Keeping time on a Pi

Posted on May 18, 2018 by Brian

Again, this blog fall into the category of a ‘I need to remember how I did this’ post. I actually started writing this months ago but never finished it. Good job as some of what I originally said was wrong and we have moved on from Jessie to Stretch.

This started when I accidentally discovered that the date/time on one of my Pi’s was significantly out. I won’t bore you with the things that I tried and failed, rather just tell you how to fix it.

You can check the date&time your own system by simply typing date and seeing if it is right or not.

Finding out how to get the right time set is remarkably difficult. In the end I discovered that you can simply use this command (reference);

sudo sntp -s time.google.com

NTP is the protocol that synchronises time across the internet. There are a number of free to use servers that you can query to get the current date & time.

One important fact is that if your date & time are significantly out, you cannot use ntp to set the time correctly. Set it manually so about the right time then sync.

On more recent versions/distributions of Linux, timedatectl may be used to query and change the system clock and its settings. Note this is not a daemon in it’s own right. On Linux there are 2 daemons you can use, ntpd or systemd-timesyncd . However you must not have both enabled at the same time. ntpd is the older daemon. systemd-timesyncd is the newer one and is a builtin service.

Now have a look at the daemon that controls synchronising your system with an ntp server.

$ timedatectl status
      Local time: Fri 2017-06-16 15:10:39 BST
  Universal time: Fri 2017-06-16 14:10:39 UTC
        RTC time: n/a
       Time zone: Europe/London (BST, +0100)
     NTP enabled: yes
NTP synchronized: yes
 RTC in local TZ: no
      DST active: yes
 Last DST change: DST began at
                  Sun 2017-03-26 00:59:59 GMT
                  Sun 2017-03-26 02:00:00 BST
 Next DST change: DST ends (the clock jumps one hour backwards) at
                  Sun 2017-10-29 01:59:59 BST
                  Sun 2017-10-29 01:00:00 GMT

$ timedatectl status

Local time: Fri 2017-06-16 15:10:39 BST

Universal time: Fri 2017-06-16 14:10:39 UTC

RTC time: n/a

Time zone: Europe/London (BST, +0100)

NTP enabled: yes

NTP synchronized: yes

RTC in local TZ: no

DST active: yes

Last DST change: DST began at

Sun 2017-03-26 00:59:59 GMT

Sun 2017-03-26 02:00:00 BST

Next DST change: DST ends (the clock jumps one hour backwards) at

Sun 2017-10-29 01:59:59 BST

Sun 2017-10-29 01:00:00 GMT

The best explanation of this daemon I found was here, with another guide here.

As you can see from the output above, NTP is enabled and synchronised and my timezone is set correctly, it is likely that in the response you see, they are not. Firstly, if your timezone is not right I’d suggest using sudo raspi-config to set it correctly.

In order to use ntp to keep your clock right, you first need to point the daemon to a time server. The recommendation is to use a suitable one out of the ntp pool project; http://www.pool.ntp.org/en/use.html . To do so you need to edit the file sudo nano /etc/systemd/timesyncd.conf so that your Time section looks something like this;

[Time]
Servers=0.uk.pool.ntp.org 1.uk.pool.ntp.org 2.uk.pool.ntp.org 3.uk.pool.ntp.org

1 2	[Time] Servers=0.uk.pool.ntp.org 1.uk.pool.ntp.org 2.uk.pool.ntp.org 3.uk.pool.ntp.org

Interestingly, the guide above suggested the line should start ‘NTP’ but as all I did to the standard line was uncomment it and change debian to uk.

To now switch on the synchronising use the command sudo timedatectl set-ntp true

One final thing, if you are using UFW you may find you need to open the port with sudo ufw allow 123/udp

You may also need to restart the services sudo systemctl restart systemd-timesyncd.service

Your system should now keep time 🙂

HTH

Posted in Technical - General | 1 Comment

Old DietPi Orange Pi Images

Posted on February 28, 2018 by Brian

If, like me, you are a fan of DietPi, you may well have been irritated by recent events. The decision to move to a completely new image that could not be upgraded to, was understandable, if unfortunate. These things happen to any OS with Windows being the obvious comparison.

Unfortunately this upgrade coincided with a spat with armbian which seemed to revolve around a school yard ‘my image is better than yours’ Twitter argument. The end result is that DietPi have dropped links with Armbian and so there are no new images for the OrangePi family which Fordee readily admits were some of the most popular downloads.

Fortunately, I had (unusually) quickly upgraded to the last Jessie version (v158) on my OrangePi Zero+ before it all went pete tong. Although no new images are being made available, it is not as if there is no further support in that you can still do the apt-get update/upgrade so it is only the added support DietPi brought that has been lost. DietPi did host some specific packages in their repositories so that will become an issue as time goes on.

