hosed system. Looking into mondo rescue for this as well. A strategy that I have employed @ work (day job not my business) is: Friday- full backup to DAT. Monday - swap tape, Friday's full job goes offsite (2 month rotation), Monday PM full backup Tue-Thur incremental backups on to monday PMs tape (these are also on a 2 month rotation but retained securely onsite) Friday AM swap tape to run the PM full that goes offsite. Rinse repeat etc.. This has worked out well since having the fairly granular full + increment tapes onsite allows recovery of general user (usually) or systems weirdness (not all that often) spanning 8+ weeks of time while the weekly stuff that goes offsite is intended more for when we need to recover from something catastrophic (think northridge quake / 9-11) in which case losing a week's user data is, in the greater scheme of things, not all that critical. Note that this schema applies primarily to user data, we have a different implementation in place for system specific data and use redundant (almost live) mirrored systems in separate geographoc locations for the really criitical stuff. All I need to do is a. find the time to finish the homework & b. get off my sorry ass and make it happen @ home as well. - Bests, Jon ----
