18 thoughts on - CentOS Backup Tools

• I’ve been using BackupPC for years. I currently have it running on a small CentOS system that mainly does backups.

  I like it because it’s agentless ( it uses ssh/rsync ). The Pooling and Data-deduping is also nice, and saves on space.

  – Derrik

• Try BackupPC first. And don’t overthink the full/incremental numbers until you understand how it compresses and pools the data. You will be able to keep much more history than you expect on line. BackupPC
  has its own mail list – you can ask there if you have any problems but the EPEL package should ‘just work’. Just look at
  /etc/httpd/conf.d/BackupPC.conf for an example of how to set the web user passwords.

• Hi Fred,

  I am using BackupPC here to backup 5 servers to a dedicated backup-machine. In parallel, I am using http://storebackup.org/ . Storebackup uses hard links to reduce space requirements.

  I have to say that BackupPC has the nicer restore tool.

  best regards
  –

• Hi Fred,

  BackupPC is really good solution for Small Office/Home office enviroment. If you don’t need anything complex but want to manage 2-3
  workstation’s backup than you can use BackupPC without any issues.

  16.05.2014 19:33, Michael Schumacher ?????:

• BackupPC works great. I’m using it to back up about 20 servers. The pooling allows you to keep many more backups online than you expect.

  A couple of things to watch for:

  1) The data directory must be on a filesystem that supports hardlinks as that is how the pooling is done.

  2) Due to the massive number of hardlinks used in the pool, it can be very difficult to backup or copy the backup server itself depending on the number of files in the pool. If you want an offsite copy, I would suggest breaking the mirrored pair, sending one of those disks offsite, and then rebuilding to a new drive. I actually have 3 drives in my raid1 setup so that there is still redundancy while it is rebuilding.

• There are a number of good ready made choices (BackupPC seems to be the most suggested so far) but you can always grow your own, it just depends on how “sophisticated” you want to get. In the office I built a backup server and a private subnet (using a second gigE interface on each server) to create a backup network. Then it was just a little scripting using nfs to connect and tar for the backups. It’s basic, it’s simple and it works.

  What ever you chose be sure that you can do recoveries without having to install the entire application again. If the application stores data in a proprietary format you can be screwed when it comes time to recover.

  If you don’t already have it, a good read on the subject is:
  Backup & Recovery by W. Curtis Preston
  http://shop.oreilly.com/product/9780596102463.do

  Good luck!

• We’re using rsnapshot.

  My colleague set it up, but I will be taking over administration soon.

  Mark

• That is kinda of what I do at home: I have a 5W openwrt(!)
  device which does the backup using rsync and hard links for the incremental crap. It backs up my fileserver and then certain hosts
  (and specific directories since users are NFS mounted). During daytime it also backups the, well, backup drive to an external drive. Not that fancy but does the job.

  There are a lot of people who like Amanda, but it might be a bit more time consuming to setup properly

• Amanda is sort-of tuned to work with tapes with the unique feature of being able to pre-estimate the sizes of full and incremental runs and pick the right mix across a set of hosts each day to fill a tape. If you don’t need that feature it probably won’t be your first choice.

  The most unique thing about BackupPC is that it has its own implementation of rsync that can work with the compressed archive files on the server and a stock remote rsync version accessing the target files. Or, it can use tar or samba to transfer the files, with all duplicate files pooled regardless of the location or transfer method. And it has a nice web interface for configuration/browsing/restores.

• One thing I really like about rsnapshot is that you can take the latest snapshot and almost literally drop it in to replace the original. This is appealing to me if, for example, the mobo on the main fileserver dies; I can simply change IP addresses, run the right daemons, and my users are back up without too much data loss and without having to wait a long time for a restore process. Is this possible with BackupPC? I
  don’t know enough about how the backend data store is organized to know if this is a reasonable use case.

  –keith

• The internal storage format on the BackupPC server is compressed files with slightly munged filenames so you can’t quite use them ‘as is’ or use the usual tools to copy them back out. However, there is a web browser view of the backups where you simply select the backup/directory/file(s) you want and can either restore them back where they came from (or to some other configured target) or download to the browser (single file or tar/zip archive). And there are command line tools to generate tar/zip images if you prefer or want to use an SSH pipeline.

  So, you do have to wait for a restore to get usable files but the process is convenient and the tradeoff is that the same disk space can hold a much longer history (or more targets) due to the compression and pooling of duplicate content matched in different locations. If you anticipate needing an instant replacement you might want a separate disk kept current with rsync and use BackupPC for those cases where something is accidentally deleted and you don’t notice for weeks.

• BackupPC looks like a good tool, but… 1) it requires I configure Apache, and this is just my personal/home workstation where I don’t really have any other use for Apache, and I don’t really feel like having to learn Apache just to do backups. 2) BackupPC config, itself, looks
  (potentially) complicated, especially since all I want to backup is my own PC, to a raid-1 drive set that’s locally connected via USB3 or esata.

  and BackupPC seems (in my advanced state of ignorance) like one of those tools you can’t use to recover without first making your system, once again, bootable, then reinstalling and reconfiguring BackupPC.

  that’s an interesting way of doing it! I have a router with USB
  capability,… but I don’t think I want my internet gateway to also be responsible for storing important data–it’s just a “thing” I have… but if I had an extra one I could certainly do that, just turn off the wireless.

  The device I plan to use (if Fedex ever gets it here–can you sense some impatience? :) ) this:

  http://www.newegg.com/Product/Product.aspx?Item=N82E16817332028

• Remember that we are talking about rpm packages here, so that complicated install process will be
  ‘yum install BackupPC
  and the config is mostly done through the web forms. I don’t think you need to know anything about apache itself other than running htpasswd to make a web login/password.

  But if you intend to run it on the same machine as the as the stuff you want to back up it probably is the wrong tool. Or the wrong idea in the first place.

  Yes, you really want it on a different system from the one with the valuable data. However, it does have an option to generate tar archive snapshots compressed/split so you take them offsite for archival storage and you can restore from those with just standard tools.

• If you think that you’ll expand out to more machines, you may also want to consider Bacula. It’s a very stable and capable solution with enterprise grade features.

• Another solution could be rsync with zfs or btrfs.

• From: Fred Smith

  You could also have a look at bacula…

  JD

• Having looked more deeply into rsnapshot, I think I’ll try it, as it appears it’ll do what I need.

  My thanks to ALL OF YOU who have replied with suggestions. Ain’t Community Great? :)

  Fred

• +1 to BackupPC.

  A word of caution – database backups should be done with their respective native tools. A colleague, was backing up /var/lib/mysql/
  thinking he could restore the db from the backup!

  — Arun Khan