DietPi are also being robust and deleting links to old images out there that are posted on the forum. One user has kindly provided a download of the images, the link for which I got before it was removed. If you want the last Jessie OrangePi image from DietPi you can get them here.

HTH

Posted in DietPi, OrangePi | 2 Comments

Adding a DietPi VM to an EXSi Host

Posted on January 30, 2018 by Brian

This blog very much falls into the ‘how in the hell did I do that’ category.

A bit of background. I have had an HP MicroServer for a number of years just sitting there not doing a lot except running an XP VM with a mailserver on it called Mailtraq. This was a hangover from when mail servers were pretty poor and GoogleMail was a twinkle in someone’s eye. It also helped with dial on demand (who remembers that!), NNTP and Demon Internet. Suffice to say life has moved on so the XP VM needed to be binned and so, sadly, did the Mailtraq installation.

What then to do with the MicroServer. It runs a free version of VMWare’s EXSi server (5.x) which I happily connect to via the old vSphere client. Whilst this is a little old hat, I know it and can find my way around it so ‘it ain’t broke, don’t fix’ is the mantra. I then realised that DietPi have some VM images so I thought I’d give them a go. Whilst running things on RaspberyPis seems like a good idea, for somethings it really isn’t. The network speed is poor (shared USB Bus), SD Card wear and a lack of grunt. The DietPi images are built for the VMware Workstation but I was able to convert them using the VMware vCenter standalone converter with little problem. The speed of the application was fast compared to a B+ Pi so all was well and good.

DietPi have just released a new platform. This has caused some angst as the older platform is now simply not supported. However, it is a step change for them, so I picked up the new V6 image and tried to convert it. No joy. I kept getting an error “Unable to Obtain Hardware information for the selected machine”. At first I thought it was an error with the new image file but on trying an image I had previously converted successfully, it also failed.

Following some searching, I am sure that the problem is with security on Windows 10. I do not know if it is MS or BitDefender. I had a brief foray with BitDefender but it interfered with Bash on Windows (no network) and they were unwilling to investigate without me doing all the work so it go canned. Getting rid of all the BitDefender garbage seems a bit tricky! I did try opening the converter as Administrator, disabling UAC and disabling MS Defender all to no avail.

So plan B was to upload the VMDK and VMX files. I was sure this had failed before but I had a go. First boot failed on OS type. This was fixed by selecting ‘otherLinux-64’.

Next boot did not like number of CPUs. Changed that in setting.

Next boot did not like the disk type. Error message “Failed to open disk scsi0:0: Unsupported or invalid disk type 7”. The issue here is that the disk type is for VMware Workstation. However, fixing this was relatively simple. I had SSH enabled on the EXSi host so SSH in and use vmkfstools -i <sourceVM> <targetVM> (KB item) to clone the disk to the right format.

The VM now booted and DietPi started. One final issue was the network did not want to configure. I fixed this by deleting the network adapter and installing a new one.

Having now got a DietPi VM that has run once, I will clone this and create the various VMs I need. http://www.mustbegeek.com/create-copy-of-existing-virtual-machine-in-esxi-server/

Posted in DietPi, Linux | Leave a comment

Displaying buttons on the TFT

Posted on June 5, 2017 by Brian

Having successfully hooked my small TFT up to the Raspberry Pi, I now wanted to use it! As it was a touch screen version what I wanted was buttons.

This is the end result;

TFT initial state

TFT display updated after a button press

The code is to be found in this gist:

import pygame
# Needs this to get the event types
from pygame.locals import *
import os
import logging
from time import sleep

log_format = '%(asctime)-6s: %(name)s - %(levelname)s - %(message)s'
console_handler = logging.StreamHandler()
console_handler.setFormatter(logging.Formatter(log_format))
logger = logging.getLogger()
logger.setLevel(logging.DEBUG)
logger.addHandler(console_handler)

#Colours
red = (255,0,0)
green = (0,255,0)
blue = (0,0,255)
black = (0,0,0)

display_width = 320
display_height = 240

# Assume all buttons of equal size (allow for a 5mm border all round to be added
button_dim = (0,0,120,80)
# define the centres for the buttons rather than top left
button_ctrs =[(85,65),(85,175),(235,65),(235,175)]
# I'm going to save the state of each button
button_state = [False , False, False, False]

# functions from https://pythonprogramming.net/displaying-text-pygame-screen/
# create a text object
def text_objects(text, font):
    textSurface = font.render(text, True, black)
    return textSurface, textSurface.get_rect()

