Re: [fuse-devel] NFS - corrupt files?

[Valient G. <vg...@po...>]

So can I correctly characterize the data flow in the test setup as:
	local server FS -> encfs -> FUSE -> NFS server=20
		-> NFS client -> cp -> local client FS

Or is encfs being run on top of another type of filesystem (like an NFS=20
mount)?

On Sunday 12 June 2005 22:36, Matthias Rieber wrote:

> > the resulting encfs filesystem (plaintext data)?   Or are you having nfs
> > export enciphered data and then running encfs on the remote system to
> > decipher the data?
>
> no. Maybe I should try that, but I'm a bit afraid of file locking problems
> and damaging the encrypted files.

=46rom a security standpoint, it is usually better to have the client handl=
e the=20
encryption, rather then the server..

If there is a problem with the interaction between NFS and FUSE, then this =
may=20
avoid that problem.  If you export the encrypted data through NFS, then=20
you're just exporting a normal filesystem, so no NFS <-> FUSE interaction o=
n=20
the server. =20

The data flow would then look like:
	local server FS -> NFS server=20
		-> NFS client -> encfs -> FUSE -> cp -> local clientFS
=09
Of course this doesn't help finding the source of the problem, just an idea=
=20
for a workaround.

regards,
Valient