RAM usage on large files/folder


I am currently running Nextcloud on my raspberry pi 4 which has 4Gb RAM. I noted important RAM usage when uploading large files. I guess this is ok (php storing file in ram…). But now I wonder if I could migrate my nextcloud instance to a real NAS such as a Synology DS218+.

Here is my question :
Since the Synology would have 2Gb RAM, will I face problems with large files/folders transfer (>2Gb, >4Gb…)?


Can you post an image of top? You might just be seeing kernel cache usage.

Well, I cant right now, but I noted this happens when i upload a large file (for instance 2-3gb). I guess this is php putting this file in RAM before writing it to the disk?
This happens aswell when downloading large folders --> but this is php zipping this folder so that’s normal aswell, i guess

But then you mean that 2Gb RAM shouldn’t be a problem for transferring >4gb files right?

I am by no means an expert, but afaik RAM is in fact the limiting factor for browser-based uploads (and moving of files). I ran into this limitation recently since my odroid hc2 has 2 GB RAM. The desktop client uses a different method, so RAM is not the limiting factor. A way to solve this RAM based limit is using the app “flowupload”, this allows you to upload files larger than your RAM.

Yes, I discovered flow upload last week, and this is totally game changer. To be honest, this plugin is insane and solved me a lot of problems.
However, I guess i would still get the ram problem when downloading folders larger than ram

no, downloading via browser works fine in my experience.

Even if you try to download an entire folder bigger than your actual amount of RAM?
Btw, are you running NC on a 32 or 64bits instance?

I tried downloading files several times larger than my RAM, it works. The Odroid hc2 is a 32 bit platform, afaik.

I mean, if its a simple large file, then nextcloud doesn’t need to zip the file. However if its a big folder, it needs to .Zip it and I’m pretty sure you would get an error

yes, that’s sound reasonable! Single files, which are larger than RAM, however work well