# display text
def message_display(text,button):
    largeText = pygame.font.Font('freesansbold.ttf',75)
    TextSurf, TextRect = text_objects(text, largeText)
    TextRect.center = (button_ctrs[button])
    lcd.blit(TextSurf, TextRect)

# work out which quarter of the screen was pressed
def which_button ():
	pos = pygame.mouse.get_pos()
	#Find which quarter of the screen we're in
	x,y = pos
	button = 5
	if x < 160:
		if y < 120:
			button = 0
		else:
			button = 1
	else:
		if y < 120:
			button = 2
		else:
			button = 3

	return button

#########################
# Start doing things
#########################

# setup LCD screen
os.putenv('SDL_FBDEV', '/dev/fb1')
os.putenv('SDL_MOUSEDRV', 'TSLIB')
os.putenv('SDL_MOUSEDEV', '/dev/input/by-path/platform-20204000.spi-event')
#os.putenv('SDL_MOUSEDEV', '/dev/input/event0')

pygame.init()
pygame.mouse.set_visible(False)

# create a screen
lcd = pygame.display.set_mode((display_width, display_height))

# set up the initial screen
# Add a background
lcd.fill(red)

# build the array of rects with centers defined
buttons = []
for button in range (0,4):
	buttons.append(pygame.Rect(button_dim))
	buttons[button].center = button_ctrs[button]

#draw the buttons with a border
for button in buttons:
	# make the border by drawing a bigger rect first
	pygame.draw.rect(lcd, black, button.inflate(5,5), 0)
	pygame.draw.rect(lcd, blue, button, 0)
	message_display("off",buttons.index(button))

# update the whole screen as the borders are outside of the rects defined 
# so if we just updated those, the borders do not show.
pygame.display.flip()

# enclose loop in a try - catch to give a clean exit
try:
	print ("Press CTRL+C to exit")
	while True:
		# Scan touchscreen events
		for event in pygame.event.get():
			# Just use a button down event.
			if(event.type is MOUSEBUTTONDOWN):
				# which button was pressed
				but_pressed = which_button()
				# update the buttons state
				button_state[but_pressed] = not button_state[but_pressed]
				# update the display based on the state
				if button_state[but_pressed]:
					pygame.draw.rect(lcd, green, buttons[but_pressed], 0)
					message_display("on",but_pressed)
				else:
					pygame.draw.rect(lcd, blue, buttons[but_pressed], 0)
					message_display("off",but_pressed)
				# just update the pressed button rect	
				pygame.display.update(buttons[but_pressed])
				# put a sleep here to de-bounce
				sleep(0.2)
				# clear the queue of events that may have happened while we were sleeping
				pygame.event.clear()

except KeyboardInterrupt:
	# graceful exit (running interactively)
    print ("Goodbye.")

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

import pygame

# Needs this to get the event types

from pygame.locals import *

import os

import logging

from time import sleep

log_format = '%(asctime)-6s: %(name)s - %(levelname)s - %(message)s'

console_handler = logging.StreamHandler()

console_handler.setFormatter(logging.Formatter(log_format))

logger = logging.getLogger()

logger.setLevel(logging.DEBUG)

logger.addHandler(console_handler)

#Colours

red = (255,0,0)

green = (0,255,0)

blue = (0,0,255)

black = (0,0,0)

display_width = 320

display_height = 240

# Assume all buttons of equal size (allow for a 5mm border all round to be added

button_dim = (0,0,120,80)

# define the centres for the buttons rather than top left

button_ctrs =[(85,65),(85,175),(235,65),(235,175)]

# I'm going to save the state of each button

button_state = [False , False, False, False]

# functions from https://pythonprogramming.net/displaying-text-pygame-screen/

# create a text object

def text_objects(text, font):

textSurface = font.render(text, True, black)

return textSurface, textSurface.get_rect()

# display text

def message_display(text,button):

largeText = pygame.font.Font('freesansbold.ttf',75)

TextSurf, TextRect = text_objects(text, largeText)

TextRect.center = (button_ctrs[button])

lcd.blit(TextSurf, TextRect)

# work out which quarter of the screen was pressed

def which_button ():

pos = pygame.mouse.get_pos()

#Find which quarter of the screen we're in

x,y = pos

button = 5

if x < 160:

if y < 120:

button = 0

else:

button = 1

else:

if y < 120:

button = 2

else:

button = 3

return button

#########################

# Start doing things

#########################

# setup LCD screen

os.putenv('SDL_FBDEV', '/dev/fb1')

os.putenv('SDL_MOUSEDRV', 'TSLIB')

os.putenv('SDL_MOUSEDEV', '/dev/input/by-path/platform-20204000.spi-event')

