The rx listener thread in a fileserver can float from thread to thread, but even if it couldn't you don't necessarily get off: work for any given client is not necessarily handled all within a single thread, and thus not necessarily on a single core. What this means is you get to incur context switches when one packet is handled on the first core and the next packet from the same client is on another core. So there's no one "right" answer. And as long as but a single port (7000) handles all traffic there is not a convenient way to bind streamless (UDP) traffic from a single client to a single core...
On Mon, Jul 22, 2013 at 1:21 PM, Rich Sudlow <[email protected]> wrote: > Are there any recommendations on why/why not to use hyperthreading on > Intel machines used as fileservers with Red Hat 6.4 OS? > > Thanks, > > Rich > > -- > Rich Sudlow > University of Notre Dame > Center for Research Computing - Union Station > 506 W. South St > South Bend, In 46601 > > (574) 631-7258 (office) > (574) 807-1046 (cell) > ______________________________**_________________ > OpenAFS-info mailing list > [email protected] > https://lists.openafs.org/**mailman/listinfo/openafs-info<https://lists.openafs.org/mailman/listinfo/openafs-info> > > -- Derrick
