Re: nfs server performance (netbsd-5)

[Andrew Doran <ad%NetBSD.org@localhost>]

On Mon, Apr 05, 2010 at 10:47:54PM +0200, Michael van Elst wrote:
> On Mon, Apr 05, 2010 at 09:15:44PM +0300, Jukka Marin wrote:
> 
> > Okay, thanks.  It seems the NFS client is retrying writes, sometimes
> > a fragment or two is missing (so a retry is needed), but sometimes
> > I can't seen anything missing and the client still retransmits the Write
> > command: [RPC retransmission of #1657]V3 WRITE Call
> 
> Then the client missed the answer.
> 
> > The client is a Dreambox 7025 running linux, mount options look like this:
> > 
> > foo:/media on /media/hdd type nfs 
> > (rw,v3,rsize=32768,wsize=32768,hard,udp,lock,addr=foo)
> 
> UDP mount and rsize/wsize=32768 is ambitious...
>     
> You may try to cut this down to 8192 (common) or even 1024 (avoids
> fragments) or better switch to TCP if your client support this.

In case it's not obvious, TCP is likely to give much better performance
in some configurations.