JTK – Notes and Ramblings Things we have found …

After an update to Raspberry Pi OS Trixie and a great deal of hassle with sox not being willing to handle mp3 format in spite of libsox-fmt-mp3 being installed the Weather Radio scripts have been updated again.

The start.sh script is unaltered from before:

cd weatherRadio
rm -f one two three; mkfifo one two three

./to-same.sh <one &
./to-lame.sh <two &
./to-udp.sh <three &
rtl_fm -f 162550000 -s 22050 -p 14 | tee one two three |multimon-ng -t raw -a EAS /dev/stdin

The tosame.sh script is also unaltered:

multimon-ng -t raw -a EAS /dev/stdin | python3 ~/tg-dsame/dsame.py --mqtt your-mqtt-broker --json output.json --call /home/pi/weatherRadio/playRadio.sh --command boo

The to-lame.sh script was updated to once again use lame instead of sox for the mp3 conversion. While ez-stream can handle other formats mp3 works well from lame.

lame --bitwidth 16 --signed -s 22050 --lowpass 3500 --abr 64 --scale 8 -r -m m - - |ezstream -c ezstream.conf

The to-udp.sh script was greatly simplfied and sox was removed completely from it:

socat -u - udp4-sendto:localhost:5555

The playRadio.sh script was updated to handle playing a raw stream instead of mp3:

#!/bin/bash
nc -lu 5555|play -t raw -r 22050 -b 16 -e signed -c 1 -v 7 -&
sleep 120
killall "play"

For completeness, the ezstream.conf file:

<?xml version="1.0" encoding="UTF-8"?>
<ezstream>

  <servers>
    <server>
      <protocol>http</protocol>
      <hostname>somemachine</hostname>
      <password>hackmeplease</password>
    </server>
  </servers>

  <streams>
    <stream>
      <mountpoint>/weather</mountpoint>
      <format>MP3</format>
      <stream_name>The Weather Radio</stream_name>
      <stream_url>yoururl</stream_url>
      <stream_genre>weather</stream_genre>
      <stream_description>NOAA Radio KHB59 (162.55MHz)</stream_description>
      <stream_quality>2.0</stream_quality>
      <stream_bitrate>32</stream_bitrate>
      <stream_samplerate>48000</stream_samplerate>
      <stream_channels>1</stream_channels>
    </stream>
  </streams>

  <intakes>
    <intake>
      <type>stdin</type>
      <stream_once>yes</stream_once>
    </intake>
  </intakes>
</ezstream>

The Weatherflow Tempest unit has an API based on local broadcast UDP packets. There is also a websocket and a REST API.

This article will focus on using the UDP API with MQTT for use in Openhab. There are a number of resources online. There is a binding that started in 2020 but the discussion stops in 2022. There is a different binding from Bill Welliver that seems more complete.

Additionally there are resources for Home Assistant. https://github.com/briis/hass-weatherflow2mqtt has been depreciated and the git set to read only. It has been replaced with https://github.com/briis/weatherflow_forecast. The newer integration uses a combination the Weaterflow APIs for a more complete setup and also works with Home Assistant discovery. There is also https://github.com/gualandd/WeatherFlow-Tempest-UDP using NodeRed.

All of the options mentioned above seem a bit too heavy for a simple sensor to MQTT path. Here is a bit of python3 code to listen to the UDP port and send to a MQTT broker.

import socket
import paho.mqtt.publish as mqtt
import json

# Define the UDP port to listen on
UDP_IP = "0.0.0.0"  # Listen on all available interfaces
UDP_PORT = 50222
mqtt_host = "your.mqtt.broker"

# Create a UDP socket
sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)

# Bind the socket to the specified address and port
sock.bind((UDP_IP, UDP_PORT))

print(f"Listening on {UDP_IP}:{UDP_PORT}...")

