NFS Share between NAS and 2 clients - why DOES it work?

I know the title may seem confusing but let me explain. I am using OMV 7 on a NAS and exported a NFS share without modifying any options:

On my clients I created the following mount points and /etc/fstab entries:

Nobara:

192.168.178.4:/nfs_share /mnt/nfs-share nfs defaults 0 0

Fedora Kinoite:

192.168.178.4:/nfs_share /var/mnt/nfs-share nfs defaults 0 0

The users on both clients have the same username (which is different from the OMV user) and do own the mount points (UID:GID). UIDs are the same, GIDs differ on the two clients:

Nobara: UID=1000, GID=1001

Kinoite: UID=1000, GID=1000

The exported folder "nfs_share" is owned by root and group users which has GID 100.

After mount -a I expected to see some error messages but there was nothing to see. The nfs share is mounted on each client and is RW.

Back to the title of the topic: why does this work? It´s nice it does - but I expected it not to which means I don´t fully understand NFS

Any idea?

Quote from andrzejls

although I would not use "/var" to mount remote nfs share(s). I, personally, use "/media/nfs/"sharename" on the client machine.

Thank you for your reply.

Since I was just playing around I chose /mnt for its temporary character over /media or any other directory. Regarding the mountpoint on /var:

Quote from Fedora Docs

On Fedora Kinoite, the root filesystem (/) is immutable. The /usr directory and everything below it is read-only.

The /etc and /var directories are respectively used to store configuration files and runtime state and are thus writable. Symlinks are used to make traditional state-carrying directories available in their expected locations. This includes:

• /home → /var/home
• /opt → /var/opt
• /srv → /var/srv
• /root → /var/roothome
• /usr/local → /var/usrlocal
• /mnt→ /var/mnt
• /tmp → /sysroot/tmp

Quote from andrzejls

Why did you expect not to work. NFS was design and implemented originally by SUN Microsystem for their UNIX OS.

You have install "nfs-common" client

I am wondering why my clients have read-write access on the share. The UIDs and GIDs don´t match and I did not set any options like all_squash, anonuid, anongid, etc. Since mounting works I expected to only see the share but with read-only at best.

Are there any export options set by default not visible in the GUI?

Quote from andrzejls

I also use "sync,no_subtree_check,insecure,all_squash,anonuid=1000,anongid=100" in "Extra Options" field.

This is exactly what is confusing me. I would expect my clients to have rw-access on my NFS share with these options set even if UID and GID are different between client and server. But I don´t use these options and my clients still do have rw-access.

Can anyone point me in the right direction?

I think I solved it. It is the way NFSv4.x handles permissions. Versions prior to 4 worked the way I had in mind. I wasn´t aware of the changes in v4.x:

Quote

Added clarifications and error conditions for the handling of the owner and group attributes. Since these attributes are string based (as opposed to the numeric uid/gid of previous versions of NFS), translations may not be available and hence the changes made.

After double checking names, UIDs and GIDs I found that I randomnly created a case which works.