Quick and dirty gateway metrics for telegraf on pfSense

Additional configuration for Telegraf

[[inputs.exec]]
        commands = ["/usr/local/libexec/telegraf/gateways.py"]
        timeout = "5s"
        data_format = "influx"

gateways.py (for pfSense < 2.4.5)

#!/usr/local/bin/python2.7

import glob, os, socket

DPINGER_SOCK_PATH = "/var/run/"

os.chdir(DPINGER_SOCK_PATH)

for sock_name in glob.glob("dpinger*.sock"):
    sock = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
    sock_path = DPINGER_SOCK_PATH+sock_name
    s = sock.connect(sock_path)
    line = sock.recv(1024).split('\n', 1)[0]
    values = line.split()
    print "gateways,gateway_name=" + values[0] + " rtt=" + str(int(values[1])/100.0) + \
          ",rttsd=" + str(int(values[2])/100.0) + ",loss=" + str(int(values[3])) + "i"
    sock.close()

gateways.py (for pfSense >= 2.4.5)

#!/usr/local/bin/python3.7

import glob, os, socket

DPINGER_SOCK_PATH = "/var/run/"

os.chdir(DPINGER_SOCK_PATH)

for sock_name in glob.glob("dpinger*.sock"):
    sock = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
    sock_path = DPINGER_SOCK_PATH+sock_name
    s = sock.connect(sock_path)
    line = sock.recv(1024).decode().split('\n', 1)[0]
    values = line.split()
    print("gateways,gateway_name="+values[0]+" rtt="+str(int(values[1])/100.0)+ \
          ",rttsd="+str(int(values[2])/100.0)+",loss="+str(int(values[3]))+"i")
    sock.close()

Trying to use this on the new version of pfsense which is now python 3.7 updating just the python line at the top produces

File "telegraf_gateways.py", line 17 print "gateways,gateway_name=" + values[0] + " rtt=" + str(int(values[1])/100.0) + \ ^ SyntaxError: invalid syntax
Adding parentheses around the print part brings up this error.

Traceback (most recent call last): File "./telegraf_gateways.py", line 15, in <module> line = sock.recv(1024).split('\n', 1)[0] TypeError: a bytes-like object is required, not 'str'
some googling I found that this new byte-like option was introduced in 3.x so I am guessing this script will need to be updated?

I've created a python 3.7 version of this script for pfSense 2.4.5 and above

I've created a python 3.7 version of this script for pfSense 2.4.5 and above

Thank you so much! I was really struggling trying to convert this to Python3.

Works like a charm fastjack! thanks!

Reading https://forum.netgate.com/topic/152132/grafana-dashboard-using-telegraf-with-additional-plugins/19 I noticed that some people were getting ping times that were orders of magnitude too small. This problem never occurred to me. I had set the unit of my RTT/RTTsd graph to "Time/miliseconds (ms)" in Grafana.

The script works correctly and the data goes back to grafana. The script stops executing after a while. I have to relaunch the gateway.py script and the data is going up for a while. I just checked my telegraf.conf file. It has not undergone any modification.

This file is automatically generated by pfSense

[agent]
interval = "10s"
round_interval = true

[[inputs.cpu]]
percpu = true
totalcpu = true
fielddrop = ["time_*"]

[[inputs.disk]]
ignore_fs = ["tmpfs", "devtmpfs"]

[[inputs.diskio]]

[[inputs.kernel]]

[[inputs.mem]]

[[inputs.net]]

[[inputs.processes]]

[[inputs.swap]]

[[inputs.system]]

[[inputs.pf]]
[[outputs.influxdb]]
urls = ["http://grafana:8086"]
database = "influx_db"
username = "xxxxx"
password = "xxxxx"

Additional Raw Options

any idea how to get the gateway data without the dpinger service?
I had to deactivate the gateway lookup, because of some strange conection problems, as discribed here https://be-virtual.net/pfsense-arpresolve-cant-allocate-llinfo-for-x-x-x-x-on-emx/

@corgan2222 You probably could cobble together something with ping. I used dpinger because it already comes configured through pfSense's admin interface.

But if you're having problems pinging your already configured gateway you can choose a custom ping target (the user interfaces calls it a "Monitor IP"). Just go to System -> Routing -> Gateways and then edit your gateway from there.

Why the /100 - the default output is microseconds and that seems fine?

I believe I had my Grafana already set up for seconds instead of microseconds so I added it here.

Ah okay, that would be /1000 but either way thanks for saving me some time with this. Works well on opnsense too