On Sat, 31 Mar 2007 16:49:02 +0200
Johnny Billquist <bqt@softjar.se> wrote:

> If the problem is reception, then this is usually the trick you need.
> If data transmitted to the VAX is in more segments than the receive 
> buffers, then it will basically never succeed. One or another of the 
> segments will always be lost.
> 
> However, I think that George thought his problem was transmissions, in
>  which case this won't help. But he might be wrong, so it's worth a
>  try...

I think it is a transmit problem so I've tried reducing the NFS write
size to 2K and haven't had any lockups yet. However, I'm now getting
file corruption instead :( Building pkgtools/digest fails with syntax
errors during the configure stage, looking at the configure file I see
that it has chunks missing and is much shorter than it should be. The
same source, on the same server, builds with no problems on my SGI
Challenge S so I'm fairly confident that it's not a bug on the server.
I'll do some more investigating...

Thanks,
George

> 	Johnny
> 
> Ken Seefried wrote:
> > George Harvey writes:
> > 
> >> I'd
> >> particularly like to get NFS working, are there any
> >patches/workrounds > that might help?
> > 
> > 
> > Not that I've got insight this particular problem, and this won't
> > get to  your root cause, but my personal first dirty trick to NFS
> > problems is to  reduce the NFS read and write block size to 4K, 2K
> > or 1K.  Might let you  limp along until you figure out what's really
> > wrong. Ken
>