#os.putenv('SDL_MOUSEDEV', '/dev/input/event0')

pygame.init()

pygame.mouse.set_visible(False)

# create a screen

lcd = pygame.display.set_mode((display_width, display_height))

# set up the initial screen

# Add a background

lcd.fill(red)

# build the array of rects with centers defined

buttons = []

for button in range (0,4):

buttons.append(pygame.Rect(button_dim))

buttons[button].center = button_ctrs[button]

#draw the buttons with a border

for button in buttons:

# make the border by drawing a bigger rect first

pygame.draw.rect(lcd, black, button.inflate(5,5), 0)

pygame.draw.rect(lcd, blue, button, 0)

message_display("off",buttons.index(button))

# update the whole screen as the borders are outside of the rects defined

# so if we just updated those, the borders do not show.

pygame.display.flip()

# enclose loop in a try - catch to give a clean exit

try:

print ("Press CTRL+C to exit")

while True:

# Scan touchscreen events

for event in pygame.event.get():

# Just use a button down event.

if(event.type is MOUSEBUTTONDOWN):

# which button was pressed

but_pressed = which_button()

# update the buttons state

button_state[but_pressed] = not button_state[but_pressed]

# update the display based on the state

if button_state[but_pressed]:

pygame.draw.rect(lcd, green, buttons[but_pressed], 0)

message_display("on",but_pressed)

else:

pygame.draw.rect(lcd, blue, buttons[but_pressed], 0)

message_display("off",but_pressed)

# just update the pressed button rect

pygame.display.update(buttons[but_pressed])

# put a sleep here to de-bounce

sleep(0.2)

# clear the queue of events that may have happened while we were sleeping

pygame.event.clear()

except KeyboardInterrupt:

# graceful exit (running interactively)

print ("Goodbye.")

A few comments on the false starts and things I learned (code is well commented I think):

I tried to install pygameUI but this failed because of the way I had implemented the TFT itself.
pygame is structured in a way I was unfamiliar with (fairly new to python). Key concepts are:
- The screen (the area you update)
- Rect – an object that defines a rectangular area of the screen. I created an array of Rect objects then gave each a centre. Once defined they will always display at that location unless you change it.
- You output various layers then update the display.
- Updating the display is best done on the rects that have changed although in this case, frame rate is not an issue.
- I used centres as I found that trying to create a border using a top left reference did not result in a constant thickness of the border all round.

I now need to find a way to mount the TFT permanently. I also really would like an enclosure – I’m going to hunt down a 3D one or pay a visit to my local hacklab.

I also need to expand my code to pull instructions off and send instructions to the main IOT server I have to control the heating.

HTH

Posted in Coding and Programming, Python | Tagged Python, RaspberryPi, TFT | Leave a comment

Forcing npm to Update Packages to Latest

Posted on May 31, 2017 by Brian

I have had a problem getting npm to update packages to the latest version. The packages involved are packages that have been installed directly and have no dependencies. It it is down to the use of semantic versioning in npm / node.js but knowing that does not help!

I have a package, canvas-gauges, installed via npm, that I wanted to update. Running

npm outdated

1	npm outdated

told me that;

Package             Current  Wanted  Latest  Location
canvas-gauges        2.1.3   2.1.3   2.1.4

1 2	Package Current Wanted Latest Location canvas-gauges 2.1.3 2.1.3 2.1.4

I wanted to update it to the latest version but despite repeated searches, on how to do this, I could not force npm to update the package.

Trying these did not work;

npm update canvas-gauges@latest
npm update canvas-gauges@2.1.4

1 2	npm update canvas-gauges@latest npm update canvas-gauges@2.1.4

I also tried installing and running ncu but that complains there is no package.json file.

There is nothing wrong in what npm is doing, it is following the ‘Wanted’ version even though I wanted the latest evrsion.

Ultimately the solution was simple (just not easily found);

npm install canvas-gauges@latest

1	npm install canvas-gauges@latest

HTH

Posted in Linux, Node-RED | Tagged npm | Leave a comment

Using a Small TFT with a Raspberry Pi

Posted on May 21, 2017 by Brian

Introduction

As I have said before part of the point of this blog is so I have a record of how I did things and this definitely fits into that category. If others find it useful, that is a bonus.

I have had a small 2.4″ TFT with touch sitting around for ages. I looked at using it when it arrived, but compling a kernel put me right off. So this week the notion took me to try again and I discovered it was actually very easy with newer versions of Raspbian.

[Edit 16 Dec 17] I had made an error in the table below putting both clocks on GPIO08. I have had some PCBs made up to simplify the connection – the error blew the TFT I was using sadly.

