[SLUG] Copying files from one machine to another

[Paul Teasdale]

On Friday 23 Dec 2005 10:35, john at johnallsopp.co.uk wrote:

Hi John,
>
> So, NFS, according to the RH9 GUI list of services, NFS is ticked so
> it should be up and running, I've added the share to /etc/exports,
> I've exportfs'd it, and then on the FC4 machine, I try to mount it
> (mount 192.168.0.69:/home/j /home/J/j) and it says, after a ponder,
> mount to NFS server (IP) failed, server is down.
>
> So I'm all washed up and need a hint.
>
On the NFS 'serving' PC:

Firstly and foremostly Jamie is right in saying check if there is a firewall 
running. I had a brief encounter with RedHat 8 and by default it did have a 
firewall running so I guess RedHat 9 will have. If a firewall is running 
temporarily stop it (easiest route for testing). Bear in mind however that if 
you plan to leave the NFS services running it is best to 'poke' the necessary 
holes in the firewall to allow NFS connections.

Secondly, bearing in mind that I have not used RH for ages, doesn't RH have 
SELINUX present. If so this may be a factor. You may need to temporarily 
disable this also (or otherwise find out how to configure to allow NFS 
connections).

Thirdly, most distros (again I don't know about RH) tend to compile the NFS 
parts of the kernel as modules so do an 'lsmod' to see if modules are loaded. 
Look for something like (from my Debian box so it may not match RH exactly):

nfsd                  221760  8
exportfs                5888  1 nfsd
lockd                  66152  3 nfs,nfsd
sunrpc                143300  3 nfs,nfsd,lockd

If you don't see these lines then NFS may be compiled as part of the kernel 
(as opposed to kernel modules) so don't get too mislead by the above tip.

You can also try (which I think you said you have done) to do:

$ sudo ps aux | grep nfsd
$ sudo ps aux | grep lockd
$ sudo ps aux | grep portmap

You should get a return otherwise NFS server is not running (e.g.).

nfs     23875  0.0  0.0     0    0 ?        S    15:45   0:00 [nfsd]

Fourthly, check that the expected exports are present by typing 'showmount 
-e'. Here is a list from my Debian box:

/archive        192.168.1.0/24
/archive        192.168.0.0/24
/work           192.168.1.0/24
/work           192.168.0.0/24

Also ensure your /etc/exports file has the correct permissions on the share. 
You didn't post your /etc/exports file so I can't comment. While testing you 
can always put something like:

/home/j	*(ro,no_root_squash)

where the asterix means allow anyone from anywhere. Never leave exports file 
like this however when testing is done.

/home/j	192.168.0.0/24(ro,no_root_squash)

will allow connections only from 192.168.0.?

Fifthly, before trying to mount across a network try a local mount thus 
mounting the filesystem back onto a different directory but still on the same 
PC.

$ sudo mkdir /mnt/tmp
$ sudo mount -t nfs 127.0.0.1:/home/j  /mnt/tmp

If this doesn't work then a mount from another PC never will. Also try the 
same but with the IP address of the current PC.

$ sudo mount -t nfs 192.168.0.69:/home/j  /mnt/tmp

If this works here are some tips for the remote PC:

1) Check it's firewall is not thwarting you.
2) SELINUX? - Definitely in FC4
3) Is portmap running (it must be to allow mount to work - i.e. ps for it).
4) Try: showmount -e <server_ip>. Does it show the mounts you are expecting?

Hope some of these tips help,

Regards,

Paul.

PS: Just thought. Does RH use tcpwrappers by default?

If so (and for testing only) add:

ALL:ALL

in to your /etc/hosts.allow file.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://mailman.lug.org.uk/pipermail/scarborough/attachments/20051223/a35547ae/attachment.bin