# Listen for incoming packets and print them
while True:
    data, addr = sock.recvfrom(1024)  # Buffer size is 1024 bytes
    js = json.loads(data)
#    print(json.dumps(js, indent=4))
    match js['type']:
        case "obs_st":
            mqtt.single('Tempest/obs',json.dumps(js['obs']), hostname=mqtt_host)
            print(f"OBS: {js['obs']}")
        case "rapid_wind":
            mqtt.single('Tempest/wind',json.dumps(js['ob']), hostname=mqtt_host)
            print(f"Wind: {js['ob']}")

This code could easily be expanded to handle more that the obs_st and rapid_wind messages. For now it’s enough. Be sure to update the mqtt_host to point to your broker.

For sending the MQTT messages, the paho-mqtt library is used. The publish.single function can take additional parameters beyond those seen in the script:

• topic: (Required, string) The topic to publish the message to.
• payload: (Required, string, bytes, or None) The actual message content.
• qos: (Optional, int, default 0) The Quality of Service level (0, 1, or 2).
• retain: (Optional, bool, default False) If set to True, the broker will retain the message as the last known good value for the topic.
• hostname: (Optional, string, default 'localhost') The IP address or domain name of the MQTT broker.
• port: (Optional, int, default 1883) The network port of the broker.
• client_id: (Optional, string) A unique identifier for the client.
• auth: (Optional, dict) A dictionary containing username and password for authentication (e.g., {'username': "myuser", 'password': "mypass"}).
• tls: (Optional, dict or None) A dictionary for configuring TLS/SSL secure connections. 

A great intro to using MQTT in python shows installing paho-mqtt using pip3. For a Raspian Buster system apt install python3-paho-mqtt is simpler. Alternatively a virtual environment could be created.

In Openhab, a Generic MQTT thing pointing to your broker is ideal. For most of the sensor data your channels will have the MQTT Topic State set to Tempest/obs. An Incoming Value Transformation using JSONPath will parse the individual readings from the JSON array: JSONPATH:$[0][7], for example, will extract Air Temperature. Refer to the Weatherflow UDP API of the array index of each value. The channels and linked items can be created using the appropriate units for proper conversions.

For Wind Direction a Scale Profile can be created on the Item. This allows for displaying wind direction with compass directions in addition to degrees. For example, E (82 °). The degrees unit is set in the Channel. Create direction.scale in /etc/openhab/transform/ with this content:

[0..11]=N
[12..33]=NNE
[34..56]=NE
[57..79]=ENE
[80..102]=E
[103..125]=ESE
[126..147]=SE
[148..170]=SSE
[171..191]=S
[192..214]=SSW
[215..237]=SW
[238..260]=WSW
[261..283]=W
[284..306]=WNW
[307..329]=NW
[330..352]=NNW
[353..360]=N
NaN=Non-numeric state
format=%label% (%value%)

After updating TrueNAS let’s try to run Dropbox client in a Docker container rather than installing directly. Hopefully the container will survive upgrades better.

The janeczku/dropbox is one of the most downloaded images for Dropbox client in Docker. Unfortunately it hasn’t seen an update in nine years. It uses the 64-bit version 11.4.21 of dropboxd from 2016.

You must set env variable DBOX_SKIP_UPDATE to prevent loading the newer version of DB. The current version will not install in the container due to GLIB version issues.

There are several things to setup in the TrueNAS app GUI to get things going: As mentioned in the container overview you will want to set the two environment variables for user id and group ip: DBOX_UID and DBOX_GID. You also need to configure storage. The simplest storage is a Host Path mapped to /dbox/Dropbox and /dbox/.dropbox

Once it starts the log (View Logs ) will report the URL to follow to authorize the link. This is typically in the form https://www.dropbox.com/cli_link_nonce?nonce=5b59dd0928bc08b75a736935bd7d37cd. Pasting the link into a browser and confirming the login is required.

