Infinite Recursion

[Chris Cowan]

Magnus Ahltorp writes:
 > > The "OldFiles" mountpoint is itself duplicated in the backup volume, thus 
 > > creating a self-referential mount.  The Transarc client disallows this, but 
 > > Arla sees nothing wrong.  I suspect Arla should detect that a mountpoint 
 > > references the volume containing the mopuntpoint and disallow it, just as 
 > > the Transarc client does.
 > 
 > Self-referential mount points are not disallowed in the Transarc
 > client. There is a special case for backup volumes:
 > 
 > sundance:~$ fs lsm selfref
 > 'selfref' is a mount point for volume '#home.map'
 > sundance:~$ ls -ld selfref/selfref
 > drwxr-xr-x 49 root        22528 Nov  3 03:32 selfref/selfref
 > sundance:~$ fs lsm .Oldfiles
 > '.Oldfiles' is a mount point for volume '#home.map.backup'
 > sundance:~$ ls -ld .Oldfiles/.Oldfiles
 > .Oldfiles/.Oldfiles not found
 > sundance:~$ 
 > 
 > This is probably due to the fact that people are not used to non-tree
 > file systems, and Transarc is apparently trying to locally "treeify"
 > the user's home directory in case of a Oldfiles directory.
 > 
 > I don't think it's a good idea to implement this Transarc compat, but
 > we'll see, it may get implemented anyway.
 > 
 > Are there strong political feelings in this issue?
 > 
 > /Magnus
 > map at stacken.kth.se
 > 

I've been lurking mostly, but there's one thing I'd like to see.
The UNIX/POSIX find has flags to stop at a device boundary.  For AFS,
special hacks were often made to GNU find to simulate this feature
with volumes.  (Yes, I know that you can also get this by using -exec, too!)

Actually, I'm wondering what DFS did in this case?  (Don't have access to a
system anymore).  But, I think it behaved correctly.   

I think it would be nice for the ARLA client to do this, rather than
have to go fix the utilities.  

Regards,
Chris Cowan

Hopefully participating here much more since my present job isn't
related to filesystems in any immediate way.