[7.1-RELEASE-p2 amd64] NFS mount in fstab hangs during mountcritremote execution

[Arjan van der Oest]

Hi,

I’m puzzled and either I don’t understand the boot rc.d process or there is something wrong with it ☺

I have this 7.1-RELEASE-p2 amd64 machine compiled with a GENERIC kernel, so NFS support is baked right into the kernel by default. In fsstab I have this entry:

<nfs server ip>:/data/nfs-shares/S1018SR18 /nfs-mounts/backupsrv           nfs          rw            0    0

Further, as per the handbook, I’ve set nfs_client_enable to “YES” in the rc.conf. During boottime the machine hangs on the console after stating “Mounting NFS filesystems:”. I notice that _immediately_ after that line there is a console message saying “em0: link state changed to UP”. Then the system repeats this line until eternity (well, the max I’ve been waiting has been 30 minutes), no . appears indicating the system has not yet completed the 'mount -a 't nfs' command from the mountcritremote script.

[udp] <nfs server ip>:/data/nfs-shares/S1018SR18: RPGPROC_MNT: RPC: Timed out

Hitting CTRL-C forces the machine to continue the boot process, aborting the mountcritremote script. What strikes me is that the actual NFS share has been mounted, although the boot-process seems to indicate otherwise... After doing a umount and mount –a –t nfs again the machine has no problem whatsoever.

Something else that puzzles me, when I set the nfs_client_enable=”NO” line in the rc.conf, the same happens : console hangs on the mountcritremote script until I hit CTRL-C and after that the share has been mounted anyway? Shouldn’t the machine ignore nfs filesystems with this rc.conf config? 

I’ve removed the entry in fstab and set a line in rc.local and then the boot-process works fine without interruption. 

So I'm currently lost with these questions:

- why does the system tries to mount the nfs filesystem from the fstab while nfs_client_enable has been set to no in rc.conf?
- why does the system seems to hang on the mountcritremote script although there seems no valid reason for that. I can imagine the network has not been fully configured yet when executing (indicated by the link UP message right after the mouning NFS filesystems line) but why will the script not continue after a few timeouts? And more bizarre: when interrupting the mountcritremote script the share has been actually mounted, so it seems the 'mount -a -t nfs' command has actually been executed successfully.

Any fingerpoints?

-- 
Met vriendelijke groet / Kind Regards,
Worldmax Operations B.V.
 
Arjan van der Oest
Network Design Engineer

T.: +31 (0) 88 001 7912
F.: +31 (0) 88 001 7902
M.: +31 (0) 6 10 62 58 46
 
GPG: https://keyserver.pgp.com/ (Key ID: 07286F78, fingerprint: 2E9F 3AE2 0A8B 7579 75A9  169F 5D9E 5312 0728 6F78)