SNMP Bandwidth Monitor

mconway · September 23, 2017, 5:39am

No problem, glad I could help!

Vasiley · September 23, 2017, 11:55am

you can customize in a package add this to the top of Package

homeassistant:
  customize:
    <your sensor here>:   
      friendly_name: Your chosen friendly nam 
      icon: mdi: <your mdi icon>
      hasska_hidden: true  
      unit_of_measurement: '°F'  # if any or delete

mconway · September 23, 2017, 6:24pm

Thanks @Vasiley! I’ve updated the package with the customize section!

mconway · October 8, 2017, 6:58pm

Just an update… I’ve replaced the input_slider with input_number and fixed the automations based on the breaking changes in 0.55. My package file on GitHub is up to date.

fredrike · January 9, 2018, 8:53am

What is the difference between Internet Speed IN and WAN Download Average?

Is it possible to add Daily/Weekly/Monthly usage statistics too?

bonterra · January 9, 2018, 10:22am

Internet speed in is an instanteneous value whereas Download average is just that, an average.

Unfortunately the statistics component is not clever enough to handle several values so you’d have to use three different sensors for your averages. The next problem is what sample size to use as the sensor doesn’t allow you to set a time only, you need a sample size too and it has to be big enough to handle thousands of updates per day. (It defaults to 20 samples if you omit it).

fredrike · January 9, 2018, 10:37am

Ah, great then I understand, average over what time?

All of the different components in hass makes it really hard to understand. You can’t help me with a short example with say usage over day?

Thank you!

bonterra · January 9, 2018, 11:48am

That’s the tricky part, the only answer I can give is “it depends”. The statistics component requires a fixed number of samples to work, the default value is 20. Then there is also a time limit variable called max_age so that values older than that are omitted even if that makes for less than the defined number of samples.

Unfortunately there seems to be no way to use only the max_age variable, making it quite tricky to create statistics for sensors that change unpredictably or very often. An outdoor temperature sensor might change 100 or 200 times per day, that’s not too bad. A bandwidth sensor that updates every second or two on the other hand makes for a huge number of samples over a day (86400) or a month (2.7 M) and I don’t know how well HASS handles such large datasets.

A 24h sensor might look something like this:

sensor:
  - platform: statistics
    entity_id: sensor.internet_speed_in
    sampling_size: 100000
    max_age:
      hours: 24

I have created a Github issue on this matter, asking for a statistics sensor based on time rather than sample size.

fredrike · January 9, 2018, 12:07pm

Great! It should be possible to create a week sensor based on 7 samples of that sensor then. But, what I want is to show the cumulative amount of bytes transmitted (perhaps summing the snmp_wan_out directly)…

bonterra · January 9, 2018, 12:44pm

Well, yes and no. A new sample is generated whenever a monitored value changes so updating an hourly sensor would trigger an update of the daily sensor which in turn would trigger an update of the weekly sensor and so on… You’d need some way to keep the daily sensor from updating every second, I’m not sure on how to do that.

As for the cumulative value, the statistics component offers a number of attributes that you can use in template sensors, in this case you could use {{states.sensor.internet_speed_in.attributes.total}} (or something similar, I can’t test the syntax right now).

lpt2007 · March 18, 2018, 9:41am

I try to configure SNMP to get bandwith of my router similar to this configuration and I get error:

Unable to find service input_number/set_value

sensor.yaml
sensor.yaml (1.3 KB)

automation.yaml
automation.yaml (956 Bytes)

I get numbers from “snmp_wan_in” and “snmp_wan_out”. I thing the problem is in automation script?

lpt2007 · March 18, 2018, 11:19am

Now is working. I add input_number from sensor:yaml to configuration.yaml

katsaplias · May 12, 2018, 7:55pm

hi everybody, I’m trying to get a valid value on the HASSIO from a Mikrotik router through the snmp bandwidth monitor configuration, but I haven’t accomplished it yet.

configuration.yaml

  - platform: snmp
    host: 192.168.1.1
    community: public
    baseoid: .1.3.6.1.2.1.31.1.1.1.6.20
    name: 'PPoE_download'
    unit_of_measurement: 'Mbit/sec'
    value_template: '{{((value | int) / 8 / 1024 / 1024) | int}}'
    accept_errors: true
            
  - platform: snmp
    host: 192.168.1.1
    community: public
    baseoid: .1.3.6.1.21.1.1.1.10.20
    name: 'PPoE_upload'
    unit_of_measurement: 'Mbit/sec'
    value_template: '{{((value | int) / 8 / 1024 / 1024) | int}}'
    accept_errors: true

but the OID DataType is Counter64 and not an Integer

snmpwalk command:
snmpwalk -Os -c public -v 1 192.168.1.1 .1.3.6.1.2.1.31.1.1.1.6.20
ifHCInOctets.20 = Counter64: 8499094646

so how do I convert it in order to get a valid result?

lmamakos · May 12, 2018, 11:15pm

The SNMP MIB variables that you retrieving is returning a running counter of the number of octets (bytes) that have been transmitted (or received, depending on which one) over the interface. This is not a rate. To compute a rate, you need to fetch the counter, wait an interval, fetch the counter again and see how much the counter changed over what interval of time. (Often, you might also retrieve system.sysUpTime.0 each time to get the managed entity’s idea of the elapsed time between the measurements to factor our the latency of sending, processing and returning the response.)

