0.107.6 broke Device Tracker

I have 2 HA’s running. Pi4 is the main one and my old Pi3B+ is used for adguard and as a zizbee bridge. Pi3 is running 107.7 and the device tracker on that is working fine with a few test devices to track and has been for days. I tried updating the Pi4 again last night and device tracker worked for about 12 hours before failing again - back to 107.5.

Wonder if it is the amount of devices that it is tracking that is causing the issue?

There were no logs on the Pi4 when it failed as well.

Has anyone updated to 108 yet? To see if it’s fixed?

I have but too early to tell just yet. I think not but we will see.

update
Well the answer is no, it hasn’t been fixed yet

1 Like

It may be too early to say I have found a way to fix this but I may have. I installed arpscan yesterday through HACS and added the code via my device tracker file and none of my devices have gone offline yet.

It does seem to add random devices to known_devices.yaml but if it works I’m fine with that.

Afraid to say that this did not fix it - it could go 12+ hours before failing or a few hours - it is very random and annoying.

I am going to test using ping instead of NMAP and see if that is more stable. Have about 4 devices now using it and will update with the results on the next failure.

Hi - I have been having the same problem. I read on the Person integration about the clashes that can happen so I removed all persons and # out the person from config. I then stripped all the known devices apart from the house mobile phones. I then reloaded persons from the services and rebooted.

I have two nmap scanners running both are set with the scheduled scan interval 0:00:20. On reboot it showed in the error log as having 2 sets one with the default 12 and the other 20 which repeated. After 40 mins it settled down to showing both 20.

I have also noticed the known_devices has changed in the descriptions of the devices. Everything appears to be running showing "‘home’ that I want to except the error log of ‘Updating device list from legacy took longer than the scheduled scan interval’

Just thought I’d share the observations if it helps anyone.

I have removed Person myself now after your update but have finally taken everything off NMAP and have disabled it as well. Lets see how Ping works for me.

Some devices like Nest & Alexa seem to block ICMP so I cannot check their network status via Ping but that is a small price to pay if it is stable.

re-trying asuswrt myself here, but a bit puzzled by the lack of configuration options for track_new_devices, exclude, consider_home, and interval_seconds…

cant believe these device_trackers are so neglected and immature all over tbh. Aren’t we missing some fundamentals elsewhere ?

I have some good news and bad.

Good news I’ve resolved it - The only problem is I had to reinstall HA from scratch. Haven’t had a single error.

Using:

  - platform: nmap_tracker
    hosts: 
      - !secret include_subnets1
      - !secret include_subnets2
    exclude:
      - !secret exclude
    scan_options: " --privileged -sn "
    interval_seconds: 30
    home_interval: 120
    consider_home: 0:01:00

What was interesting from the loading a new HA was that a lot of the configuration in Yaml such as that within addons such as DuckDNS and Samba and MQTT had all changed. Again I no expert in Software code but thought might help.

Bad news is 0.108.3 has a massive problem with an error that is being displayed all over the logs and doesn’t appear to have definte pattern “Event is not JSON serializable” and was the reason for having to reinstall from scratch because it crashed my SD Card.

https://github.com/home-assistant/core/issues/34134

I found the only way to stop the messages was to hash out the sensors its a bit of nightmare. Patiently awaiting a fix

# binary_sensor: !include binary_sensor.yaml

It does appear to be a Bayesian sensor problem

I’m using Home Assistant (HassOS / Hassio) so what does that mean?
Do I need to re-flash HassOS and restore? Or just a full restore from a snapshot will do it?

Well I am too and still not seeing any of the errors and concerning the Bayes problem I rewrote the Bayes sensors without the use of Value Templates and works fine and now running the current latest version 0.108.4. I wasn’t happy about having to do a rebuild but as mentioned previously in this issue it appears there have been some fundamental changes to the code concerning the direction of Yaml. I may be wrong but I now have a fresh build.

Fingers crossed that 0.108.7 will fix device tracker after this issue with Modbus was resolved.

I don’t understand the relation between nmap device tracker and modbus.
I have updated and I already have lots of logs (took more than 20s). Will wait but I don’t think it is solved for nmap

I found closed issue here that mentioned Modbus affecting NMAP.

that would be surprising, since not using Modbus at all, people are still affected by Nmap being broken.

I can confirm that it is not solved with 108.7

Damn, that is annoying :rage:

confirm issue remain on 108.9
known_device.yaml is not being updated by nmap.
configuration :

 - platform: nmap_tracker
   hosts: 192.168.1.1-250
   interval_seconds: 15
   track_new_devices: true
   consider_home: 300

Since adding the scan options line mine has been stable for over 3 days. Currently running 0.108.8 as I don’t want to restart HA since I added this line.

image

1 Like

Thanks for the suggestion, still seeing issues though:
2020-04-27 15:30:00 WARNING (MainThread) [homeassistant.components.device_tracker] Updating device list from legacy took longer than the scheduled scan interval 0:00:15