[SOLVED] nfs lock problems

[billr]

I recently upgraded my development server from Mageia 2 Beta 2 to Mageia 2 final. After that, my embedded development board can no longer mount its rootfs via nfs. nfsd and portmap are running on the server and I can mount the exported fs from another Linux box running Mageia 1. This confirms the nfs server part is working fine. I managed to boot the dev board from another system and manually attempt the mount. Here is the error:

Code: Select all

/ # mount 192.168.120.6:/exports/nfsroot-ccardwmx28js /mnt
svc: failed to register lockdv1 RPC service (errno 111).
mount: mounting 192.168.120.6:/exports/nfsroot-ccardwmx28js on /mnt failed: Connection refused
/ #

If I specify the nolock option to mount, it works fine:

Code: Select all

/ # mount -o nolock 192.168.120.6:/exports/nfsroot-ccardwmx28js /mnt
/ #

This is the /etc/exports entry I am using:

Code: Select all

/exports/nfsroot-ccardwmx28js   *(rw,insecure,all_squash,async,no_subtree_check,anonuid=500,anongid=500)

What changed between Beta and final release that might explain this issue?

[billr]

Here's some additional information. Even though using "nolock" from the command line mount worked, when I passed "nolock" on the kernel bootline as a nfsroot option, it still failed to boot, hanging at looking up the RPC port:

Code: Select all

Root-NFS: Mounting /exports/nfsroot-ccardwmx28js on server 192.168.120.6 as root
Root-NFS:     rsize = 4096, wsize = 4096, timeo = 0, retrans = 0
Root-NFS:     acreg (min,max) = (3,60), acdir (min,max) = (30,60)
Root-NFS:     nfsd port = -1, mountd port = 0, flags = 00000240
Looking up port of RPC 100003/2 on 192.168.120.6

You may think this is a problem with my embedded system, but it's really not. It works fine with a non-Mageia 2 host as the server and worked find with Mageia 2 Beta 2 before I did the upgrade to Mageia 2 release.

[tom_]

Hi,

have you checked if it is due to the firewall?

[billr]

Yes, the firewall is open.

It appears to be a problem with the kernel on the embedded system. Starting with a fresh kernel config and rebuilding the kernel seems to have solved the problem, at least for now. Very strange, as it used to work....