Re: Debian replicator boot disk problem

On Sat, 31 Dec 2005 09:21:06 +0100, Christian Maier <tomtailor@xxxxxxxxxxx> wrote:

OK:
What I am  doing. All steps described at
http://replicator.sourceforge.net/documentation/Replicator-UserGuide/node5.html

I can connect from an other client in the LAN 192.16.0.12 pretty good.
Premissions in exports ist 192.168.0.* (rw, no_root_squash),  so is it
realy a serverside problem? Or is ist caus client cant detect his hd,
filesystem, controller, what so ever?


If i understand it right, you posted a syslog entry

  Dec 30 18:59:46 localhost rpc.mountd: refused mount request from 192.168.0.13
                                        for /export/miniroot (/): no export entry

This only makes sense to me if the localhost is the *server* and
192.168.0.13 is the *client*.

How could the server write this in its syslog, unless the client really
sent a request to the server?  Why would the *server* refuse the request
just because the *client* does not detect its hd, brush it teeth, or sing
negro spirituals?

There is still a potential for myself being the one who misunderstands,
(shame that I'm so sarcastic then) but you have failed give us a clear map
of the landscape of events.  Now I have one reference point: There is a
computer involved that has IP address 192.168.0.13, and that computer
has sent an NFS mount request to another computer - about which I don't
know the least, except that it has a syslog, and runs an rpc.mount
daemon. It would be nice if you had described things in such a manner that
we could refer to the computers involved using labels that you had
introduced.  Why don't you say

 "I have three computers A,B, and C, with IP addresses
  192.168.0.{10,12,13} respectively, and I am using
  A as an NFS server, with the following exports file:

    /exports 192.168.0.*(rw, no_root_squash)

  The directory /exports contains two subdirectories:

    drwx------  3 christian christian  4096 Nov 21 17:38 mypictures
    drwx------  3 root      root       4096 Dec 22 00:55 miniroot

  which are both on the same partition as /exports. The underlying
  file system is ext3.

  B has the following entry in /etc/fstab:

    A:/exports/mypictures /home/ego/photos nfs user,bg,rw,soft 0 0

  and that seems to work just fine, I have browsed the directories
  and even viewed the pictures on the screen.

  C is a diskless station, and mounts its root file system from A.
  This is how I boot C...
  This is the fstab entry..."

It would be great if you could
say: yes 100% serverside. I don't want to compile kernels anymore (it's
always a problem with the size and the supported stuff)


Agreed. It would be great. Unfortunately I can only say: I'm 100% sure that
I, in your place, would have wanted to investigate why the server refuses
the mount request.  I don't know what would be the outcome of that
investigation.  Should I come over to your place and do it?

Problems with the size and the supported stuff? You mean... Oh you need
to keep the kernel image small, and have trouble determining what to include
and what not.  (Why do you need to keep it so small?  No, let's come back
to the kernel later, if it turns out to be required.)

The Clienet is about a 1,44MB Floppy with a realy small kernel on it
(ONLY supports ip autoconfig., DHCP, NFS, ext3 and ISA support).
Compiling your own kernel isn't basic stuff for me!


OK.  We can come back to the kernel as soon as we have evidence that
it needs revision.  (But why do you need ext3 if you have no harddisk...
Oh, Enrique! Stop it. Leave the kernel alone for now. :) )

The Floppy boots and the mini kernel want to connect via NFS to my
model -> seems the request is recognised and logged at serverside. And
then the client crashes with kernel panic.


Oh, but this is new information.  This may warrant an examination of the
kernel configuration, but it may also just mean that the nfs mount is
the root file system, and the kernel panics if it cannot mount the a
root file system.  It needs a root file system, because it wants to
run a program called "init" in the root file system.  If it can't run
"init" it's not much else it can do!

Of course one could imagine a broken kernel on the client side sending
improperly formatted mount requests, just because it is so broken.
But notice that this is a kind of imprecise thinking, like "Oh, so
the car has run out of gas?  I guess the plugs are bad too. I have heard
that cars that don't run often have bad plugs."

I don't find it likely that you would modify the c code of the kernel.
If you have just made your selections in the kernel configurator, you
should either have a good NFS client support or no NFS client support
at all, and the crash should be due to something external to the NFS
support itself.

I would definitely look into the actual configuration data before taking
on a rather difficult kernel debugging task.

Don't know why. Sais
connection refused from Server but I can mount with mount -t nfs ....
fromanother linux in the LAN fine.


You don't tell what directory the other linux mounts. Is it the same
/exports/miniroot?

And watch your language: "Connection refused" has a rather specific meaning.
You are *not* seeing refused connections!

-Enrique
.

Relevant Pages

  • NFS problems with through 2.5.x to 2.6.0-test9
    ... When the server is running the ... kernel, as a client the 2.6 series seem to work perfectly, excluding ... Interesting problem arose when I attempted switch the server's kernel to ... with and without nfsv4 support compiled in (was considering testing it at ...
    (Linux-Kernel)
  • Re: Problems mounting nfs from freebsd to Mac.
    ... Problems mounting nfs from freebsd to Mac. ... I've got an nfs server that's refusing to mount one client - via one ... That elimintes NFS on the client, and -most- of the NFS config on the ...
    (freebsd-questions)
  • Re: After a full day of computers...
    ... > computers in question run SuSE 9.1 Pro and nothing but. ... > through the CUPS server). ... You need to create a mountpoint on the client machine. ... Once the shares have been created, they need to be mapped onto the mount ...
    (alt.os.linux.suse)
  • "CIFS VFS: server not responding" with some client/server combinations
    ... As it might be kernel related and others ... A Samba server running Debian Etch with latest updates. ... I don't know what's different with Etch on the client, ...
    (Linux-Kernel)
  • Re: recent nfs change causes autofs regression
    ... If the user asks for a new mount that is read-write, ... get it - ie we should not re-use the old client handles, ... everything needs to have the same flags), THOSE PEOPLE, who want the NEW ... or deleted on the server are now extended to also include the ...
    (Linux-Kernel)