[SOLVED] Cannot mount NFS shares anymore after latest updates (Dec 25)

lems · 12-26-2015, 03:58 AM

Hello,

after applying the updates from the Dec 25 update, I can no longer mount any NFS shares using my Raspberry Pi; it was working before with the updates from the 21st December. They are served by another Raspberry Pi running current (Dec 25); I can mount them just fine on my x86_64 desktop, which is also running -current (Dec 23).

This is what I get:

Code:

# mount -v 192.168.1.6:/share/nfs /mnt/tmp
mount.nfs: timeout set for Sat Dec 26 10:52:39 2015
mount.nfs: trying text-based options 'addr=192.168.1.6'
mount.nfs: prog 100003, trying vers=4956184, prot=6
mount.nfs: trying 192.168.1.6 prog 100003 vers 4956184 prot TCP port 2049
mount.nfs: portmap query retrying: RPC: Program/version mismatch
mount.nfs: prog 100003, trying vers=4956184, prot=17
mount.nfs: trying 192.168.1.6 prog 100003 vers 4956184 prot UDP port 2049
mount.nfs: portmap query failed: RPC: Program/version mismatch
mount.nfs: Protocol not supported

Any other information needed?

edit: It seems I can mount them once I use '-o rw,vers=3'. This was not needed before, though.

drmozes · 12-27-2015, 03:02 AM

Quote:

Originally Posted by lems

Hello,

Any other information needed?

edit: It seems I can mount them once I use '-o rw,vers=3'. This was not needed before, though.

I've spent a few hours now debugging this and can't work out what's wrong with it, apart from that it seems to be mount.nfs that's broken (because using just mount.nfs from the previous nfs-utils v1.3.1 works immediately).
I'm going to test some kernel changes first though out of curiosity before reverting to 1.3.1

Which kernel are you using?

Drakeo · 12-27-2015, 04:15 AM

you may need to make it executable /etc/rc.d/rc.rpc I believe this was a install-new so this would chould cause your issue.

Code:

#!/bin/sh
# rc.rpc:  start/stop/restart RPC daemons needed to use NFS.
#
# You must run these daemons in order to mount NFS partitions
# (unless you use the mount option '-o nolock', which can
# corrupt files and is not generally recommended unless you
# are mounting the partition(s) as read-only).
#
# To run an NFS server, starting these is mandatory.
#

rpc_start() {
  if [ -x /sbin/rpcbind -a -x /sbin/rpc.statd ]; then
    if ! ps axc | grep -q rpcbind ; then
      echo "Starting RPC portmapper:  /sbin/rpcbind -l $1"
      /sbin/rpcbind -l $1
    fi
    if ! ps axc | grep -q rpc.statd ; then
      echo "Starting RPC NSM (Network Status Monitor):  /sbin/rpc.statd"
      /sbin/rpc.statd
    fi
  else
    echo "WARNING:  Cannot start RPC daemons needed for NFS.  One or more of"
    echo "          these required daemons is not executable or is not present"
    echo "          on your system:"
    echo
    echo "          /sbin/rpcbind or /sbin/rpc.statd"
    echo
  fi
}

rpc_stop() {
  killall rpc.statd 2> /dev/null
  sleep 1
  killall rpcbind 2> /dev/null
  sleep 1
  killall -9 rpc.statd 2> /dev/null # make sure :)
  sleep 1
  killall -9 rpcbind 2> /dev/null # make sure :)
  sleep 1
}

rpc_restart() {
  rpc_stop
  rpc_start
}

case "$1" in
'start')
  # Warm restart by default (see "man rpcbind" for details about the -w option)
  rpc_start -w
  ;;
'cold_start') # Start without -w option
  rpc_start
  ;;
'stop')
  rpc_stop
  ;;
'restart')
  rpc_restart
  ;;
*)
  echo "usage $0 start|stop|restart"
esac

lems · 12-27-2015, 10:03 AM

Quote:

Originally Posted by drmozes

Which kernel are you using?

I'm using kernel 4.1.15+* (client+server) installed via rpi-update. Thank you for looking into this.

* uname -a output:

Code:

Linux edmond 4.1.15+ #830 Tue Dec 15 16:58:28 GMT 2015 armv6l BCM2708 GNU/Linux

@Drakeo: rc.rpc is executable on my system:

Code:

$ ls -l /etc/rc.d/rc.rpc
-rwxr-xr-x 1 root root 1526 Dec 15 08:55 /etc/rc.d/rc.rpc*

drmozes · 12-27-2015, 05:45 PM

Quote:

Originally Posted by lems

I'm using kernel 4.1.15+* (client+server) installed via rpi-update. Thank you for looking into this.

As I was about to push out the updates with a reversion to 1.3.1, it suddenly occurred to me to do this:

Code:

sed -i '/Defaultvers=3/avers=3' /etc/nfsmount.conf

This just adds vers=3 to the config file in the proper location.

I'll do some more tests on a few machines here but it seems to work so far.

lems · 12-29-2015, 12:20 AM

Your fix seems to work. Thank you, Stuart.

slugman · 05-30-2016, 12:51 PM

Hey guys, I just rsync'd slackwarearm-current yesterday, and I still encountered this issue. Thankfully I found this post and ran the command above (which simply adds the line 'vers=3' after 'Defaultvers=3'), which resolved it right away. However, you may want to go ahead and patch it.

drmozes · 05-31-2016, 01:29 AM

Quote:

Originally Posted by slugman

Hey guys, I just rsync'd slackwarearm-current yesterday, and I still encountered this issue. Thankfully I found this post and ran the command above (which simply adds the line 'vers=3' after 'Defaultvers=3'), which resolved it right away. However, you may want to go ahead and patch it.

This was fixed last year and the config now ships with that setting, so I suspect you upgraded it but the .new file is left over.

Interesting though because I thought the problem was fixed without the need for that config file change.