Debian kernel: "Server nfs-home reports our clientid is in use"

Hi folks,

I have to admit I am running Wheezy and a Debian backport kernel (3.16.7-ckt11-1~bpo70+1) on my NFServer and the clients, but maybe somebody on this list can help.

Problem: Since the kernel upgrade it happens that a client looses the NFS connection sometimes. The error message in kern.log on the client says

Jul 30 15:53:57 dpcl082 kernel: [2880660.867248] NFS: Server nfs-home reports our clientid is in use
Jul 30 15:53:57 dpcl082 kernel: [2880660.867254] NFS: state manager: lease expired failed on NFSv4 server nfs-home with
Jul 30 15:54:54 dpcl082 kernel: [2880717.424872] NFS: Server nfs-home reports our clientid is in use
Jul 30 15:54:54 dpcl082 kernel: [2880717.424878] NFS: state manager: lease expired failed on NFSv4 server nfs-home with
Jul 30 15:55:01 dpcl082 kernel: [2880724.526741] NFS: Server nfs-home reports our clientid is in use
Jul 30 15:55:01 dpcl082 kernel: [2880724.526748] NFS: state manager: lease expired failed on NFSv4 server nfs-home with

:
:

The problem does not go away on its own. I have to umount and mount /home on the client to get write access again. /var/log/kern.log on the server doesn't mention this incident.

Since I saw "NFSv4.1: Fix client id trunking on Linux" introduced early this year in the official kernel and included with 3.16.7-ckt7-1 into Debian's kernel I wonder if this rings a bell somewhere?

By now I saw this problem 4 times within the last week. I have more than 100 NFS clients running 24/7, using a static NFS mount of /home and /data. The problem appears to be hard to reproduce.

Every helpful comment is highly appreciated.

Regards
Harri

PS:

/sys/module/nfs/parameters/nfs4_unique_id returns just an empty string. I would have expected the hostname or a similar string here, according to Documentation/filesystems/\
nfs/nfs.txt .

I am absolutely sure that there are no duplicate host names in my net. The IP addresses are fixed, too. Nevertheless, to avoid problems I have changed my setup procedure to set nfs.nfs4_unique_id in grub, as suggested.

I am still hoping for valuable feedback

Regards
Harri

Hi,

I have the same issue with several NFS servers. Cannot find any help.

How did you fixed it with grub ?
And did you go further and find a solution though ?

NFS server debian 8.2
NFS client debian 8.2

Any help would be appreciate.

Pierre

Hi Pierre,

I set

	GRUB_CMDLINE_LINUX="nfs.nfs4_unique_id=e2177c92-1fe6-48db-918b-ddefba900379"

in /etc/default/grub and rebuild the grub.cfg file.

The important point is that every NFS client and every root partition get its own unique id. Be careful, especially if you clone your hosts.

You can generate a new ID using uuidgen on the command line.

Hope this helps
Harri

Hi Harri,

thank you for your time and the tips.

so, I will do that :

1. modify /etc/default/grub with :

GRUB_CMDLINE_LINUX="nfs.nfs4_unique_id=${GRUB_DEVICE_UUID}"

2. See the result :

grub-mkconfig | less -Sr

3. rebuild grub.cfg :

sudo update-grub

4. Reboot

I will let you know, if the error keep going to appear.

thanks.

Pierre