On Thu, 27 Jul 2000, Paul Blackburn wrote:
> Ken Hornstein wrote:
>
> > >I do not understand how you can characterize this
> > >as "another abuse of symblinks"?
> > >
> > >How, exactly, does one use symbolic links in a non-abusive way?
> >
> > I always thought that symbolic links were kinda like salt ... good
> > when used appropriately, but using them too much will kill you in
> > the long run.
> >
> > It always seemed to me that the right solution was to change the
> > cache manager so that a stat() of a mountpoint in /afs didn't cause
> > you to contact that cell. There's was even some code that was #ifdef'd
> > out in the cache manager that claimed to do that (but it was for
> > NeXTStep, so I suspect that it hasn't been maintained :-) ).
> >
> > --Ken
>
> Yes, except that it might take a while to get "out there" and change every
> AFS client.
>
> A least with "root.othercells", a cell administrator can provide a
> solution
> for clients of that cell without having to modify those AFS clients.
If you use something like Synctree, you can make your change in one
location, and when you come back in the next morning, the change will
have been propagated to all of your machines, in an orderly, secure
fashion.
ftp://ftp.engin.umich.edu/pub/synctree
http://www-personal.engin.umich.edu/~rmach/synctree/
--jlockard - The beatings will continue until morale improves.
------------------------------------------------------------------
John M. Lockard - SysAdmin III | Merit Network, Inc., Suite 2000
[EMAIL PROTECTED] | 4251 Plymouth Road, Bldg. 1
[EMAIL PROTECTED] | Ann Arbor, MI 48105-2785
http://www.merit.edu/~jlockard | 734-647-9811 | 734-647-3185 FAX
------------------------------------------------------------------
