old nfs client, new nfs host, incorrect mount option?

[Tom Horsley]

I've got a system running centos 7 hosting lots of filesystems
available for NFS mounting.

I've got a gazillion virtual machines that need to have their
fstab edited to mount these filesystems.

I've now discovered that mechanically editing the fstab doesn't
work. On most of the VMs, if I say something like this:

bob:/builder /bob/builder nfs rw,bg,hard,rsize=8192,wsize=8192 0 0

then everything "just works", but as I go back in time, when I get
to a machine running fedora 11, I get the highly informative
"incorrect mount option" message. With lots of experimentation,
I eventually found that I could change the type from "nfs" to "nfs4"
and the mount would work.

Is there a guide somewhere for which idioms to use for which version
of linux I want to mount in? Is there any way to automate adding
these fstab entries correctly? I thought the nfs server was supposed
to automatically fall back to old protocols if the client didn't know
something, am I missing some sort of configuration on the centos 7 host?