Replacing NFS.

Robert G. Brown rgb at phy.duke.edu
Wed Apr 11 07:25:12 PDT 2001


On Tue, 10 Apr 2001, Greg Lindahl wrote:

> [speaking of AFS]
>
> > ( non-broken caching is one of the best reasons to use AFS in a
> > cluster of workstations )
>
> One man's non-broken is another man's broken. NFS supports only a
> limited range of policies, and AFS merely supports a different range
> of policies, some of which may work well for you, but can be
> pathologically bad in some circumstances, just like everything else
> has its weak points.
>
> AFS is also a pain in the neck to compile, and the tokens are a pain
> to deal with in a batch setting -- again, only a limited range of
> policies are supported. And we won't talk about scalability.

Speaking of broken, has AFS fixed its writeback behavior so that
fflush() actually works to update the file image across all mounts?  In
the old days (5-6 years ago) one had to close and then reopen a file to
get data written to the file by one node to be visible from another
node.

   rgb

-- 
Robert G. Brown	                       http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at phy.duke.edu







More information about the Beowulf mailing list