Sources

There are a couple of key sources I used and in places I have blatantly plagiarized them (they are that good) so as to bring it all together, and in case the source disappears (it happens).

Background

The background detail, found in the first reference, is interesting. It seems the old notro type method is being phased out in preference for overlays. Worth reading.

Wiring

I was fortunate in that my board was very similar to the board in reference 1 so I could use those mappings for my wiring. If you have a different board then head over there and it is explained how to work out the wiring for your specific overlay.

Panel Pin	Function name	GPIO	Pin No.
T_IRQ	Panel Touch Interrupt	GPIO25	22
T_DO	Touch Data Out	GPIO09	21
T_DIN	Touch Data In	GPIO10	19
T_CS	Touch Chip Select SPI_CE1_N	GPIO07	26
T_CLK	Touch Clock	GPIO11	23
SDO (MISO)	TFT Data Out SPI_MISO	GPIO09	21
LED	Backlight	GPIO18– see note	12
SCK	TFT Clock SPI_CLK	GPIO11	23
SDI (MOSI)	TFT Data In SPI_MOSI	GPIO10	19
D/C	TFT Data / Command	GPIO24	18
RESET	Reset	GPIO23	16
CS	TFT Chip Select SPI_CE0_N	GPIO08	24
GND	Ground	GND
VCC	3V3 supply	+3V3

I did the wiring using a breadboard as a couple of the pins need the same GPIO pin. Not pretty but functional!

Note the original sources comment on wiring up the LED Backlight. For now I have just connected to the 3.3V.

Setting up the OS

As an aside, although I did this initially on a clean install of Raspbian Jessie, I then moved to DietPi. The main reason for this is that you can create a configuration file for such things as WiFi SSID & Password and a fixed IP address. I could then do everything headless from the go.

The OS changes themselves are simplicity itself.

Edit the file /boot/config.txt (on DietPi you need to edit /DietPi/config.txt) and include the following lines;

# TFT display and touch panel
#dtoverlay=rpi-display,rotate=0,speed=10000000,xohms=100,debug=4
dtoverlay=rpi-display
dtparam=rotate=0
dtparam=speed=10000000
dtparam=xohms=100
dtparam=debug=4

# TFT display and touch panel

#dtoverlay=rpi-display,rotate=0,speed=10000000,xohms=100,debug=4

dtoverlay=rpi-display

dtparam=rotate=0

dtparam=speed=10000000

dtparam=xohms=100

dtparam=debug=4

And reboot the Pi. That’s it. If it is OK, the screen will light up during boot, then go blank.

Next step is to check and calibrate the touchscreen. I deviate a bit from my main source as there were 2 extra packages to install (references 2 & 3).

sudo apt-get install tsconf tslib libts-bin

1	sudo apt-get install tsconf tslib libts-bin

I found I needed to do the calibration before I did the test so (remembering to change the event number and fb if necessary);

sudo TSLIB_FBDEVICE=/dev/fb1 TSLIB_TSDEVICE=/dev/input/event0 ts_calibrate

1	sudo TSLIB_FBDEVICE=/dev/fb1 TSLIB_TSDEVICE=/dev/input/event0 ts_calibrate

and then

sudo TSLIB_FBDEVICE=/dev/fb1 TSLIB_TSDEVICE=/dev/input/event0 ts_test

1	sudo TSLIB_FBDEVICE=/dev/fb1 TSLIB_TSDEVICE=/dev/input/event0 ts_test

If all is well, you can drag the cross hairs or draw in the screen.

You now need to check the touch input. to do this;

sudo apt-get install xinput evtest

1	sudo apt-get install xinput evtest

and then run evtest If you are running headless, the touch input is likely to be event0, if you are not, again head over to the first reference and see how to interpret the contents of /dev/events/

The AdaFruit tutorial on this has some instructions for creating a udev rule to map the touch screen events to a fixed point. I have not managed to get this to work and I suspect it is because I am using overlays.

pygame and TSLIB problems with Jessie

I won’t bore you with how I got here, suffice to say I realised that the coordinates of the touch input were not correctly reported by the OS to pygame. After a bit of digging I found reference 6. In summary, to fix the issue create and run the following script;

#!/bin/bash

#enable wheezy package sources
echo "deb http://archive.raspbian.org/raspbian wheezy main
" > /etc/apt/sources.list.d/wheezy.list

#set stable as default package source (currently jessie)
echo "APT::Default-release \"stable\";
" > /etc/apt/apt.conf.d/10defaultRelease

#set the priority for libsdl from wheezy higher then the jessie package
echo "Package: libsdl1.2debian
Pin: release n=jessie
Pin-Priority: -10
Package: libsdl1.2debian
Pin: release n=wheezy
Pin-Priority: 900
" > /etc/apt/preferences.d/libsdl

