Re: nfs server performance (netbsd-5)

[Jukka Marin <jmarin%embedtronics.fi@localhost>]

On Mon, Apr 05, 2010 at 12:59:36PM +0300, Jukka Marin wrote:
> On Mon, Apr 05, 2010 at 12:38:20PM +0300, Jukka Marin wrote:
> > I "upgraded" my 4.0 server to netbsd-5 by extracting the tarballs over
> > the old binaries, so old libraries and stuff were not deleted.  Maybe this
> > is causing me problems.. is it possible that the new nfsd is using libs
> > from NetBSD 4.0 and this is causing problems?
> 
> What are the RPC errors, reported by nfsstats?  The number is increasing.
> 
> RPC errors          faults
>      67923               0

I have updated all NetBSD systems to netbsd-5 built from yesterday's sources
(and cleared the old libraries).  I also changed the linux clients to use
TCP mounts with 8 k r/w sizes.

Now the NFS server performs better - CPU load is lower and the system no
longer coughs when nfsd is running.  However, I'm still getting lots of
RPC errors - something like 100 or 200 per second during a build on a NetBSD
client machine.

I'll try tcpdump and wireshark again (the problem might be easier to spot
in the logs now that it occurs this often).

The NetBSD client uses these options for the NFS mount (I have been adding
more options year by year, as you can see ;-)

  rw,-X,-i,-b,-s,-T,-x16,-D100000

I also switched from softdep to log for most partitions on most virtual
machines.  (dom0 isn't using either softdep or log.)

  -jm