There are a few places to get some status info. From the TrueNAS shell, docker commands will help you see a bit. docker ps -a will list the current containers. docker logs -f containername will tail the container logs. docker top containername will list the current processes for the container.

From the container shell , the dropbox command can display additional information. dropbox status will give a brief status. dropbox filestatus will show the current sync status of the files in the current directory. Additional help on commands can be found using dropbox help or online.

Note: The container shell is accessed from the TrueNAS Apps Gui, select the container and look for Workloads and Containers. There should be three icons for each container – Shell , Volume Mounts and View Logs . In TrueNAS 25.04.2.6 there are bugs that prevent the View Logs function from working well in many cases

Even after all of this the janeczku/dropbox container didn’t seem to be actually syncing anything. It is unclear what all of the issues were but the extreme age of the daemon is surely part of the problem.

There are several newer containers available. One is tiagovdaa/dropbox-docker. It is definitely descended from janeczku/dropbox. Future testing will show if it is more suitable. It is built on version 223.4.4909. The current version from dropbox.com as of December 2025 is 238.4.6075. At the very least, 238.4.6075 does auto install correctly inside the tiagovdaa/dropbox-docker container. It should be noted the language for the container is Portuguese. It appears there is only one volume mount point for this container: dbox. This implies that .dropbox and Dropbox are both children of this mount point.

So for now, it’s back to the previous method of installing dropbox headless and using the Gist to run it as a service as discussed in an earlier post.

A few notes from the earlier post: The first command from the earlier post is su dropbox. This is critical. The .dropbox-dist/dropboxd command will create the Dropbox folder in the home directory of the current user. You will also get the prompt to copy/paste the displayed URL to authenticate to Dropbox as discussed above. When running as a service the /etc/db/dropbox-cli commands will fail UNLESS you are running as the user you defined in dropbox-start.target. The python script checks for the dropbox-pid file in ~/.dropbox.

For services, use systemctl status servicename or journalctl -u servicename for more info. Be aware for the dropbox-start service created above, the dropbox-start.service process completes and exits leaving behind a child running dropbox.

Sometimes there is a faster way. First let’s look at a simple append one table to another:

For In loop (slowest) 100%

local t1 = {string.byte(string.rep('A',256),1,256)}
local t2
for x = 1, 1000 do
    t2 = {1,2,3,4,5,6,7,8,9,0}
    for _, v in pairs(t1) do
      table.insert(t2, v)
    end
end

Regular For Loop (slightly faster) 84.3%

local t1 = {string.byte(string.rep('A',256),1,256)}
local t2
for x = 1, 1000 do
    t2 = {1,2,3,4,5,6,7,8,9,0}
    for v = 1,  #t1 do
      table.insert(t2, v)
    end
end
log(#t1..' : '..#t2)

table.move (much faster) 24.5%

local t1 = {string.byte(string.rep('A',256),1,256)}
local t2
for x = 1, 1000 do
    t2 = {1,2,3,4,5,6,7,8,9,0}
    table.move(t1,1,#t1,#t2+1,t2)
end
log(#t1..' : '..#t2)

Table.unpack (fastest) 12.5%

local t1 = {string.byte(string.rep('A',256),1,256)}
local t2
for x = 1, 1000 do
    t2 = {1,2,3,4,5,6,7,8,9,0, table.unpack(t1)}
end
log(#t1..' : '..#t2)

Test are based on ETC Mosaic v 2.14.4 soft triggers running scripts:

Results:

for in	1,213,051 microseconds
for	1,022,291 microseconds
move	  296,931 microseconds
unpack	  151,784 microseconds

The next test is extracting a zero terminated string from a buffer

buffer = {
1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,
1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,
1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,
1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,
1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,0,
1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,
1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,
1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,
1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9,1,2,3,4,5,6,7,8,9,9}

String Concatenate (slowest) 100%

local str = ''
local i
for x = 1, 1000 do
    i = 1; 
    str = ''
    local len = #buffer
    while buffer[i] ~= 0 and i<=len do
        str = str .. string.char(buffer[i])
        i = i + 1
    end
end

Scan and copy (faster) 10%

local str = ''
local i
for x = 1, 1000 do
    i = 1; 
    local len = #buffer
    while buffer[i] ~= 0 and i<=len do 
        i = i + 1
    end
    str = string.char(table.unpack(buffer,1,i))
end

table.unpack / string:unpack (fastest) 6.6%

local str
local i
for x = 1, 1000 do
    local sbuffer = string.char(table.unpack(buffer)) -- second copy
    str, i = string.unpack('z' , sbuffer)  -- final copy
end

Results:

string concatenate:          3,114,018 microseconds
scan and copy:                 308,604 microseconds
table.unpack / string.unpack:  207,526 microseconds

The final test is float32 extraction

Native Lua code (slowest) 100%

buffer = {0x43, 0x68, 0x61, 0x6E}
local float 
for x = 1, 10000 do
    local sign = 1
    local mantissa = (buffer[i+1] & 127)<<16 | buffer[i+2]<<8 | buffer[i+3]

    if buffer[i] > 127 then
        sign = -1
    end

    local exponent = (buffer[i] & 127)<< 1 | (buffer[i+1]>>7)
    if exponent == 0 then
        float = 0.0
    else
        mantissa = (math.ldexp(mantissa, -23) + 1) * sign       -- ldexp is deprecated in Lua 5.3 -> replace ldexp(x, exp) with (x * 2.0^exp) 
        float = math.ldexp(mantissa, exponent - 127)
    end
end

Float code based on info from https://www.h-schmidt.net/FloatConverter/IEEE754.html and https://github.com/iryont/lua-struct

string.unpack() (fastest) 91%

buffer = {0x43, 0x68, 0x61, 0x6E}
local float 
for x = 1, 10000 do
    local b = string.char(table.unpack(buffer,1,4))
    float = string.unpack('>f', b)
end

Results:

Native Lua Code: 120,978
string.unpack:   110,149

We started the journey with the Geekworm KVM-A3 Kit for Raspberry Pi 4 kit from Amazon.com. Add a Raspberry Pi 4, power supply and sd card to get started.

Both the PiKVM instruction and the Geekworm instructions are handy to have. You will need to download the PiKVM image (PiKVM V3 HAT BOX Image, OLED/FAN preactivated). We created the SD using RPi Imager.

For us the RTC and the OLED display were enabled out of the box. We did need to purchase a CR1220 3v Lithium button battery for the RTC.

You will want to get the clock set and syncing correctly. The steps below must be done as root with the filessystem set to rw. Remember to set the file system back to ro when done.

• set the timezone
  • timedatectl set-timezone America/New_York)
• temporarily disable systemd-timesyncd
  • systemctl stop systemd-timesyncd
• set the correct time
  • timedatectl set-time 'YYYY-MM-DD HH:MM:SS'
• update /etc/systemd/timesyncd.conf
  • nano /etc/systemd/timesyncd.conf
    • NTP="10.101.1.101"
• restart systemd-timesyncd
  • systemctl stop systemd-timesyncd

Setting the static address is described in the PiKVM FAQ. It involves editing /etc/systemd/network/eth0.network.

The device we wanted to control had Display Port out. We added a DisplayPort to HDMI & DisplayPort adapter to allow the on-site KVM to share the screen with the PiKVM. It should be noted that with the adapter we chose the primary monitor is the HDMI output. In our case that requires a bit of fiddling after a reboot to get the two displays to mirror and not extend.

We also found that for our remote access www solution the display needed to be set to MJPEG/HTTP instead of the default H.264/WebRTC mode.

After updating the Pi to Bookworm things didn’t seem to work correctly for the Pi SDR Weather Radio (pt 2, pt 3, pt 4). The problem was with the new version of ezstream. Turns out this is solvable from with a quick search. Newer versions of ezstream(1.x) no longer support the .xml format configuration format from version 0.x.

