Re: NFS tries following remote symlinks as if local!
On Tue, 2001-11-20 at 16:04, Rich Puhek wrote:
> Don't be too hard on people who seem to underestimate your knowledge of
> a subject... often it's easiest to start from a certain point in an
> explanation to be certain everyone's on the same page... That's also
> very handy for others following along in the list or in the archives of
> a list.
No problem. I did sound incoherent at the beginning! Got to proof read
more as well.
I wonder if another network filesystem would be the answer, e.g. Coda:
http://www.coda.cs.cmu.edu/
I was going to attempt to implement it until I discovered that it hasn't
been packaged for unstable yet.
It looks like unofficial packages are available:
http://rpmfind.net/linux/coda/debian/binary-i386/
Has anyone used them? I don't want to risk destroying my data right
before I commence a big backup ;-) Having looked though the Coda user
manual it appears configuration is rather complex (plus "Coda is very
differently organized from NFS and Windows/Samba shares").
I've now just found that current NFS implementations are specially
designed to NOT cross mount points:
http://playground.sun.com/pub/nfsv4/nfsv4-wg-archive/1996/0004.html
The NFS protocol is specified to not let a NFS client cross
mount points. This is so the NFS client does not get confused
about the identity of files in the even[t] two files on two
different server file systems share the same file id (inode #).
Which is a PITA, meaning I will will have to reproduce all the mount
points on the client computer if I implement NFS. It appears my
perspective fits into this category:
This semantic is not desired by some clients, such as PC desktops.
The proposal, as described previously, is to make
mount point crossing optional.
Regards,
Adam
Reply to: