nfs: fstab will not mount at boot
 

Hello,
I have a small network of two linux computers.
I can ping and ssh: It works very well.
I have nfs running, and I would like to mount the other computer's filesystem.
I have this line in /etc/fstab :

p4.home.net:/ /mnt/nfs-p4 nfs rw,hard,intr,rsize=8192,wsize=8192 0 0

Which I hoped would mount the p4.home.net root partition at boot, but it does not.

If I type mount p4.home.net:/ as root, the remote filesystem is mounted correctly.

Why doesn't this happen automatically?

Before mounting manually.. what do your logs tell you after bootup? Always check your logs, they usually will tell you everything you need to know.

Thanks for your reply.

I did check the logs, and have now rechecked them: No obvious complaints visible in either /var/log/messages or from dmesg, except for "xinetd[2158]: warning: can't get client address: Transport endpoint is not connected". Is this relevant? If you like, I'll post the files, but they are quite l-o-n-g!

Looking at both the above, it seems to me that it is mounting my local partitions before it is finding my network card, loading the module for it, and starting up the network. Maybe whatever it is that runs through fstab and mounts things doesn't mount the network drive as the network isn't up by then? But nothing is complaining, and when I mount the networked drive manually, it just mounts as it should.

It is the MDK9.1 2.4 kernel machine that is badly behaved, the MDK10.1 2.6 kernel machine behaves as expected and mounts the other computer's drives at boot time. I don't suppose it should be a problem to have both PCs being both nfs servers and clients?

Any other ideas?

I think you have found the problem: You can't mount NFS filesystems before the hardware is initialized. On my SUSE system, this is solved by a runlevel script that runs after network initialization. I can post script if you don't find something similar on your system.

Thanks abisko00,

I thought that maybe I had done something wrong. I'll just put the mount commands at the end of rc.local and that should sort it out.

I am now getting exactly the same problem as tredegar, I am using a Red Hat 9 system.

I can mount the nfs filesystem with mount -t nfs.... command, but not in fstab.
I believe the hardware isn't initialized when the system trying to mount the nfs filesystem.

Would you mind shairing your runlevel script to us?
(I don't want to mount the nfs filesystem with rc.local...)

Thanks!!

SUSE may have a different structure than RH9, so you probably need to adjust the script:
Good luck!

A similar problem ...
 

Dear All,

Apologies if I am missing important details - I am very new at this.

I am trying to mount a windows network share on boot, with little luck. Mounting manually (with mount -t) works fine, so I added a line to fstab. Forcing a remount with mount -a also works fine (although there is an odd error "[mntent]: warning: no final newline at the end of /etc/fstab")

However, when I reboot, the network share does not mount.

I tried adding "sudo mount -a" to rc.local, but no improvement.

Oddly, I was intending to use smb4k, but although the "remount on restart" option was selected, it also failed on reboot!

I like the explanation of not being able to mount until the network has started up, but that does not explain the failure of rc.local, right?

If nothing else, could someone please advise on what logs to check for messages?

Thank you in advance for your help!


----------------------------

Oops - I see I have this in the wrong forum. My apologies - please delete.

That error means exactly what it says, that fstab should end with a <Return>, but it's not fatal. Just add a blank like at the end of fstab to get rid of that error.
rc.local is run as root, so you do not need the sudo
You say you can mount your network drive with mount -t blah blah blah, so I'd suggest you just put that mount command at the end of rc.local and it'll probably work for you.
And welcome to LQ :)

Hello,

Don't add the 'mount -a' option to /etc/rc.local. Add the full 'mount' command to /etc/rc.local ... ie.. mount -t nfs /source /destination'.

Thanks,

Vimal Kumar

Hi everyone,

I've had the same problem with my nfs shares not mounting at boot. (Had to manually mount).
In looking at /var/log/messages I found "link beat not detected". The NIC was working though. I could browse the internet.

I disabled my MCP51 ethernet controller, (Nvidia using the forcedeth module), and fired up my ADM983 Linksys card with tulip....Rebooted, and the nfs share was sitting on the desktop.

Thank you all for the quick response, and the welcome!

True to form I did find the solution, much later last night. I did not end up needing to modify rc.local, although I am sure the proposed solution above would have worked.

The (for me) ideal solution was found elsewhere in the forum. The suggestion was to use the _netdev modifier in the fstab statement. This forces the mount operation to wait until the network connection is up and running. It worked a charm.

For future noob (like me!) reference the correct use is:

//wherever/blah /somewhere/blah cifs guest,_netdev,uid=1000,iocharset=utf8,codepage=unicode,unicode 0 0

Yes, I know the guest authentication sucks, but this is for a windows share which defaults to no authentication required, and all of this is behind a firewall so I am not too worried. As my bravery increases I may fix it later.

Thank you all again. I am VERY sure I will be back!!!!

I am having a similar problem with being able to mount nfs shares manually but not via fstab. For me, it isn't an issue related to hardware initialization. I'm not sure what the problem is, but you can see my full description here:
http://www.linuxquestions.org/questi...y-help-631772/

I've got a similar problem, but I don't think it's because of the service start up order.

I have tried delaying the mount via rc.local (well /etc/rc3.d/S99local which roues there) and still no joy.

I've ried two different ways of delaying the mount:

1) mount -t //share/ etc. in S99Local
2) at now + 3 minutes < /scrips/<scriptWithMountCommand>

Neither have worked.

BUT, when I go to login and run the mount command immediately it works.

I think it's some kind of permissions thing here.

I am mounting a share on one network that it is not a domain member of to a share is then accessible to another completely separate network (via a separate network interface). The server has joined the domain of the latter and once it's mounted the share it then maps AD users from that domain to the share.


What makes me think this is a permissions thing is that the mount partially works after a reboot.

After a reboot if you log into the local server you can see the share has been successfully assigned to the AD group that is on the domain it is a member of, but the share is empty. It's like it mounts the share, assigns it to the correct user group but then can't pull the contents of the share in.

When I try looking at the logs when a user tried to access the share after it has been partially mounted you can see that it knows to map the users to their particular part of the share as errors when trying to get at that part of the share.


I am very stuck trying to work out what it would be.


Has anyone got any experience with this type of setup? I run iptables and selinux but have tried this with both off and still the same error.
I'm mapping users via the UNIX attributes of AD. Like I say all works straight away when you are logged in as root and run the mount command.

...So I've now just tried mounting the share without the part that does the user mapping and it's still an issue. /var/log/messages reports:

CIFS VFS cifs_mount failed w/return code = -13

...so It's nothing to do with the user mapping part. Again, If I then try the mount when logged into the local server it works fine. Does anyone know wha this means? I've had a quick search but couldn't find anything specific enough.

Thanks.