3

So i have a backup system running that basically stores all files in Dropbox using encfs. (In addition i do a nightly rsync of the complete Dropbox folder to Amazon S3). I'm quite happy with that solution, however i worry about data corruption.

I have Dropbox running on about 5 machines (2 of them are not online all the time). While this is good for having multiple copys of my files it also introduced multiple points of failure. If any of the harddrives corrupt files for whatever reason, Dropbox will happily sync that now corrupted file accross all machines. Since there are lot's of files that i don't use very often i would probably not notice in quite some time.

So my question is: Do you have any ideas how i can prevent this ? Notice when something happens ?

I was thinking about keeping hash sums of all files in a db for the current state (assuming all files are ok NOW) and then checking them periodically but that would mean that files i WANT to change also show up, which would make it hard to manage.

2 Answers 2

1

The problem here is HOW you can detect a file has been corrupted...

You can use checksums (MD5 for example). If a file hasn't been modified and its MD5 checksum is wrong, then it's most likely corrupt.

Sadly, this doesn't allow you to know if modified files have been saved corrupted except opening them with the correct software.

5
  • An even bigger problem is that i would have to manually check that list an remember which files i actually modified and which might just be in that list due to corruption...
    – joekr
    Feb 9, 2012 at 9:56
  • You can easily automate that using a scripting language. The problem is automating DropBox itself...
    – m0skit0
    Feb 9, 2012 at 9:57
  • No i can't because the script doesn't know which files i modified - how should it ?
    – joekr
    Feb 9, 2012 at 10:00
  • Of course he knows because OS (filesystem) knows when a file has been modfied (Modified Date)
    – m0skit0
    Feb 9, 2012 at 10:06
  • Hmm that might be an idea also i will have to try and test what dropbox does if a file was modified but the modfied date has not changed.
    – joekr
    Feb 9, 2012 at 12:02
0

My solution is Apple-based: I simply let Time Machine make its incremental backups. This automatically includes Dropbox folder.

I have implemented nothing to detect corrupted files, but at least correct versions can be retrieved.

I also clone my time Machine Backup periodically.

By the way, it is worth noting that non-free offers of Dropbox include full versionning, whereas free offers provide versionning only for the last month.

3
  • Actually even for payed account you have to add the "Packrat feature for 39$/year to get more than 30days of history for your files.
    – joekr
    Feb 9, 2012 at 11:59
  • Of course there are other ways to do backups that protect you a bit better from it by doing full versioning but what i like about my dropbox setup is that i can access all the files from everywhere.
    – joekr
    Feb 9, 2012 at 12:01
  • Backups don't prevent you from accessing Dropbox files from everywhere.
    – mouviciel
    Feb 9, 2012 at 13:26

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .