I am trying to network 2 Redhat 9 boxes connected to the Internet through a Netcomm NB1300plus4 ADSL Modem/Router/Switch.

This thing has DHCP and has given the IP addresses 192.168.1.13 to the "server" and
192.168.1.6 to the "client"

Kernel is 2.4.20-6 It seems I have to use NFS
NFS not running at boot-up (built as module)
I followed the steps described in 'http://nfs.sourceforge.net/nfs-howto/' but when I try to mount a directory on the client I get the following message:
"mount: 192.168.1.13:/home failed, reason given by server: Permission denied"

root name, user name and passwords are exactly the same on the 2 boxes

Server side
=========
this is the "ls -la" entry from the server from /
drwxr-xr-x 4 rene rene 4096 Sep 13 22:22 home

this is the /etc/exports file on server:
/var 192.168.1.6(ro)
/home 192.168.1.6(ro)

this is the /etc/hosts.deny on server
portmap:ALL
lockd: ALL
mountd: ALL
rquotad: ALL

this is the /etc/hosts.allow on server
portmap: 192.168.1.6
lockd: 192.168.1.6
rquotad: 192.168.1.6
mountd: 192.168.1.6
statd: 192.168.1.6
ALL: localhost
rpc.mountd: ALL

This is the script I used to start it up after booting up
#!/bin/bash
/sbin/insmod nfs
/usr/sbin/rpc.mountd
/usr/sbin/rpc.nfsd
/sbin/rpc.statd
/sbin/rpc.lockd
/usr/sbin/rpc.rquotad

and the result of rpcinfo -p from the server after running previous script
program vers proto port
100000 2 tcp 111 portmapper
100000 2 udp 111 portmapper
391002 2 tcp 32769 sgi_fam
100003 2 udp 2049 nfs
100003 3 udp 2049 nfs
100021 1 udp 32770 nlockmgr
100021 3 udp 32770 nlockmgr
100021 4 udp 32770 nlockmgr
100005 1 udp 32772 mountd
100005 1 tcp 33101 mountd
100005 2 udp 32772 mountd
100005 2 tcp 33101 mountd
100005 3 udp 32772 mountd
100005 3 tcp 33101 mountd
100011 1 udp 1017 rquotad
100011 2 udp 1017 rquotad
100011 1 tcp 1020 rquotad
100011 2 tcp 1020 rquotad
100024 1 udp 32773 status
100024 1 tcp 33102 status

Client side
========
Script used on client after booting up
#!/bin/bash
/sbin/insmod nfs
/sbin/rpc.lockd
/sbin/rpc.statd

the result of rpcinfo -p from the client after running previous script
program vers proto port
100000 2 tcp 111 portmapper
100000 2 udp 111 portmapper
391002 2 tcp 1025 sgi_fam
100024 1 udp 1028 status
100024 1 tcp 3590 status

result of "ls -la" from /mnt from client
total 23
drwxr-xr-x 5 root root 4096 Oct 19 20:15 .
drwxr-xr-x 19 root root 4096 Oct 19 18:49 ..
drwxr-xr-x 2 root root 4096 Oct 19 10:33 cdrom
drwxr-xr-x 5 rene rene 7168 Oct 20 00:19 floppy
drwxr-xr-x 2 rene rene 4096 Oct 19 20:15 home

Command used on client:
[root@hp rene]# mount 192.168.1.13:/home /mnt/home

I have checked and re-checked many times and I can't work out what is wrong.
Any help or hint most appreciated. If necessary I can disconnect one box from the net and
connect the 2 boxes with a cross-over cable to achieve networking (if this makes a solution easier)

On the server side, in your NFS share

-tyring chaning its permission to 777,
-change its owner to root for a while....

On the client side mount it by root user


-jack

Thanks to 'emailssent'.

Tried your suggestion, the result is the same.
(I've edited my original post a fraction for clarity of what's been done)

show here ur output of ls -l of ur NFS share ...... of the server


-jack

This is the relevant resulting line from 'ls -la' issued from / (root - lines for other directories ignored)
drwxr-xr-x 4 rene rene 4096 Sep 13 22:22 home

I've found that running the following lines on the client fixes the problem
"/etc/rc.d/init.d/nfs stop"
"/etc/rc.d/init.d/nfs start"

I am now able to see the shared files from (server) /home on the client immediately (/mnt/home) after rebooting and without having to run any script as explained earlier but a new addition (/var) to this system is not giving the expected result, so I'll investigate a bit more and update this post accordingly.

