re-export nfs possible?

re-export nfs possible?

[nbecker]

Is it possible to mount a fs via nfs, and then reexport it via nfs?

Re: re-export nfs possible?

[Neil Brown]

On  August 1, nbecker@fred.net wrote:
> Is it possible to mount a fs via nfs, and then reexport it via nfs?

No.

The protocol doesn't really support it, and the (Linux-kernel)
implementation definately doesn't support it.

I think the user-space nfs server can do it.  It has other problems,
but it might work for you.

NeilBrown

Re: re-export nfs possible?

[Luigi Genoni]

On Thu, 2 Aug 2001, Neil Brown wrote:

> On  August 1, nbecker@fred.net wrote:
> > Is it possible to mount a fs via nfs, and then reexport it via nfs?
>
> No.
>
> The protocol doesn't really support it, and the (Linux-kernel)
> implementation definately doesn't support it.
definitelly the kernel implementation cannot manage the loops a re-export
would imply.
>
> I think the user-space nfs server can do it.  It has other problems,
> but it might work for you.
yes, you have to run both rpc.mountd and rpc.nfsd with the -r option.I
used this one year ago, and it was really stable with 10 nfs clients
(physicaly the partition was on an AIX server, mounted on a linux server
and then re-exported on many linux clients), but the physical FS status
could not be really coherent in front of what the nfs re-export server
thinks it should be.

Luigi

Re: re-export nfs possible?

[Albert D. Cahalan]

nbecker@fred.net writes:

> Is it possible to mount a fs via nfs, and then reexport it via nfs?

Currently, no.

With some restrictions, it could be done. For any given IP address,
you could export or re-export filesystems from _one_ source server.
Anything more is like trying to cram N+1 bits into an N-bit sack,
and no you can't assume the bits are compressible.