#install
apt-get update
apt-get -y --force-yes install libsdl1.2debian/wheezy

#!/bin/bash

#enable wheezy package sources

echo "deb http://archive.raspbian.org/raspbian wheezy main

" > /etc/apt/sources.list.d/wheezy.list

#set stable as default package source (currently jessie)

echo "APT::Default-release \"stable\";

" > /etc/apt/apt.conf.d/10defaultRelease

#set the priority for libsdl from wheezy higher then the jessie package

echo "Package: libsdl1.2debian

Pin: release n=jessie

Pin-Priority: -10

Package: libsdl1.2debian

Pin: release n=wheezy

Pin-Priority: 900

" > /etc/apt/preferences.d/libsdl

#install

apt-get update

apt-get -y --force-yes install libsdl1.2debian/wheezy

This fixed the coordinates for me.

Python and TFT

The aim of all this is to have a small touchscreen that I can use to provide simple control of my heating and hot water – basically a boost function. I’ll blog about that when I get a bit further. In the meantime, the following code will test out the touch responses.

#!/usr/bin/python                                                                                               
import os
import locale
import logging as log
import pygame
from time import *

# Tell the RPi to use the TFT screen and that it's a touchscreen device
os.putenv('SDL_VIDEODRIVER', 'fbcon')
os.putenv('SDL_FBDEV'      , '/dev/fb1')
os.putenv('SDL_MOUSEDRV'   , 'TSLIB')
os.putenv('SDL_MOUSEDEV'   , '/dev/input/event0')

class test():
	def __init__(self):
		pygame.init()
		self.screen = pygame.display.set_mode((320, 240))

	def main_loop(self):
		while True:
			event = pygame.event.wait()
			if event.type == pygame.MOUSEBUTTONDOWN:
				log.debug("MOUSEBUTTONDOWN:{}".format(pygame.mouse.get_pos()))
				pygame.event.clear()

if __name__ == "__main__":
	log.basicConfig(filename="/tmp/log",level=log.DEBUG)
	main_window = test()
	main_window.main_loop()

#!/usr/bin/python

import os

import locale

import logging as log

import pygame

from time import *

# Tell the RPi to use the TFT screen and that it's a touchscreen device

os.putenv('SDL_VIDEODRIVER', 'fbcon')

os.putenv('SDL_FBDEV' , '/dev/fb1')

os.putenv('SDL_MOUSEDRV' , 'TSLIB')

os.putenv('SDL_MOUSEDEV' , '/dev/input/event0')

class test():

def __init__(self):

pygame.init()

self.screen = pygame.display.set_mode((320, 240))

def main_loop(self):

while True:

event = pygame.event.wait()

if event.type == pygame.MOUSEBUTTONDOWN:

log.debug("MOUSEBUTTONDOWN:{}".format(pygame.mouse.get_pos()))

pygame.event.clear()

if __name__ == "__main__":

log.basicConfig(filename="/tmp/log",level=log.DEBUG)

main_window = test()

main_window.main_loop()

Edit 5 June 2017

I have now sorted out some more useful code to display buttons which you will find here.

HTH

Posted in Hardware, Linux, RaspberryPi | Tagged pygame, Python, TFT | Leave a comment

Bash History

Posted on April 3, 2017 by Brian

On several occasions, I have lost some of my bash history, as the ssh session has not been exited cleanly. This annoyed me as I use the history as a means to remember what I have done. It prompted me to investigate if I could fix it and Unix being Unix, of course you can.

The History part of my .bashrc file now looks like this;

# don't put duplicate lines or lines starting with space in the history.
# See bash(1) for more options
HISTCONTROL=ignoreboth

# append to the history file, don't overwrite it
shopt -s histappend

# BPO force all commands to be saved on one line
shopt -s cmdhist

# BPO append the command to the history after every command
# prevents lost history when exit is not used
PROMPT_COMMAND='history -a'

# for setting history length see HISTSIZE and HISTFILESIZE in bash(1)
HISTSIZE=1000000
HISTFILESIZE=2000000

# BPO Save a time stamp
# HISTTIMEFORMAT='%F %T '

# BPO Don't save common commands
HISTIGNORE='ls:bg:fg:history'

# don't put duplicate lines or lines starting with space in the history.

# See bash(1) for more options

HISTCONTROL=ignoreboth

# append to the history file, don't overwrite it

shopt -s histappend

# BPO force all commands to be saved on one line

shopt -s cmdhist

# BPO append the command to the history after every command

# prevents lost history when exit is not used

PROMPT_COMMAND='history -a'

# for setting history length see HISTSIZE and HISTFILESIZE in bash(1)

HISTSIZE=1000000

HISTFILESIZE=2000000

# BPO Save a time stamp

# HISTTIMEFORMAT='%F %T '

# BPO Don't save common commands

HISTIGNORE='ls:bg:fg:history'

I tried adding a timestamp but that just looked messy so I took it out again.