There is an included utility, ezstream-cfgmigrate, to solve the problem.

ezstream-cfgmigrate -0 ezstream.xml >ezstream.conf

will create a newly formatted file, ezstream.conf, to use.

I also realized I had never documented the current version of the scripts.

The first file is start.sh. This is typically going to be left running, typically with screen.

cd weatherRadio
rm -f one two three; mkfifo one two three

./to-same.sh <one &
./to-lame.sh <two &
./to-udp.sh <three &
rtl_fm -f 162550000 -s 22050 -p 14 | tee one two three |multimon-ng -t raw -a EAS /dev/stdin

Everything else happens in a subfolder. We first delete and creates three bash fifos, each processed by a separate script.

to-same.sh handles the actual SAME messages, MQTT and starting the speaker output:

multimon-ng -t raw -a EAS /dev/stdin | python3 ~/tg-dsame/dsame.py --mqtt jtk-openhab.home.arpa --json output.json --call /home/pi/weatherRadio/playRadio.sh --command boo

to-lame.sh (named before we switched to sox) handles the streaming via ezstream:

sox -t raw -r 22050 -b 16 -e signed -c 1 -v 7 - -r 22050 -t mp3 -c 1 -C 64 - sinc -3.5k |ezstream -c ezstream.conf

to-udp.sh creates the udp stream for playRadio.sh:

sox -t raw -r 22050 -b 16 -e signed -c 1 -v 7 - -r 22050 -t mp3 -c 1 -C 64 - sinc -3.5k|socat - udp4-sendto:localhost:5555

playRadio.sh pipes the udp stream from to-udp.sh to play using netcat and plays for two minutes to the attached speaker:

#!/bin/bash
nc -lu 5555|play -t mp3 -&
sleep 120
killall "play"

After getting TrueNAS setup I was time to create a FTP user for the cameras to record. First a local user is created. The user’s home directory is where the files go and will probably also wants to be a Dataset and configured as a Share. The user is configured with a shell of /usr/sbin/nologin.

The FTP service is then enabled. The service is configured to always chroot.

Now it is time to create a cron job to cleanup these folders. The following command is set to run each night at midnight as the camera ftp user:

find /mnt/tank1/video/// -type f -mtime +30 -print0 | xargs -0 rm -f

This should find and delete all of the files in the folder older than 30 days.

There has been an issue with some of the Amcrest camera and FTP but for now ours seem to be uploading correctly.

Unfortunately, TrueNAS Scale doesn’t natively support the Dropbox service. It does support Dropbox through Cloud Data Protection however that only provides for one way sync. The gist found at https://gist.github.com/kbumsik/b7cc243e297a3a66837151024049f43c provides an option. The gist will not survive a system update but can be re-enabled.

Make sure you create a regular user for dropbox and give them a home directory in your pool/home. For example, user dropbox with a home directory of /mnt/tank1/home/dropbox. The gist will have you the dropbox headless linux install directly from Dropbox along with the handy dropbox.py script from Dropbox before adding their service script. A summary of the commands from Dropbox and the gist:

su dropbox
wget https://www.dropbox.com/download?plat=lnx.x86_64
tar -xzf name_of_downloaded_file
.dropbox-dist/dropboxd
# this can take a long time especially on reinstallation

wget https://linux.dropbox.com/packages/dropbox.py
sudo mkdir /etc/db
sudo cp dropbox.py /etc/db/dropbox-cli
sudo chmod +x /etc/db/dropbox-cli

wget https://gist.githubusercontent.com/kbumsik/b7cc243e297a3a66837151024049f43c/raw/b339fbeee2ba1081723612bec5aacf92cc60e7c2/dropbox-start.target

# Edit the service unit for your needs - be sure user id is correct and adjust path for dropbox-cli
pico dropbox-start.target

# do not start the service unless you have stopped the manually started process
sudo cp dropbox-start.target /etc/systemd/system/dropbox-start.service
sudo systemctl enable --now dropbox-start.service
sudo systemctl status dropbox-start.service

I have saved dropbox.py and dropbox-start.target at truenas dropbox.zip in case the gist goes missing.

