Hi cheesemarathon, and thanks for your quick response.
I think the cache setup itself is not the issue. I also think that I get how it works (not meant offensively).
I copied from an external samba share to /mnt/usr/media (media is my media share, set to "high water" with 2 disks). So I did not copy directly to a specific disk. The cache was used (as was intended), but when the cache filled up to 100%, the copy processed stopped with an error. I would have expected that the copy process just continued with writing directly to the actual disks (which would be normal behavior for a copy process to a cached array).
The question is... when is it determined where the files are actually written onto?
Start of the copying process for all files
Whenever a new file is copied (that would be the expected behavior).
If 1) is the case, then that probably explains my issue: The first files would fit onto the cache, but the rest won't, so an error occurs.
Say you start a copy process for 5TB, each file is 100GB and the target is large enough, but with a 200GB cache. When you normally copy something to the array, the first 2 files are copied to the cache and the rest is written directly to an array disk.
With Cloud Commander, it seems like it wants to copy everything where the first file would fit, circumventing (skipping?) the balancing check that usually takes place for each file when you copy something to an unraid share via SMB (e.g. from a Windows PC).
It may of course be possible that copying to /mnt/usr/media is not the correct way to do it if I want to use the balancing feature.