HTH

Posted in Linux | Tagged bash, history | Leave a comment

Installing pywws and running it as a Daemon

Posted on March 28, 2017 by Brian

I have been using pywws for a number of years to download the data from my weather station and it simply just works. There is good support from the author though there are some gaps in the documentation. Moving pywws to the new Pi gave me the opportunity to update the package which had not been done for a while (ain’t broke don’t fix).

The weather station is a WH1080 compatible system from Maplin. It is often on offer for about £60 and you can buy spares for all the external parts including the transmitter at a very reasonable price which is a major bonus.

Installing pywws via pip is relatively painless but what I would urge, is that you check out the dependencies first and get them installed before the main package. It is easy to miss or skip over this step. There are 2 options for USB libraries and I plumped for the libsub pip install libusb1 .

One of the beauties of this weather station and pywws combination is that the weather station itself stores data so you can disconnect the USB from the base station to play and not lose any data as it just catches up once reconnected.

Once installed, I plugged in the weather station to the new pi to check that it was communicating – I had to do the changes suggested as it would only work as sudo otherwise. Doing this test also created some of the system files.

I then pulled across all the data files using rsync plus a couple of system files I had modified. These modifications allowed pywws to communicate to EmonCMS and to put data out on mqtt. The way this is setup is a little convoluted (to my mind), not that well documented, and it had taken me a while previously to work it out. To do it you need to create a new ‘service’.

New pywws service

To create a new pywws service you need to create 2 files in the service folder(/usr/local/lib/python2.7/dist-packages/pywws/services), a service file and a template. The template is used to create the format of the message you wish to send and the service file defines some of the parameters that the template may use. By defining the parameters correctly, you can also change these in the main configuration file weather.ini in the data folder.

I have created a GitHub repository for the new service files for EmonCMS. I am using the http API for pushing the data to EmonCMS. One of the problems I had in setting this up is that EmonCMS does not accept valid JSON so it took a bit of time to get the template right.

Running pywws live-logger as a Daemon

Pywws has a daemon version of the main logging script using python-daemon. I have never felt completely comfortable with this as it seems to me that by following the instructions, you can end up with a daemon starting a daemon. The original install was Wheezy based so I had created a file for init.d to use (available from the GitHub repository) but on the newer Jessie based system, I thought I’d try a systemd setup.

After reading this excellent guide on systemd, I generated the following service file and placed it in the systemd folder (/etc/systemd/system). After a couple of minor teething problems, I got it running and it has just worked since. This seems a better solution.

[Unit]
Description=PYWWS-Livelog daemon
After=syslog.target
After=network.target
Wants=systemd-networkd-wait-online.service
Wants=mosquitto.service

[Service]
Type=simple
User=pi
#Group=<run as group group>
ExecStart=/usr/bin/python /usr/local/bin/pywws-livelog -v -l /home/pi/weather/data/live-logger-new /home/pi/weather/data

Restart=always
RestartSec=90
StartLimitInterval=400
StartLimitBurst=3

# Give the script some time to startup
TimeoutSec=300

[Install]
WantedBy=multi-user.target

[Unit]

Description=PYWWS-Livelog daemon

After=syslog.target

After=network.target

Wants=systemd-networkd-wait-online.service

Wants=mosquitto.service

[Service]

Type=simple

User=pi

#Group=<run as group group>

ExecStart=/usr/bin/python /usr/local/bin/pywws-livelog -v -l /home/pi/weather/data/live-logger-new /home/pi/weather/data

Restart=always

RestartSec=90

StartLimitInterval=400

StartLimitBurst=3

# Give the script some time to startup

TimeoutSec=300

[Install]

WantedBy=multi-user.target

Monitoring pywws

On the old Wheezy system, I used Monit to monitor both the daemon and a specific folder. For a while, pywws had a habit of just stopping the writing of data. The daemon was still running just no data appeared. I therefore set Monit up to monitor a specific folder to check it was modifed regularly. As I have installed Monit I brought this across (also on the GitHub repository).

However, one of the advantages of systemd is that it has built-in monitor and restart functionality so Monit cannot be used to monitor the daemon itself.

Final Steps

Once I was happy that the system was working fine, I plugged the weather station back into the original Pi, restarted it and left it running for a while to catch up. I then stopped the daemon, used rsync to copy across the files again (I blatted that data on the new Pi first just to be sure) and plugged the station into the new Pi. A quick reboot and it all seems very happy.

