Re: Win32 Cache Question

The Windows and Unix implementations are independent.
While they share some core libraries the actual client implementations 
are very different.
The cache on Unix is not the same as the cache on Windows.  When 
persistence was
added to the Unix implementation there was no one working on the Windows 
implementation.

The Windows cache uses a memory mapped file to store the file data 
segments.  It keeps the vnode
data in memory.  In order for a persistent cache to be implemented three 
things must be done:

   1. synchronize the startup of all threads in the afs client service
   2. store the vnode data in a memory mapped file
   3. synchronize the shutdown of all threads in the afs client service

The startup/shutdown synchronization is required to preserve the 
integrity of the cache data.

Jeffrey Altman


Tommie Gannert wrote:

>Just read the status list. One item was about making a persistent cache.
>
>So, how does the win32-cache work now, and why is the cache
>implementation different on Windows vs. Solaris et al.?
>
>/Tommie Gannert
>_______________________________________________
>OpenAFS-Win32-devel mailing list
>[email protected]
>http://lists.openafs.org/mailman/listinfo/openafs-win32-devel
>  
>