UPDATE: after a short while the process gave me an error that it couldn’t monitor the entire folder structure and suggest I run the following command to resolve:

echo fs.inotify.max_user_watches=100000 | sudo tee -a /etc/sysctl.conf; sudo sysctl -p

UPDATE #2: After updating to TrueNAS Scale Dragonfish-24.04.0 /usr is Read Only. The commands above have been updated. You will need to edit dropbox-start.target with the proper path from dropbox-cli.

There are some resources online for getting the Airthings Wave Plus to communicate with your Raspberry Pi and Openhab. The first article is from Airthings and is a bit out of date but a very good starting point. The find_wave.py script they refer to is no longer on their site but I found it at https://cdn2.hubspot.net/hubfs/4406702/Tech/find_wave.py or at https://github.com/Airthings/waveplus-reader. An newer Python Bluetooth library from the Airthings folks is also available.

This all takes us down a rabbit hole of python 2.7 lack of current support. “sudo apt-get install python-2” got the proper version of Python itself installed but that still leaves some dependencies missing. I found a note on installing the no longer supported version of pip for Python 2.7 for Raspbian Bullseye.

sudo curl https://bootstrap.pypa.io/pip/2.7/get-pip.py --output get-pip.py
sudo python2 get-pip.py

However, I was still not able to get bluepy installed for Python 2.7.

So now we try to adapt the script for Python 3. Not too bad, we need parenthesis for the print statements/functions. Once the syntax was correct my first run “sudo python3 find_wave.py” or “sudo blescan” rewarded me with the error message:

BTLEManagementError(“Failed to execute management command ‘%s'” % (cmd), rsp)

“bluetoothctl power off” followed by “bluetoothctl power on” seems to resolve the issue. There is quite a bit of discussion on this error on github.

Once we got find_wave.py working it was time to try read_waveplus.py. Once again this is written for Python 2.7 but a quick fixup of all of the print statements got it working for Python 3.

~/waveplus-reader $ sudo python3 read_waveplus.py 2930165120 5

Press ctrl+C to exit program

Device serial number: 2930165120
??????????????????????????????????????????????????????????????????????????????????????????????????????????
?     Humidity ? Radon ST avg ? Radon LT avg ?  Temperature ?     Pressure ?    CO2 level ?    VOC level ?
??????????????????????????????????????????????????????????????????????????????????????????????????????????
?     34.0 %rH ?     74 Bq/m3 ?     74 Bq/m3 ?   19.43 degC ?   977.64 hPa ?    546.0 ppm ?     46.0 ppb ?
?     34.0 %rH ?     74 Bq/m3 ?     74 Bq/m3 ?   19.43 degC ?   977.64 hPa ?    546.0 ppm ?     46.0 ppb ?
?     34.0 %rH ?     74 Bq/m3 ?     74 Bq/m3 ?   19.43 degC ?   977.64 hPa ?    546.0 ppm ?     46.0 ppb ?

There are lots of folks who have various version of Airthings to MQTT and airthings-wave-plus-reader. There is even a nifty ESP32 gateway project. But the project that interested me most for the Pi also creates a mini dashboard and was recommended in the Openhab forums.

Different readers all seem to return the same data for the sensors(Temp, Humidity, Pressure, CO₂, VOC and Radon). However, there are differences in the handling of access control data (illuminance, ambient light, measurement periods and battery level). The access control packets seems to start with the sent command followed by, in Python struct pack format, <L12B6H Or <L2BH2B9H. The battery level appear to be millivolts as a 16 bit unsigned word 3rd from the end(17 or 13).

