Suse 11.2 NFS fails

English Network/Internet
1

I set up my 11.2 NFS server the same way I set it up in 10.3. The client side is unchanged. When I boot 10.3 it works. When I boot 11.2 it doesn’t. The only error message I could find is…

Nov 16 20:59:45 linux-3hev exportfs[2678]: /etc/exports [1]: Neither ‘subtree_check’ or ‘no_subtree_check’ specified for export “192.168.0.201:/home/ion/SHARE”.#012 Assuming default behaviour (‘no_subtree_check’).#012 NOTE: this default has changed since nfs-utils version 1.0.x

…and it seems somewhat benign.

Any ideas would be appreciated.

2

Hi,

Interesting, I am checking this out now using a virtual box. I’ll see if I get the same and post back!

cheers
J

3

Ok,

I have tested this from my a brand new 11.2 64bit install

My /etc/fstab entry on the NFS client is:

192.168.1.220:/srv/www/htdocs /mnt/nfswww nfs defaults 0 0

and my /etc/exports on the NFS server is:

/srv/www/htdocs 192.168.1.0/255.255.255.0(rw,root_squash)

All works as expected with no errors.

I do not have a 10.3 setup to check against, but this should also work.

How are your conf files looking ?

hth
J

4

Thanks for the quick response.

My client fstab reads:-

192.168.0.201:/home/ion/SHARE /home/ingrid/ION-IN nfs rw 0 0

My server /etc/exports reads:-

/home/ion/SHARE 192.168.0.202(rw)

The only difference I see is that I assigned a single client IP rather than a range. By the way I did try it using a mask as well (192.168.0.0/255.255.255.0).

Am I missing something obvious?

5

Here’s some error output I found…

Nov 19 10:07:17.240983000> service nfs start

Nov 19 10:07:17.249384000> Shutting down NFS client services:

Nov 19 10:07:17.349825000> checkproc: /usr/sbin/rpc.statd 6327
umount: /var/lib/nfs/rpc_pipefs: device is busy.failed

Nov 19 10:07:17.364723000> service nfs done

Nov 19 10:07:17.365480000> service nfsserver startShutting down kernel based NFS server: nfsd

Nov 19 10:07:17.758602000> service nfsserver done

6

Let me guess: you copied over /etc/exports from 10.3 ?

I would reconfigure NFS from scratch, to avoid conflicts with old settings. This is a line from my /etc/exports on my server, which has 11.2, 11.1, a 10.3 and a 9.3 client(s):

/home *(rw,root_squash,sync,no_subtree_check)

The /etc/fstab on the 11.2 client to mount /home is like this:

192.168.1.100:/home /home nfs defaults 0 0

This combo works flawlessly. Change lines in your config and you should be OK.

BTW I assume 11.2 is a clean install…

PS. Perform a reboot, to make sure NFS is starting clean.

7

Thanks for the suggestion.

No, I didn’t.

I have reconfigured it from scratch several times, but just in case I did it again.

I changed my /etc/exports to reflect your settings, just in case. It now reads “/home/ion/SHARE *(ro,root_squash,sync,no_subtree_check)”.

Made no difference at all.

Yes it was.

Rebooted. NFS still not running. Same error messages. A scan from the client reveals no NFS server on the LAN.

8

Pardon me. Not the same error messages.

Nov 19 15:37:03.545175000> service nfsserver start
Nov 19 15:37:03.717894000> service network-remotefs doneShutting down kernel based NFS server: nfsd
Nov 19 15:37:03.894098000> service nfsserver done
Nov 19 15:37:03.895575000> service nfs startShutting down NFS client services:
Nov 19 15:37:04.38070000> service nfs done

9

After fruitless repeated attempts to connect via NFS, I am in the process of re-installing 11.2 from DVD, hoping that my problem was caused by something I installed. This time I will not install anything until I have resolved my NFS problem.

Would someone using 11.2 who has successfully connected via NFS to a client be kind enough to guide me through the steps.

Thank you.

10

Hi,

I have just installed and setup a simple nfs-server with a basic export on 11.2 with no problems.

Steps taken:

Installed nfs-kernel-server package via yast

Edit of /etc/exports to read

/mnt/bob (rw)

So my server directory /mnt/bob will be available to all clients. Not recommended but a good test to check things are functioning.

