The problem is that every inode in a qtree has a qtree id in it. You'd have
to update the contents of the inode, adjusting the id and updating quotas.
Of course, that would be faster (in most cases) than actually copying the
data for all files. I'm sure there are other issues, such as the fact that
any NFS filehandles referring to the old qtree would be stale, but it's an
idea that keeps coming up so I've submitted an RFE for this: 22411.
Mark Muhlestein -- mmm(a)netapp.com
-----Original Message-----
From: Graham C. Knight [mailto:grahamk@ast.lmco.com]
Sent: Friday, February 25, 2000 7:38 AM
To: toasters(a)mathworks.com
Subject: mv'ing between qtrees
Has anyone figured out a way to trick the
system into allowing a lightning fast "mv" of
directories/files in between qtrees that are
on the same volume?
It seems to me that the filesystem is actually
the volume, and that qtree's are just built on
top of that - so a "mv" *might* be possible with
the right craftiness. Anyone done it? Can anyone
tell me beyond a shadow of a doubt that it's impossible
so i'll quit trying to figure out a way to do it??
The error i'm trying to work around is this one:
doc*> qtree
Volume Tree Style Oplocks
-------- -------- ----- --------
vol0 unix enabled
vol0 doc-1 mixed enabled
vol0 test unix enabled
vol0 test2 unix enabled
doc*> mv /test/testing1 /test2
rename: Cross-device link
doc*>
Thx,
Graham