Configure NFS ports on client VMs

This page shows you how to set the ports for the statd and nlockmgr daemons on your client VMs to make firewall configuration easier.

Filestore uses the nlockmgr and statd daemons to enable file locking. The ports for these services need to be properly exposed on your client VMs through firewall rules so that clients can properly use locks. We recommend setting the statd and nlockmgr ports so that they are consistent on all client VMs, making it easier to configure ingress firewall rules.

For more information on determining if you need to configure a firewall rule for the VPC network, see Configure firewall rules.

Check port settings

To check what values the statd and nlockmgr ports are currently set to, run the following commands on the client VM instance. If the files don't exist, or if the options don't have values, then the ports aren't set. In that case, the daemons are dynamically assigned arbitrary available ports.

Debian/Ubuntu

To determine the statd port, run the following command and look at the STATDOPTS value:
```
cat /etc/default/nfs-common
```
To determine the nlockmgr port, run the following command and look at the nlm_tcpport and nlm_udpport values:
```
cat /etc/modprobe.d/lock.conf
```

RHEL/CentOS

To determine the statd port, run the following command and look at the STATD_PORT value:
```
cat /etc/sysconfig/nfs
```
To determine the nlockmgr port, run the following command and look at the nlm_tcpport and nlm_udpport values:
```
cat /etc/modprobe.d/lock.conf
```

SUSE

Run the following command:

cat /etc/sysconfig/nfs

The statd port is listed under STATD_PORT and the nlockmgr port is listed under LOCKD_TCPPORT and LOCKD_UDPPORT.

Windows

Setting NFS ports is not required on Windows.

Set ports

To set the statd and nlockmgr ports, run the following commands on the client VM instance. These examples use the nano text editor, but you can use any text editor. These examples also use 2046 as the port for statd and 4045 as the port for nlockmgr because these values are common choices. You can use different ports based on your network configuration. In this case, the ingress firewall rules must allow traffic to the specific ports that you use.

Debian/Ubuntu

Set the statd port:
1. Open the /etc/default/nfs-common file for editing:
```
sudo nano /etc/default/nfs-common
```
2. Set the STATDOPTS option:
```
STATDOPTS="-p 2046"
```
3. Save the file and exit.
Set the nlockmgr port:
1. Create the /etc/modprobe.d/lock.conf file:
```
sudo nano /etc/modprobe.d/lock.conf
```
2. Set the nlm_tcpport and nlm_udpport options:
```
options lockd nlm_tcpport=4045
options lockd nlm_udpport=4045
```
3. Save the file and exit.

RHEL/CentOS

Set the statd port:
1. Open the /etc/sysconfig/nfs file for editing:
```
sudo nano /etc/sysconfig/nfs
```
2. Set the STATD_PORT option:
```
STATD_PORT=2046
```
3. Save the file and exit.
Set the nlockmgr port:
1. Create the /etc/modprobe.d/lock.conf file:
```
sudo nano /etc/modprobe.d/lock.conf
```
2. Set the nlm_tcpport and nlm_udpport options:
```
options lockd nlm_tcpport=4045
options lockd nlm_udpport=4045
```
3. Save the file and exit.

SUSE

Set the statd and nlockmgr ports:

Open the /etc/sysconfig/nfs file for editing:
```
sudo nano /etc/sysconfig/nfs
```
Set the STATD_PORT, LOCKD_TCPPORT, and LOCKD_UDPPORT options:
```
STATD_PORT=2046
LOCKD_TCPPORT=4045
LOCKD_UDPPORT=4045
```
Save the file and exit.

Windows

Setting NFS ports is not required on Windows.

Verify ports are open

To verify NFS ports have been opened properly, complete the following steps.

Install the following dependencies.
Debian/Ubuntu
From the command line, enter the following command:
```
sudo apt install nfs-common tcpdump tshark
```
RHEL/CentOS
From the command line, enter the following command:
```
sudo yum install nfs-utils tcpdump wireshark
```
SUSE
From the command line, enter the following command:
```
sudo zypper install nfs-client tcpdump wireshark
```
Windows
This verification process is not supported on Windows.

Create a script file called verify-nfs-port-script.sh, copy and paste the following script within it, and save it locally to your machine. Note the location of the file and save it for the next step.

