just a small suggestion for Rock-ons backup provider: Altdrive (Altdrive.com).
This backup provider has private encryption and fileversioning. Pretty much what you need with unlimited storage for a cheap cost. Also native Linuxclient.
Be Interesting if they get Funny about people backing up their NAS/Rockstor from it as it looks like it’s aimed at your home pc.
Been trying using Duplicati 2.x and Hubic (Duplicati running in a VM connecting to Rockstor over NFS, not tried running it on Rockstor itself yet) Seems to work other Than Hubic throttling to 10Mbit/s
Can I ask, how did you use Duplicati with Hubic?
If you download the version 2 preview hubic is one of the options it gives you when selecting the storage backend.
Seems stable enough thesedays and for the most important data I have other backups as well just in case something did go wrong with it.
One thing to note is Duplicati can upload to locations the Hubic webUI can’t see so if you ever want to delete that data from Hubic you’ll probably have to use something like Swift explorer (http://www.619.io/swift-explorer)
I’ve been playing with it (having downloaded a ton of mono and gtk stuff. I seem to be plagued by assorted Mono User errors, file not found and SQLLite lock errors errors at this end, It has made a change turning off metadata. Lets see how that goes.
I think maybe running Duplicati on the Rockstor box was ambitious!
I used a ubuntu 14.04 lxc container, although I think Mono packages for Centos 7 exist, isn’t there a repo for that?
My options on the backup job are as follows (Reduce memory usage + metadata doesn’t play nice with NFS permissions)
I have to confess that was hopeless for me. As broken as it could be in my environment. I have now started to try the onedrive java client I have compiled and now used on three different instances. I didn’t write it (its on github) and it makes no bones about being beta. However it works! Its will only take until Christmas to finish my photos. I don’t yet know what happens when the token expires.
This leaves me wondering why on earth I’m using Onedrive - ah yes, Its cheap and you get office with it. However, Hubic at 5 euros/month for 10TB is even better. I wonder how fast it is.
https://github.com/wooti/onedrive-java-client You will have to compile it. The provided jar is broken (see issues)
Footnote: Unless I am very much mistaken neither Duplicati will not create a folder hierarchy from this linux source. In fact from the Centos Desktop, neither will the Onedrive web interface. The onedrive java client has a switch for that. As my Picture library is deeply nested in years and months, that’s a rather big issue. Why cant these guys provide a regular sftp interface?
Onedrive has some nasty throttling once you start uploading a load of data, I know the devs of stable bit cloud drive specifically banned it from their beta because of it. Apparently it limits the api requests or something.
The business version isn’t much better (Sharepoint eww!), It’s definitely meant for uploading your photos or docs, not for trying to backup 10’s or 100’s of GB of data.
Hubic is throttled to 10Mbit/s at least some of the time, but then again having a backup of the data that takes ages to restore is still better than having NO backup at all.
When one’s photos and vids add up to more than 500Gb it’s a hassle though. I was lucky enough yesterday to have my laptop and my old hdd at a place with unlimited data and fibre. At least that got rid of the pics. Now the video…
Then we have the issue that the way these things are meant to work is mirroring a folder. That’s not what I want (and many others, I guess). The web interface wont let me pull folders of a usb drive (and fails to construct the full folder structure including empty ones properly from the root drive) , plus I don’t want to dump 300Gb of data onto an ssd for nothing.
No wonder NAS is become more popular. Cloud storage is more hassle than its worth. What I am considering is setting up an NAS at my friends house and synchronising them. Then I just have to hope there is no volcano in Auckland (earthquakes are not the risk here)!!
At least with owncloud providing you’re not using it’s encryption the files are dumped as IS to the underlaying file system, one of the reasons I went back to using it from seafile
Seafile was better at handling web uploads, and it had some cool de-duplcation stuff baked in, but it used it’s own Git like storage format and it was quite possible to corrupt it’s repo if the HDD went awol during a write update (One of my earlier attempts at a storage server used a USB3 multibay caddy for accommodating extra disks what a mistake that was)
Hubic seems ok if used with something like swiftexplorer or duplicati although it’s slow and it will take a long time to backup/restore large amounts of data, I also wouldn’t want to trust it with the only copy of my data thats for sure.
My Rockstor Install is actually offsite in a co-lo facility, I don’t have much in the way of local storage at home other than the computers themselves I was toying with the idea of building another one (Maybe on my own hardware, maybe on rented hardware) and placing it elsewhere to mirror the important data but at the moment Hubic works out cheaper.
Also the stuff I really care about is on my computer, on my Rockstor box, on Hubic and on Spideroak, if something takes all 4 of those out I’m probably being targeted.