[nfsv4] NFSv4 performance issues on a lagged network

[Bryce Harrington]

On Wed, Aug 30, 2006 at 05:01:30PM -0400, Trond Myklebust wrote:
> On Tue, 2006-08-29 at 17:07 -0400, Trond Myklebust wrote:
> > See http://crucible.osdl.org/runs/1621/sysinfo/nfs05.console
> > 
> > It looks like the networking layer is passing back an ENETUNREACH error
> > that we weren't expecting, and consequently weren't handling.
> > 
> > I'll look into drafting a patch.
> 
> OK... Please could you see if the attached patch has an effect on those
> errors.

Yes, it appears that this patch fixes the problem.

First, we reran a case using hard mounts instead of soft:

   $server:$share /mnt/$server $nfs_version hard,intr,noauto,sec=$sec 0 0
   http://crucible.osdl.org/runs/1724/test_output/iozone.sys.log.png

As you can see, it didn't solve the problem.

Next, we added Trond's patch (and kept hard mounts):

   http://crucible.osdl.org/runs/1725/test_output/iozone.sys.log.png

For comparison, here is what we were getting before:

   http://crucible.osdl.org/runs/1582/test_output/iozone.sys.log.png

I'm running a few more cases overnight just to be certain, but so far
the data is pointing to this patch being a good fix to the problem.

Bryce