All that remained was to set up some web access but that is another blog.

Posted in Technical - General | Leave a comment

Node-RED and icons

Posted on March 28, 2017 by Brian

I’ve been looking at Pete Scargill’s thermostat and got stuck on icons. He blogged about icons recently but that didn’t say how to install them. So on the basis of ‘if I didn’t know there will be others who don’t know’, here’s how to install some icons in Node-RED.

Pete’s suggestion is to create a static folder under the node-red folder and put the icons in there such as home/pi/.node-red/public/myicons .

To make these available you need to edit the file /home/pi/.node-red/settings.js and add the line

httpStatic: '/home/pi/.node-red/public',

1	httpStatic: '/home/pi/.node-red/public',

note the closing comma! (there is an example in the default file – just search for it).

I’ll quote a comment Pete made on this (explains it better than I could);

“So then the web pages think that the root page for all assets is the above – hence your reference in the web starts at /myicons.

The reason I use myimages and myicons etc is that Node-Red already has some directories it uses for UI and images is one of them – if you make an “images” directory it conflicts – so I just prefix any of my stuff with “my” – so there you go.

That solves where to put them, but what to put there. Well in his blog he directs you to this open icon set on SourceForge https://sourceforge.net/projects/openiconlibrary/.

I downloaded the file from SourceForge and copied it across to the Pi using WinSCP into a folder in my home directory I called ‘icons’. It is a tar.bz2 file but that is easily unpacked in one go with:

tar xvfj somefilename.tar.bz2

1	tar xvfj somefilename.tar.bz2

I prefer to keep things like this icon pack separate and link to them so if you want to change the icon set, you just change the link. To do that move into the public folder you have created cd /home/pi/.node-red/public and run the following command to create a symbolic link (adjusting paths as necessary) to your icon set;

ln -s /home/pi/icons/open_icon_library-standard/icons/ myicons

1	ln -s /home/pi/icons/open_icon_library-standard/icons/ myicons

If you do ls -la you will see the link. Restart Node-RED

sudo systemctl restart nodered

1	sudo systemctl restart nodered

and wait until it has properly started. Refresh the dashboard page and you should have icons!

HTH

Posted in Node-RED, RaspberryPi | Tagged icon set, icons | 3 Comments

Node-RED, npm, sudo and permissions

Posted on March 27, 2017 by Brian

I ran into a problem on my Pi yesterday; I could not update any npm installed packages. If I did try, it simply failed, quietly. Some searching led me to believe that it was because I had installed npm using sudo and possibly other packages as well (I’d lost some Bash history so I cannot be sure – a blog for another time).

I dug around a little and found this comment on StackOverflow;

“Recent versions of npm switch to user “nobody” as a security measure if it’s being run with sudo privileges.”

I could see exactly this within /usr/lib/node_modules.

On the npm site itself, there is an item on this issue which suggests some modifications which I really did not look the like of, as npm config get prefix returned just /usr.

Another item I found suggested changing permissions in a similar manner to the npm docs but I felt it was rather over the top and changed too much. So I took a middle road and did the following;


sudo chown -R $(whoami) /usr/lib/node_modules

1 2	sudo chown -R $(whoami) /usr/lib/node_modules

This changed all the ‘nobody’ owned files and folders to owned by pi (which I was logged in as).

What I cannot remember (lost history again) is if I did it for the folders in my home folder. If I had found any files owned by ‘nobody’ I would have done so.

This has solved the problem and now I can use npm without sudo and update (both locally and globally) work fine.

HTH

Posted in Linux, Node-RED | Tagged permissions, sudo | Leave a comment

The (Tech) Life of (a) Brian

Keeping time on a Pi

Old DietPi Orange Pi Images

Adding a DietPi VM to an EXSi Host

Displaying buttons on the TFT

Forcing npm to Update Packages to Latest

Using a Small TFT with a Raspberry Pi

Introduction

Sources

Background

Wiring

Setting up the OS

pygame and TSLIB problems with Jessie

Python and TFT

Bash History

Installing pywws and running it as a Daemon

New pywws service

Running pywws live-logger as a Daemon

Monitoring pywws

Final Steps

Node-RED and icons

Node-RED, npm, sudo and permissions

About This Site

Meta

Recent Posts

Recent Comments

Archives

Categories

Be Sociable

Introduction

Sources

Background

Wiring

Setting up the OS

pygame and TSLIB problems with Jessie

Python and TFT

New pywws service

Running pywws live-logger as a Daemon

Monitoring pywws

Final Steps

About This Site

Meta

Recent Posts

Recent Comments

Archives

Categories

Tags