[KLUG Members] Odd NFS Mount problem.
Adam Williams
members@kalamazoolinux.org
16 Jan 2002 08:22:22 -0500
>>I can then remount /usr/pcnet and it is ok, and /home
>>becomes a stale handle.... round and round we go...
>>Anyone seen or heard about this?
>Couple of suggestions ...
>1. Try starting more NFS daemons in your server's
>/etc/rc.d/init.d/nfs script.
Well, I oops-ed and restated nfs on the client. Remounted and both
mounted fine! Then I realized what I had done and said to myself "How
the &&@) did that help anything?". But it works now, I unmount and
mounted each NFS volumes a couple of times to be sure. This worked on
two clients, both of which are NFS servers themselves (not the same
volumes of course).
>2. Recommend making sure your "nfs-utils" version is the same on
>all Linux clients.
On the two clients and the one server they are. All the other ones are
diffrent versions and work great. Head scratcher.
>3. Recommend making all your RedHat versions the same, especially
>since RedHat 7.0 runs kernel 2.2.x and RedHat 7.1/2 runs kernel
>2.4.x.
Well, I'd love for that to be true. But it isn't going to happen for
various ugly reasons.
>3a. What happens if you use NFS v2 for just the 7.0 systems? BTW,
>NFS v3 support differs _radically_ between kernel 2.2.x and 2.4.x
>(the former, at least as of 2.2.18, is a limited backport of the
>latter).
Yep, I know about that. All the 7.0 boxes are standard RH kernels, but
they all work fine. And the first thing I tried (see #1) seems to have
fixed the problem.
>4. Any reason you are using 16KB block sizes? I usually stick with
>the 8KB default of NFS v2, or 32KB default of NFS v3. BTW, I think
>Alan Cox _ignored_ the specs, and make 1KB the "default" in 2.2.18+
>(same in 2.4.x???).
AIX. It really wants a 16KB block size. So I just tend to do that.
Actually it is probably more the funky app than AIX itself, but I've got
to play along.