Thanks to emailssent.
I've found the server is not functional under Redhat 9 unless I run the following line which I have added to the start of the script described above.
/usr/sbin/exportfs -ra

There are probably more elegant ways of doing it but I find it premature to spend time on it when one major problem still persists. The problem is, regardless of what I do, I can only manage to see the first level of directories from the client. For example when I export from the server and mount on the client "/var", I can only see the contents of "/var" on the client, not the content of the sub-directories in "/var" (like "/var/www"). If I try to export from the server and mount on the client "/var/www" then I get a message on the client saying it cannot mount /var

Since Redhat 9 is a dead end, I decided to install Fedora core2, I'll wait till I see how NFS go with it and Ill then update this post.

@rblampain


Try this,

You have make another share on the NFS server
for /var/www/

(becz. as far i remember they appear on the the same disk, if they appear on another disk then u will be able to mount it also without creating another share)



-jack

To Jack:
Yes that fixed it. I can now see all the subdirectories, sub-subdirectories and files in /var/www ( and others). Great stuff.
So, for posterity, I've got now in my "/etc/exports" (on server)
/var 192.168.1.6(ro,no_root_squash)
/var/www 192.168.1.6(ro,no_root_squash)
and in "/etc/fstab" on client
192.168.1.13:/var /mnt/var nfs ro 0 0
192.168.1.13:/var/www /mnt/var/www nfs ro 0 0
also created directories "/var" and "/var/www" in "/mnt"

Thank you very much for your persistence.

Let's backup to the first set of exports /var and /home

As you export the /var directory you can mount that, but you don't export the /var/www as a resource so therefore you can't mount it.
You can still reach the /var/www directory using the /mnt/var/www path (provided that you mount /var to /mnt/var on the client).
Compare to mounting a hard disk partition. You mount the entire partition with all of it's subdirectories, but you can't mount a directory of a partition.

@rblampain

Anothere solutions now I remembered as I am reading ugge post, that if u don't want to make seperated export for ur www then u can try to mount it directly from the NFS client.

That's without exporting www u can just mount it from NFS client
#mount -a -t nfs
or
#mount /var/www /var/www
(mount <sorce directry on NFS server> <destination directory on NFS client>

This all above is a NFS rule provided by nfs protocol.



-jack

You can't mount what you don't explicitly export.
If you don't export the /var/www then you can't mount that directly, because that resource doesn't exist as far as exports go.
You will have to mount the /var directory and then cd to the right subdirectory on the client.

Thanks to you people for all your suggestions. I definitely have to streamline the set up but I will do it slowly, one thing at a time after I install the new distro as explained before.

I can now see all the shares from the client, form all the directories listed although I am getting the following message after runnung the script "nfs.sh" on the server:
192.168.1.6:/home Invalid argument
which doesn't seem to affect anything and which I can't explain.

Also I have to run the scripts a couple of times after re-booting.
Here are the latest files I used.

Server side
=========
# /etc/exports
/home 192.168.1.6(ro,sync)
/home/rene 192.168.1.6(ro,sync)
/var 192.168.1.6(ro,sync)
/var/dir1 192.168.1.6(ro,sync)
/var/dir2 192.168.1.6(ro,sync)
/var/dir3 192.168.1.6(ro,sync)
/var/dir4 192.168.1.6(ro,sync)
/var/dir5 192.168.1.6(ro,sync)

# hosts.deny
portmap:ALL
lockd:ALL
mountd:All
rquotad:ALL

# hosts.allow
portmap: 192.168.1.6
lockd: 192.168.1.6
rquotad: 192.168.1.6
mountd: 192.168.1.6
statd: 192.168.1.6
ALL: localhost
rpc.mount:: ALL

# "ls -la" of script on server (can only run as root)
-rwx------ 1 root root 169 Oct 24 12:23 nfs.sh
# script nfs.sh
#!/bin/bash
/usr/sbin/exportfs -ra
/sbin/insmod nfs
/usr/sbin/rpc.mountd
/usr/sbin/rpc.nfsd
/sbin/rpc.statd
/sbin/rpc.lockd
/usr/sbin/rpc.rquotad

Client side
========
Appropriate mount points in /mnt to match the exports file on server
Appropriate mount entries in /etc/fstab

# script nfs.sh (can only run as root as above)
#!/bin/bash
/sbin/insmod nfs
/sbin/rpc.lockd
/sbin/rpc.statd
(this script appears unnecessary)

# script stst.sh
/etc/rc.d/init.d/nfs stop
/etc/rc.d/init.d/nfs start
(this script seems necessary)