Subject: Re: Miserable NFS performance between NetBSD and MacOS X
To: Matthias Scheler <email@example.com>
From: Greg Oster <firstname.lastname@example.org>
Date: 12/14/2005 08:45:07
Matthias Scheler writes:
> On Tue, Dec 13, 2005 at 11:15:25PM +0000, email@example.com wrote:
> > Try something like this in your fstab when mounting nfs directories:
> > nfs_server:/remote_dir /local_mount_point nfs tcp,intr,-w=32768,-r32768,rw
> 0 0
> That doesn't help unfortunately. From "tcpdump" output it looked like
> MacOS X used 16KB by default anyway which isn't to bad.
If you do a 'tcpdump' to watch the NFS transfer, what sorts of "spacing"
do you see between packets? Who is being "slow" to respond? (e.g.
are there retransmits or delays in sending ACKs or anything)
You can likely use a ttcp transfer to guestimate what the traffic
flow should look like...
Also: is file transfer slow for both reads and writes, or just writes
to the server?