no, don't copy NC's data directory into OpenCloud's storage with rclone (or cp, or rsync). It will look like it worked and then nothing will be visible in the UI.

Why: OpenCloud doesn't store files in a human-browsable user/files/... tree the way Nextcloud does. It uses the decomposedfs backend — every file/space has a UUID, and the actual content lives under nodes/<xx>/<xx>/<uuid> with separate metadata (extended attributes + sidecar files) describing the tree, owners, share state, ETags, etc. Files dropped onto disk without those entries are invisible to OpenCloud, and hand-crafting the metadata is not realistic.

That's exactly why the official guide tells you to go through rclone over WebDAV — the WebDAV PUT is what makes OpenCloud create the node, assign a UUID, write the metadata, and index it. The NFS share is the destination filesystem under the hood, not an API.

So the realistic options are:

1. Use the documented rclone WebDAV → WebDAV path. Run it on the NC or OC host (or anywhere with network to both), e.g.:

   rclone copy \
     --webdav-url https://nc.example.tld/remote.php/dav/files/<ncuser> \
     --webdav-vendor nextcloud --webdav-user <ncuser> --webdav-pass "$(rclone obscure <pwd>)" \
     :webdav: \
     :webdav,url=https://oc.example.tld,vendor=owncloud,user=<ocuser>,pass=$(rclone obscure <pwd>): \
     --transfers 8 --checkers 16 --fast-list

   Loop it per user. Slow-ish, but it's the only path that preserves user/space ownership and makes files actually appear.

2. Local shortcut, same idea: mount the NC user's files directory locally (you said you already have NFS access) and rclone copy /mnt/nc/<user>/files :webdav: into that user's OpenCloud space. Same effect, no NC server round-trip, still goes through OC's WebDAV so metadata is created correctly.

3. Shares, comments, tags, versions, calendars/contacts: rclone copies file bytes only. If you need any of that, plan for it separately — there is no first-class NC→OC migrator that handles it today.

TL;DR: keep the NFS share as your backing storage for OpenCloud if you want, but ingest the data through WebDAV, not by writing into the storage directory.

Why: OpenCloud doesn't store files in a human-browsable user/files/... tree the way Nextcloud does. It uses the decomposedfs backend — every file/space has a UUID, and the actual content lives under nodes/// with separate metadata (extended attributes + sidecar files) describing the tree, owners, share state, ETags, etc. Files dropped onto disk without those entries are invisible to OpenCloud, and hand-crafting the metadata is not realistic.

DecomposedFS is not the default. The default is the POSIX filesystem mode where the backend file tree is the same than on the frontend.
https://docs.opencloud.eu/docs/next/admin/configuration/storage/storage-posix