Let's make Time Machine

[BDNiner]

There is a restore Tab in the application. That you can use to restore specific files. I have never actually had to restore a backup though. So how it works in practice I don't know.

[BDNiner]

Um - I disagree - I started writing a Time Machine clone, complete with plans to add a time slider into Nautilus if a backup existed on a given folder hierarchy. I stopped after my laptop was stolen, and I realised I did not have a backup of my code to date (oh the irony).

I will start re-writing it pretty soon - I don't need donations (although they would be nice) - but I would need help to do all the change management stuff - I used to be able to wrap my head around it - but it confuses the hell out of me now.

Do keep us informed. I am also interested in this since it doesn't look like fwbackups is being developed anymore.

[Silverflyer]

Backintime works pretty well once set up.

[Q-collective]

Um - I disagree - I started writing a Time Machine clone, complete with plans to add a time slider into Nautilus if a backup existed on a given folder hierarchy. I stopped after my laptop was stolen, and I realised I did not have a backup of my code to date (oh the irony).

I will start re-writing it pretty soon

Yes, please keep us posted. Your specs sound pretty good!

I don't need donations (although they would be nice)

I keep my 50 euro's in check

but I would need help to do all the change management stuff - I used to be able to wrap my head around it - but it confuses the hell out of me now.

What concretely do you need?

Also, do you consider rsync as a backend? Using its differences function would make any backup app very efficient on disk space as it would only store the differences that are made on the files and directories.

[Tony Flury]

Yes, please keep us posted. Your specs sound pretty good!

I keep my 50 euro's in check

What concretely do you need?

I need someone to guide me through setting up an easy to use environment where I (and eventually others) can check files in and out etc. As I say I used to know my way around sccs, but I know stuff has moved on considerably and I will need to create project files etc online.

Also, do you consider rsync as a backend? Using its differences function would make any backup app very efficient on disk space as it would only store the differences that are made on the files and directories.

I was thinking about using a zip file backend - every backed up file will have it's own zip file at the backup destination. I definitiely did not want to be backing up entire folders on the fly. My idea is when I need to back up a file :

1) Grab entry from the queue and identify the zip file.
2) Rename old zip file to a backup version
3) Copy backup file back to correct version
4) Add file to be backuped up zip file - with the time stamp as the file name in the zip file.
5) remove the backup
6) Mark the queue entry as closed

Thus any interruption between 1 and 4 would result in a still open queue entry and a intact backup version - i.e. a failsafe version.
An interruption between 5 & 6 would result in an open queue item but with no backup (again easily detectable and resolvable).
I am assuming that file renames are atomic (i.e. a simple change to a directory and therefore high unlikely to be interrupted). The long bits of the process are items 3 and 4.

[markbl]

I've been using rsnapshot for a couple of years to backup my linux pc and my wife's windows pc. Simple and works really well.

[Tony Flury]

I've been using rsnapshot for a couple of years to backup my linux pc and my wife's windows pc. Simple and works really well.

The problem with rsnapshot is that it can take a long while to do a full backup (specially over a network), and then you end up with a massive archive which is almost impossible to open. Also any interruption (power, network etc) corrupts your backup arhcive and wastes all of that time.
I was using a similar tool - and ended up having to leave my laptop on for over 24 hours to do the initial backup, only to end up with archive I could not restore anything from, because it was too big to open across the network.
The rationale for what I was trying to write was to end up with much smaller archives, and to only ever do incremental backups, and crucially build in protection from interruptions. With my scheme the backups run in the background, but if i shutdown for the night - at most I end up with one or two partial backup files which can be recovered (deleted and re-built). If my network dies (for instance a power cut), again i would loose the backup for one or two files, not my whole archive. Restoring becomes easier due to the need to restore one file from a small archive, rather than one file from thousands in a massive archive.

[canhoto]

Today, I've found out about Redo Backup & Restore, which is a livecd that simply clones your hard drive to another drive. It is simple and restoring is also pretty straightforward. But I don't believe it is incremental, so it'll just clone your entire drive every time, wasting huge amounts of space on your backup drive. Also, rebooting for a restore is a very good idea, but I'm not too sure about the need for rebooting just to make a backup...

Unfortunately it doesn't run on ppc, apparently.

[canhoto]

Backintime works pretty well once set up.

Yes, I'm using Backintime.

However, I've found two problems:

- It takes a long time to make each new snapshot

- It doesn't run by itself. I have to launch it if I want it to run the scheduled backups. Once set up, it should run by itself, I souldn't care about it. At least that's how Time Machine works.

I can always make it launch when I start up the computer, of course. But I still have the slow snapshot's issue.