It might be that your router also computes a rate as some proprietary MIB variable, but that’s not the one that you’re retrieving in your example.

I don’t have a specific recommendation on how best to implement this rate computation in Home Assistant.

katsaplias · May 15, 2018, 11:19am

ok thanks Louis, I will dig in more to that

SFoskett · June 7, 2018, 9:59pm

Is anyone using this with the data provided by the UPNP sensors? sensor.igd_received_bytes and sensor.igd_sent_bytes

I imagine it should work, but I haven’t gotten any value out of it yet.

SFoskett · June 8, 2018, 1:13pm

Yay! I got this working with IGD stats provided by upnp: (which is a much simpler and easier data sensor since there are no SNMP strings)!

TL;DR: IGD stats are in Megabytes not Bytes so you have to multiply by 8000000 instead of 8 in automations, but otherwise everything works pretty much the same.

Here’s the relevant entry to configure IGD instead of SNMP data. Everything else is the same as @mconway’s package on GitHub.

automation:
  - id: snmp_monitor_traffic_in
    alias: Monitor Traffic In
    trigger:
      platform: state
      entity_id: sensor.igd_received_bytes
    action:
      - service: input_number.set_value
        data_template:
          entity_id: input_number.internet_traffic_delta_in
          value: '{{ ((trigger.to_state.state | int - trigger.from_state.state | int) * 8000000 ) / ( as_timestamp(trigger.to_state.last_updated) - as_timestamp(trigger.from_state.last_updated) ) }}'
  
  - id: snmp_monitor_traffic_out
    alias: Monitor Traffic Out
    trigger:
      platform: state
      entity_id: sensor.igd_sent_bytes
    action:
      - service: input_number.set_value
        data_template:
          entity_id: input_number.internet_traffic_delta_out
          value: '{{ ((trigger.to_state.state | int - trigger.from_state.state | int) * 8000000 ) / ( as_timestamp(trigger.to_state.last_updated) - as_timestamp(trigger.from_state.last_updated) ) }}'

Juggler · July 5, 2018, 2:35pm

Thanks for your work, and thanks to @mconway for the package!

Trying to read through the .yaml file and understand what is happening. Two questions:

Why are you using an “input_number” (slider) for the delta. This creates a visible element on the Dashboard; is there any reason I would ever need to manually change this?
@mconway has a sensor.ping_time_mean, but I don’t see any associated code? Is this contained elsewhere?

ariel · July 11, 2018, 4:58am

I implemented bandwidth monitoring, based on the @mconway github sample (thanks!), and it worked awesome.

Now - my Asus RT3100 router reports strangely similar IfInOctets and IfOutOctets on the WAN (eth0) interface which makes HASS report similar in and out bandwidth, specially noticable when streaming netflix (17Mbps in and about the same out, the “in” value is accurate but the “out” is not) - which is obviously wrong.
Turns out there is an explanation for the strange behavior, details here:

eth 0, ifInOctets 1.3.6.1.2.1.2.2.1.10.5
eth0 ifOutOctets: 1.3.6.1.2.1.2.2.1.16.5

In my case I opted to show 3 “out” values, one calculated using “eth2” “in” (wifi 5g), “eth1” “in” (2.4Ghz) and the other vlan1 (the 4 port wireline switch). All measurement tests for inbound and outbound from either radio or the wired connections are now OK (validated with speedtests in different devices).

ariel · July 11, 2018, 9:34pm

Well the above 4-OID solution worked perfectly but it turned out to be too much for the RPi3b, HASS CPU utilization went throught the roof together with temps.

Turns out that the snmp platform is very inefficient (there is a pull request meant to help on that front). I changed the approach and used “command_line” (snmpget command) instead. Then to further optimize I merged all 4 OIDs queries into one command and used a template to split the output. CPU utilization and temps decreased a lot. Finally, to further minimize CPU use (I use the same RPi as a headless Kodi music appliance, with a hifi-berry card), I went back from 4 sensors to 2, by using a template to add up Wired and both Wifi ‘output traffic’ into a single sensor.

Really pleased with the result, CPU load increased only slighly even with 10-second polling and 4 OIDs, this is the relavant bit:

  - platform: command_line
    name: snmp_multiple_oids   # scan 4 OIDs in one shot
    command: snmpget -v 2c -c your_community -Oqv 192.168.1.1 1.3.6.1.2.1.2.2.1.10.5 1.3.6.1.2.1.2.2.1.10.9 1.3.6.1.2.1.2.2.1.10.8 1.3.6.1.2.1.2.2.1.10.7  | tr '\n' ':'
    scan_interval: 10
  - platform: template
    sensors:
      snmp_wan_in:
       value_template: "{% set list = states.sensor.snmp_multiple_oids.state.split(':') %} {{ list[0] }}"
      snmp_wan_out:   # adding up all three 'output traffic' OIDs to avoid having to present them separately (high cpu)
        value_template: "{{ states.sensor.snmp_multiple_oids.state.split(':')[1] | float +
        states.sensor.snmp_multiple_oids.state.split(':')[2] | float +
        states.sensor.snmp_multiple_oids.state.split(':')[3] | float | round (0)}}"