For now the Openhab Bluetooth binding extension is working. If issues develop the MQTT gateway will probably be the solution.

UPDATE(July 2024): after updating the Pi to Bookworm bluepy was no longer installed. Currently it is not available as a Debian python3-xxx package and it considered an externally-managed-environment by pip3. This means sudo pip3 install bluepy will fail. There are a couple of choice to work around this. The recommended method is to create a virtual environment and install there. If I had the patience and time I would have chosen this path. Instead I chose the --break-system-packages route. This option will allow you to install into the global python environment with pip3.

sudo pip3 install bluepy --break-system-packages

In my situation this was expedient.

On my Raspberry Pi running Bullseye after running “sudo apt-get dist-upgrade” dhclient failed to start at boot. A local login and one could run “”sudo dhclient” and all would be well.

Fortunately the good folks at stackexchange had seen and solved the problem. Edit /etc/systemd/system/dhcpcd.service.d/wait.conf to change “/usr/lib/dhcpcd5/dhcpcd” to “/usr/sbin/dhcpcd”.

In the process of the upgrades discussed in the previous post – Ubuntu 18.04(Bionic Beaver) to 22.04(Jammy Jellyfish) there were additional issues for Webtrees and PHP.

The current release of webtrees (2.1) requires PHP 7.4, 8.0, 8.1 or 8.2. In the process of upgrading, PHP 8.1 was installed but lacked the required modules to function properly. This can be solved by a simple apt-get command to install the required pieces:

sudo apt-get install php8.1-{cli,common,curl,zip,gd,mysql,xml,mbstring,imagick,intl}

There is a discussion of this at https://www.webtrees.net/index.php/forum/help-for-2-0/35292-update-php-error-solved and https://www.webtrees.net/index.php/forum/help-for-2-0/35369-upgrade-ubuntu-to-20-04-02-broke-webtrees.

It is also worth noting the apache2 commands to enable/disable Apache PHP modules:

sudo a2dismod php7.4
sudo a2enmod php8.1

There is a discussion of installing multiple versions of PHP at https://tecadmin.net/how-to-install-php-on-ubuntu-22-04/ and https://linux.how2shout.com/how-to-install-php-7-4-on-ubuntu-22-04-lts-jammy-linux/. The articles discuss a PPA for the versions not in the standard apt sources. Once multiple versions are installed installed you can use

sudo update-alternatives --config php

to change versions. You can easily confirm the default version using “php -v”.

It was time to update the out of date server that is running the MythTV backend, so we went from 18.04(Bionic Beaver) to 20.04(Focal Fossa) to 22.04(Jammy Jellyfish). This is quite a jump. Most things survived the upgrades. According the Ubuntu we should have standard support until June 2027.

MySQL had an issue starting. The log at /var/log/mysql/error.log showed “unknown variable ‘query_cache_limit=1M'”. Turns out that query_cache_limit and query_cache_size are no longer valid in my.cnf for version 8.x. Removing these from the config file solved the problem. Read more at https://serverfault.com/questions/1042327/after-an-upgrade-to-ubuntu-20-04-lts-why-does-mysql-not-start.

There is the problem that after the upgrades the schema of the MythTV database was updated to 1376 instead of the 1348 on the clients. Off to do more upgrading…

It was also necessary to get a GUI view of the MythTV backend system. This machine hasn’t had a monitor connected in years, so the solution was VNC. The instructions at https://www.digitalocean.com/community/tutorials/how-to-install-and-configure-vnc-on-ubuntu-20-04 got me started. The short version is “sudo apt-get install tightvncserver”, minor adjustment to ~/.vnc/xstartup and “vncserver” along with TightVNC from the PC with the correct IP:port from the command line got it working.

There’s a bit more in the tutorial if you want the VNC server running as a service or need to mess with resolution… For example, “vncserver -geometry 1280×800” to set the screen dimensions or “vncserver -kill :1” to close the server.