#!/bin/bash

# This script is intended to run on client machines to verify that the ports
# are properly open to allow the reception of NLM GRANT messages from the server.

set -eu

function kill_descendants() {
   for pid in $(ps -o pid= --ppid "$1")
   do
   kill_descendants "$pid"
   done
   if [[ $1 -ne $$ ]]; then
   kill "$1" 2>/dev/null | true
   fi
}

function cleanup {
   set +eu

   # Kill all background jobs and wait for it to end, makes sure locks are released
   kill_descendants $$

   # Wait for jobs to die and locks to be released, so mount is not busy
   sleep 2

   umount -f "$MNT1"
   umount -f "$MNT2"

   rmdir "$MNT1" 2&> /dev/null || true
   rmdir "$MNT2" 2&> /dev/null || true
}

function print_help {
   echo "$0 [server_ip] [mount_path]"
   echo -e "\t For example, if you mount a server using:"
   echo -e "\t\t \"mount 10.0.0.1:share /mnt/mount_point\""
   echo -e "\t Run the script: "
   echo -e "\t\t \"$0 10.0.0.1 share\""
}

if [ $# -ne 2 ]; then
   print_help
   exit 1
fi

if [ $(id -u) -ne 0 ]; then
   echo "Failure! This script needs to run as root, use \"sudo $@\""
   exit 1
fi

if ! [ -x "$(command -v tshark)" ]; then
   echo "The 'tshark' command does not exist and is needed for the script. Please install it"
   exit 1
fi

if ! [ -x "$(command -v tcpdump)" ]; then
   echo "The 'tcpdump' command does not exist and is needed for the script. Please install it"
   exit 1
fi

SERVER_IP=$1
MOUNT_PATH=$2

MNT1=$(mktemp -d)
MNT2=$(mktemp -d)

trap cleanup EXIT

echo "Mounting..."
mount -o nosharecache "$SERVER_IP":"$MOUNT_PATH" "$MNT1"
mount -o nosharecache "$SERVER_IP":"$MOUNT_PATH" "$MNT2"

REC_FILE=$(mktemp /tmp/nlm_recording_XXXXXXXX.pcap)
tcpdump -i any -s0 -w "$REC_FILE" "host $SERVER_IP" &
TCPDUMP_PID=$!
echo "Recording TCP dump to $REC_FILE"

sleep 5 # wait for tcpdump to start running

echo "Running test..."
flock "$MNT1"/lock_file -c "echo -n \"Got first lock: \" && date && sleep 5 && echo -n \"Releasing first lock: \" && date" &
sleep 2 # Wait for the first lock to actually be taken

echo "Waiting for second lock: $(date)"
flock "$MNT2"/lock_file -c "echo -n \"Got second lock: \" && date"

sleep 2 # Wait for tcpdump to record everything
kill $TCPDUMP_PID

# For quick analysis inspect recording with tshark, if you don't have it just inspect with Wireshark
echo "Inspecting results in $REC_FILE with TShark"
tshark -r "$REC_FILE" -Y nlm # First, print the output

tshark -r "$REC_FILE" -Y nlm 2>/dev/null | grep -q GRANTED
EXIT_CODE=0
if [ $? -eq 0 ]; then
   echo "The NLM GRANT message is working properly!"
   EXIT_CODE=0
else
echo "The NLM GRANT message is not working properly!"
EXIT_CODE=1
fi
echo "For debugging, please provide the printed output of the script, and $REC_FILE"
exit ${EXIT_CODE}

Enter the following command:
```
chmod +x SCRIPT_PATH
```
Replace the following:
- SCRIPT_PATH: the path where your script file is located. This should be run as root, otherwise add sudo to the beginning of the command.
Enter the following command:
```
SCRIPT_PATH INSTANCE_IP SHARE_NAME
```
Replace the following:
- SCRIPT_PATH: the path where your script file is located. This should be run as root, otherwise add sudo to the beginning of the command.
- INSTANCE_IP: the IP address of the Filestore instance
- SHARE_NAME: the name of the file share
If the port is open, the script returns the following response:
```
The NLM GRANT message is working properly!
```
If the port is not open, the script returns the following error:
```
The NLM GRANT message is not working properly!
```

Configure NFS ports on client VMs

Check port settings

Debian/Ubuntu

RHEL/CentOS

SUSE

Windows

Set ports

Debian/Ubuntu

RHEL/CentOS

SUSE

Windows

Verify ports are open

Debian/Ubuntu

RHEL/CentOS

SUSE

Windows

What's next