NFS problem in 2.0.32
16 Dec 1997 11:13:00 +0100


I'm turning, at last, to this list, since I am unable to find any
information that seems to cover my case. It is, of course, NFS again.

I have a Linux server that talks NFS to a Solaris server. Mounting up
disks from the Solaris box works fine. Mounting up the other way works
fine - as far as I can tell - for reading, but bogs down horriby when
writing. I get copious "not responding"-messages on the Sun, and things
go terribly slow, although they eventually work.

Snoop/tcpdump seems to indicate a host of packets from the Sun, with
only a scarce reply of WRITE2 OK from the Linux - is this right? (I'm
obviously no expert on this)

The network isn't very compex, just a TP hub between the machines, and
there doesn't seem to be a lot of errors on either machine's interface.
The network config is a 3C905 100bTX (rev 0) according to proc.

Any suggestions? Is the knfsd in 2.1.* ready for serious use yet? I'm
getting somewhat desperate, and as I introduced Linux as an alternative
to another OS from a large vendor (you know the one), there's much at
stake - including my reputation. :-)

If there's anything sensible I can do to debug, please point in the
right direction, even if it's just an RTFM.


If I haven't seen further, it is by standing in the footprints of giants