<html> <head> <meta content="text/html; charset=utf-8" http-equiv="Content-Type"> </head> <body bgcolor="#FFFFFF" text="#000000"> <div class="moz-cite-prefix">On 12/5/2014 11:00 AM, John Hupp wrote: </div> <blockquote cite="mid:5481D689.5060607@prpcompany.com" type="cite"> <meta http-equiv="content-type" content="text/html; charset=utf-8"> <div class="moz-text-html" lang="x-unicode"> I'm still working on a solution for the problems I raised in the thread "A survey of GUI-based free online backup." I have swung this way and that looking for the best approach. Time and again, I have found something that is promising in one regard but undesirable in another. Here is where I am right now. Copy.com offers 15GB of free storage with a Linux client, but the client is really just a sync program, and you have to place all your files in a designated sync folder. If you don't want to do that, you could backup all your user files to the designated sync folder. Box.com offers 10GB of free storage. They don't have a Linux client, but they do support WebDAV. So you need either a backup program that supports WebDAV, or you can use davfs2 to map a local drive to the WebDAV resource, and then back up to the mapped drive. In either case you need a backup program. So with either Copy.com or Box.com, the backup program is a critical ingredient. <hr size="2" width="100%"> Notes on a couple of promising backup programs: Duplicity (say, with front end Deja Dup) has lots of strengths, but a full backup should be run periodically, and you really don't want to do that because of the difficulty of a multi-GB upload. Duplicity -- why a periodic full backup? See <a moz-do-not-send="true" href="http://www.linux-magazine.com/Online/Features/Cloud-Backup-with-Duplicity">http://www.linux-magazine.com/Online/Features/Cloud-Backup-with-Duplicity</a>, where it says: <blockquote>Thus, in principle, you could just create one full backup and then use incremental backups for the changes. The developers of Duplicity warn customers, however: Not only can a mistake in one incremental part ruin the entire backup, but restoring files takes quite a long time if the software needs to run through all the incremental backups. </blockquote> Duplicati 2 (based on Duplicity) overcomes that problem with an approach that reliably merges 100KB diffs into an existing full backup (a shameful imprecise description, but at least brief). The 100KB chunk size keeps uploads reasonable. It supports Google Drive, MS Onedrive, and other destinations. But the Linux version requires installing Mono, which can run the Duplicati C# code that also runs on .NET Framework under Windows. This is convenient for a small project to extend its platform reach, but it seems to me that it introduces security risks, since .NET Framework malware does exist. [Note: Whether it poses a real risk is still under debate. If the risk is imaginary, then Duplicati 2, though still a Preview version, is the best solution I have seen.] This may be an impossible spec, but I'd like to find a backup program that: - will run in Lubuntu without installing a load of dependencies for another environment - has a GUI (perhaps largely for the sake of Restores by average users) - doesn't require a periodic full backup - supports good encryption - supports compression - supports breaking the backup into nicely uploadable small chunks - is not buggy - preferably does auto-deletion of older backups, or supports versioning, or will send an email when a backup fails If that is an impossible spec in the current state of affairs, I'd like opinions on the best compromises. (I've been using SpiderOak, which has a nice Linux client and 2GB of free storage, but that's not much storage these days, and I'd like to hit on a better free solution that I can set up on systems that I put together for people.) </div> </blockquote> I've wearied myself nicely, studying this issue for some days. Some interesting nuggets: Rsync and rather new front end Grsync are both very actively developed. Dirvish is another front end. Rsync is of course very proven and widely acclaimed. It is nonetheless a sync tool rather than a backup tool. To use it for backup, you have to script a workaround that employs hard links e.g. <a href="http://www.sanitarium.net/golug/rsync_backups_2010.html">http://www.sanitarium.net/golug/rsync_backups_2010.html</a> and you would also have to script something to manage the full and incremental "backups." Maybe one of the front ends knows how to handle those chores, but I have not read any mention of it. Rdiff-backup is a back end that aims to provide mirror + incremental backup capabilities as part of its native design. Sounds good. But it doesn't seem to be actively developed, and there are some problems reported. There have been a couple front end projects, both of which have been dormant. But the Rdiffweb front end has a new maintainer with ambitious plans, including a 2015 new release (<a href="http://www.patrikdufresne.com/en/rdiffweb/">http://www.patrikdufresne.com/en/rdiffweb/</a>). In any case, when you are weary the thought of custom-scripting stuff (especially for a script weakling) or dealing with a buggy old back end seems extra daunting. There is extra allure for something that just works, even if it is slightly off-goal. So I returned to Copy.com, which I previewed above. It turns out that you get an extra 5GB just for installing the Linux client, so now my free allocation is 20GB. That's a lot! It also turns out that you don't have to actually move your files to the designated sync folder. Copy supports symlinks, so if in the sync folder you just create a symlink to your home folder, Copy will pick that up. Cool! Furthermore, Copy supports 30-days of file versions and also allows you to un-delete files. Quasi-backup! Copy is a product of Barracuda Networks, so though this service is relatively new, it has good backing. So for the time being, I'm running with Copy.com. </body> </html>