Coda File System

Re: Coda status quo and future

From: Jan Harkes <jaharkes_at_cs.cmu.edu>
Date: Wed, 27 Oct 2010 14:25:21 -0400
On Wed, Oct 27, 2010 at 02:05:48PM -0400, Phil Nelson wrote:
> Jan has suggested changing venus to check and see if a file that
> was closed had changed by storing a hash of each file and recomputing
> that hash at close time.   If the hashes match, then the file is
> unchanged and not sent back to the server.   This would allow unchanged
> files to not be send on any use of venus, not just on Windows.
> 
> Neither change has been made to venus and one of the two changes is
> required to be able to start the change to the Coda Kernel Module for
> Windows.

It isn't as simple as just adding hashes, there is an interaction with
potentially pending write operations in the modification log that are
removed when we reintegrate while the file is open for writing. So there
has to be some care taken to weave this correctly around CML store
optimization logic to avoid losing data.

But having a valid hash for any not-currently-owrite file would have
several other benefits, such as improving reliability and performance of
lookaside caching and revalidation of cached data after f.i. a system
crash. A disadvantage is the noticable performance hit when closing a
file that was open for writing.

Jan
Received on 2010-10-27 14:25:28