Fwd: [Gluster-devel] proposals to afr

Kevan Benson kbenson at a-1networks.com
Thu Oct 25 16:11:15 UTC 2007


Alexey Filin wrote:
> order of operations can be guaranteed on ideal hardware and software only,
> for example:
> 
> * operations over a file and its extended attributes can be considered by
> backend FS independent and reordered (I don't know is it true or not on all
> FS's),
> * disk driver can reorder operations over different files,
> * hard disk controller caches and reorders operations to optimize disk heads
> movement

As I said, I was referring to atomic in the context of GlusterFS.  Using 
locks (thread locks in his case), they can easily prevent any other 
GlusterFS threads from performing an operation on the file).  In that 
respect, it's atomic for that file with respect to GlusterFS.  Other 
processes on the server could of course perform operations at this time, 
which makes the operation not TRULY atomic, just atomic for glusterfs on 
that file, on that system. Perhaps "atomic" was a bad choice of a word 
to convey this.

-- 

-Kevan Benson
-A-1 Networks





More information about the Gluster-devel mailing list