Opened nfs server port in firewall on server

start nfsserver using rcnfsserver start or /etc/init.d/nfsserver start

check output, a warning was thrown similar to the following:

Starting kernel based NFS server: idmapdexportfs: No host name given with /mnt/bob (rw), suggest (rw) to avoid warning
exportfs: /etc/exports [1]: Neither ‘subtree_check’ or ‘no_subtree_check’ specified for export ":/mnt/bob".
Assuming default behaviour (‘no_subtree_check’).
NOTE: this default has changed since nfs-utils version 1.0.x

Illustrating that default behavior for my export is assumed, which for the purpose of test is fine.
The nfsserver daemon has started correctly despite the warning.

A quick check of /var/log/messages confirms that the server is waiting for requests.

I then check from my client that I can see the export

showmount -e x.x.x.x

this returned the correct export from the server.

using yast and the nfsserver config applet, I can check the config, stop and start the daemon, open the firewall port etc. I prefer the cli but yast can help if you have problems with this.

My nfs client is 11.1 connects fine.

mount -t nfs x.x.x.x:/mnt/bob /home/testuser/bob

All is ok.

Can you try this simple export ?

thanks
J

11

I followed your instructions up to this point EXCEPT that I disabled the firewall totally as that is my preferred way of running. I hope that is not what is causing the problem because unlike you I get the following from /var/log/messages…

Nov 23 17:31:59 linux-kc3l exportfs[24494]: /etc/exports [2]: Neither ‘subtree_check’ or ‘no_subtree_check’ specified for export “*:/home/ion/SHARE”.#012 Assuming default behaviour (‘no_subtree_check’).#012 NOTE: this default has changed since nfs-utils version 1.0.x
Nov 23 17:32:00 linux-kc3l nfsd[24498]: nfssvc: Setting version failed: errno 16 (Device or resource busy)
Nov 23 17:32:00 linux-kc3l sm-notify[24499]: Already notifying clients; Exiting!

That “Setting version failed: errno 16 (Device or resource busy)” is pretty cryptic to me.

Here is my /etc/exports…

/home/ion/SHARE *(rw)

I have not installed anything except nfs-kernel-server, and I have not configured anything either, except for the firewall. The only activity has been to create the SHARE directory in /home/ion and a test text file to reside in it.

If I need to re-enable the firewall I could use a link on how to open a nfs server port.

Thanks for all the advise and hand-holding. I am sure there must be a resolution.

12

Still struggling with this.

Could someone explain the following? When I boot 10.3 and “ifconfig” it looks like this…

eth0 Link encap:Ethernet HWaddr 00:11:5B:AB:DC:BB
inet addr:192.168.0.201 Bcast:192.168.0.255 Mask:255.255.255.0

When I boot 11.2 and do the same it looks like this…

eth0 Link encap:Ethernet HWaddr 00:11:5B:AB:DC:BB
inet addr:192.168.0.101 Bcast:192.168.0.255 Mask:255.255.255.0

Why are the inet addresses different? Same box, same card, same cables, same router.

Thanks. I think I’m on to something here.

13

Have you configured the Firewall to open up local LAN?

My only problem mounting an NFS export, was due to a typo on my part.
I copied a working 10.3 configuration to.

The no_subtree_check is interesting, it suggests better performance & security if you partition the system more, generally avoid the “difficult” partial checks.

14

Either you set them different, or it’s DHCP request being made and not trying to recover the old leased address.

So long as you’re mounting from the right IP address (do you need to update hosts(5)) then it ought not matter.

15

My firewall is disabled.

16

Thanks. I’ve been focusing on the wrong part of the problem. I should just accept it’s a different address. I’ll try it out.

17

Hi,

apologies for not posting to this thread for a while, I have been a bit snowed under. I did manage to replicate the same problem you are reporting.

Can you try a mount of your nfs export onto the client using the following command as root:

mount -o vers=3 nfsserver:/path/path /client/mountpoint

basically forcing version 3 of the nfs protocol (Note no need for the -t nfs)

thanks
J

18

Thanks for staying with the problem. My host computer is down for hardware changes, but I will try your recommendation as soon it’s back together and report back on the results.

19

I never got to try the suggested “mount -o vers=3 nfsserver:/path/path /client/mountpoint”, the reason being that I stopped taking my stupidity pills.

First mistake - I assumed that because I had directly set the router to use addresses 192.168.0.201 and 202 for server and client they would stay that way. The fact is that the server is now on 192.168.0.101.

Second mistake - As I was using the client on a daily basis with the server running 10.3 (and it was working) the entry in the client’s “fstab” was not removed.

Third mistake - When I tested 11.2 I was mounting to the same directory on the client side that I used under 10.3, BUT the client had already tried to mount 192.168.0.201 (no nfs) to that directory. Hence the error “device busy”.

I think I have things under control now other than the embarrassment of having asked so many pointless questions. On the positive side, I do understand a little bit more about NFS.

Thanks everyone.