4 ms·
Ask HN: What's your preferred open-source backup solution?
I'm currently using backuppc [1], which I love because it can easily do both my servers (Linux) and laptops (OSX), isn't too complex and does great pooling / deduplication.
It's also totally unmaintained, so it's time to move on.
What do you use, what should I use?
Huge preference to open-source tools, I'd like to host this on my own NAS (Linux-based).
[1] http://backuppc.sourceforge.net/
- drakmail 12y agobackupninja
- nodata 12y agoobnam is pretty good.
- DanBC 12y agohttp://obnam.org/ http://obnam.org/
- fundamental 12y agoHm, first time I've heard about this tool and I have searched around for backup solutions similar to it a number of times. It strikes me as a project that needs better word of mouth advertising, though it might just not have as big of a userbase due to the fact that it is relatively young. Have you observed any major drawbacks in your use? After going through the documentation, the only major turnoff that I see is that it fails to provide explicit documentation on the internals (ie when a backup occurs A->B->C happens to your data using formats very well documented in X, Y, & Z)
- pwg 12y agorsnapshot
- DanBC 12y agohttp://www.rsnapshot.org/ http://www.rsnapshot.org/
- fundamental 12y agoCurrently rsnapshot, however I'm considering a move to duplicity (perhaps via http://chrispoole.com/project/horcrux/ http://chrispoole.com/project/horcrux/ ) in order to get the additional feature of offsite encrypted backups
- nodata 12y agoBeware of huge cache directories.
- rubenv 12y agoWith rsnapshot of duplicity?
- fundamental 12y agoI'm pretty sure that he means for duplicity. I'd imagine that this comes into play when duplicity puts everything into a single file archive (tar IIRC). I haven't seen any issues with rsnapshot in regards to this.
- nodata 12y agowith duplicity.
- LarryMade2 12y agoFor the servers I use a bash script utilizing TAR activated by a chron job. If I were going to go deeper I'd be looking at bacuppc Beside the maintenance regularity on it, what's wrong with it?
- JustinGarrison 12y agoI've been using duplicati[1] for a while now. I usually just backup via ssh to a server(s) for multiple copies but it can also use cloud storage providors (S3, Google Drive, Box, etc). It's locally encrypted so it really doesn't matter where the data ends up. [1] http://www.duplicati.com/ http://www.duplicati.com/
- Nowaker 12y agoBacula. http://bacula.org/ http://bacula.org/ At my company we are using it for server backups. Linux desktop would do too. Google says it's possible to use Bacula Client on Mac, but you'd need to figure out yourself. Bacula is actively developed. http://www.bacula.org/git/cgit.cgi/bacula/log/?h=Branch-7.0 http://www.bacula.org/git/cgit.cgi/bacula/log/?h=Branch-7.0 No risk of being unmaintained at some point, since they build commercial stuff on top of community Bacula.
- Someone1234 12y agoBecula is fine on Linux [server], but I'd never run it on Windows Server again, it has no effective IO throttling (just number of concurrent jobs throttling, which is ineffective). On Linux you can fix it using security groups with limitations placed on IO specifically (at the kernel level). It is better at several large files than thousands of small files. In fact when you throw millions of small files at it it chokes pretty bad (CPU skyrockets). So if you need to backup an SQL server data store you'll be just fine, but thousands of XML files? Maybe not.
- Nowaker 12y agoThanks. OP wasn't interested in Windows. I have never been either, so I didn't even know. Thanks for the info. Regarding the small files performance - thanks for pointing it out. We are using Bacula to backup JIRA & Confluence data stored directly on the disk as files. Some customers have thousands of small attachments in their instances. I haven't seen any performance problems though. Maybe because not only there need to be thousands of small files, but also they need to change often, which is not a case for us. Will keep this in mind. Thanks.
- derekp7 12y agoIf you don't mind, I'd like to plug my own open source tool, Snebu (http://www.snebu.com http://www.snebu.com). It is a snapshot-style backup system which uses a sqlite database for its data catalog (stores the list of files/metadata associated with each backup set). Individual files are stored using SHA1 checksums for the file names, so you get file-level dedup, even across multiple systems. I've just got it to a point where I've been dogfooding it for a year or so, and feel comfortable promoting it more, but I'm not sure how to go about doing that (other than packaging it for various distros, and submitting it to them).
- ThatGeoGuy 12y agoI actually wrote an article 8 months ago on how I do it (https://thatgeoguy.ca/blog/2013/12/26/encrypted-backups-in-debian/ https://thatgeoguy.ca/blog/2013/12/26/encrypted-backups-in-d...). Granted, this is for personal use, and really relies on cron / rsync to work well. Let me know if you have any questions about how I got it to work (for the most part, the biggest issue I see you encountering is encrypting drives on OSX).
- mobiplayer 12y agoDéjà-vu because it uploads the backups to my Cloud Files account at Rackspace (I have a private container for this)
- rakoo 12y agobup (https://github.com/bup/bup/ https://github.com/bup/bup/) is good, although I haven't personnally tested the restoration.