DD-WRT device_tracker multiplies known_devices entries

Configuration
1

Hi,

I’m using the ddwrt component to track devices in my local network:

platform: ddwrt
host: 192.168.a.b
username: xxx
password: yyy

this works, but apparently all devices that are discovered that way are added to the known_devices.yaml file, regardless if they were already in there or not. This seems to happen when HA is restarted, maybe even at other times, too.
I just found over 50 instances of my fridge’s network interface in there…

This also means, I can’t manually assign attributes like name:, icon: etc. since it’s all superseded by another vanilla copy of the device entry on restart.

This looks like a bug to me - to my understanding, HA shouldn’t re-add a device that’s already in the known_devices file.

Sebastian

2

To clarify:

# date
Do 13. Jul 16:37:21 CEST 2017
# grep 68a86d2xxxx: known_devices.yaml  | wc -l
68
# docker restart hass
hass
# date
Do 13. Jul 16:37:41 CEST 2017
# sleep 60; grep 68a86d2xxxx: known_devices.yaml  | wc -l
69
# date
Do 13. Jul 16:38:57 CEST 2017
3

I had an issue like this but it turned out I corrupted the known_devices file by adding a blank in a name. HA would then stop reading the file and just add all devices to the end, over and over again.

Check your file and I think there was some warning in the log as well.

4

Thanks for that tip! I’ll look into it.

Sebastian

5

Ok, you were right!
Turning up the log level revealed some “duplicate entry” messages for the known_devices.yaml file.

I wrote a little python script to clean up the file. Maybe someone else finds it useful too:

#!/usr/bin/python3
#
# clean up duplicate entries from home-assistant's known_devices.yaml file
#

import sys
from collections import defaultdict

def main(known_devices_file):
  known_devices = defaultdict(dict);
  current_dev = None;

  # build the data structure; remove most if the identical dupes
  for line in open(known_devices_file, 'r'):
    if line.startswith('  '):
      try:
        (attr, value) = (line.rstrip().lstrip()).split(': ', 1)
      except:
        # attributes that don't have values set
        attr = line.rstrip(': \n').lstrip()
        value = ''
      if (attr == 'name' and not value.startswith('\'')):
        value = '\'' + value + '\''
      known_devices[current_dev][attr] = value
    else:
      current_dev = line.rstrip()


  # weed out duplicates of entries with custom names set
  keys_to_delete = set();
  for dev in known_devices.keys():
    dupelist = [];
    mac = known_devices[dev]['mac']
    if not mac == '':
      for otherdev in known_devices.keys():
        if (known_devices[otherdev]['mac'] == mac):
          dupelist.append(otherdev)
      if len(dupelist) > 1:
        # drop entries that only have names derived from mac address
        mac_name = mac.replace(':','').lower() + ':'
        dupelist.remove(mac_name)
        keys_to_delete.add(mac_name)
        if len(dupelist) > 1:
          print ('still having issues with this one: ' + dupelist)

  for key in keys_to_delete:
    del known_devices[key]

  for key in known_devices.keys():
    print(key)

if __name__ == "__main__":
    main(sys.argv[1])

Usage: cleanup_known_devices.py /PATH/TO/known_devices.yaml

The script will read the file and print out a cleaned up version of it to STDOUT, which can be redirected to a new known_devices.yaml file.

There’s not much error checking - use at own risk.

Sebastian