I really love Nextcloud over Owncloud. Especially the functional Android App. Way to go! I am using it as a way of Backing Up my taken images and videos when I get home to my local network.
I am using it on a Banana Pro with an external HDD. I was planing on linking some files from the HDD into the files folder of Nextcloud. So I googled nextcloud sym links and stumpled upon these posts:
and some others.
People there mentioned that Owncloud/Nextcloud is not meant to be a backup system.
Can someone tell me why?
I always though of Owncloud/Nextcloud as cloud like Onedrive or Dropbox but just self hosted. Why is it considered a bad practice to back up images from my phone with Nextcloud or in general back up things with Nextcloud?
A backup system should be simple, fast and should allow you to do differential backups. Nextcloud is huge, complex (take a look at the source) and relatively slow due to PHP which is perfectly fine for usecases such as syncing and sharing files.
Think of Nextcloud more like a tool that lets you quickly share a file with a friend or co-worker or syncing some files from your computer to your phone. Nextcloud is an absolutely HORRIBLE choice for uploading your 100gb+ music collection or your filesystem backup.
You just need to backup your Nextcloud, then you are fine
A single disk or single device can easily fail, in synced solutions there is a great risk that you distribute the broken files (or even delete them) on all your devices.
If you have a slow upload, large file collections take a lot of time to sync. If it is on your home network, I don’t see much of a problem here. Then it’s perhaps related to the audioplayer with its limited indexing performances.
If you don’t want to share it with external users, solutions like syncthing or git-annex perform better on large file collections.
Not being a backup system is not a good excuse to destroy applications passing over it. Many applications will include symlinks to system libraries, I’m very much with the group that thinks these should be preserved and never dereferenced by Nextcloud. Then on the other side I can at least see where they pointed to on the other system to figure out where I need to point them to on the system they have been copied to.
Also, 100GB+ is not a big deal for my Nextcloud install. It handles 300GB+ of music, 1 TB+ of video files and over 3TB of images without breaking a sweat. It is perhaps not good if your Nextcloud install is being handled by an old laptop, or Raspberry Pi, microserver, etc… but if you’re running it on a full blown modern desktop or server machine it is totally fine. You do not have to constantly keep all folders synced, I only need my work folder (~100GB) synced all the time.
Don’t get me wrong, I’m not trying to use Nextcloud as a backup, rather I want it to be the primary location where my files live, but that just isn’t possible if it doesn’t keep symlinks and that is a very unwelcome thing to have seen in the logs now AFTER having transferred more than 5 TB because from now till eternity every time I fetch something from Nextcloud I’ll be playing roulette for whether it will be missing files or not and the problem is that it won’t be obvious if they are missing or not because Nextcloud didn’t even leave links themselves.
Hmm, maybe. I haven’t tried yet, but maybe mounting it as WebDav allows it? I will take a while for the copy to sort out which files are missing and copy them there, but much less time than copying all the data again to try to make sure. Your comment about the storage being in /data was sort of the missing link for me to understand why people are concerned about symlinks that are not dereferenced at all. Still, any exploit there would depend on an already leaky server setup.
Edit: Maybe not though, I recall I couldn’t create a zipped file on the WebDav directly, so I assume it has the checking mechanisms where it could block a symlink… I’ll try anyway though.
