This project provides continuous monitoring of a DS18B20 temperature sensor using AWS (Amazon Web Services). All code is written in Python. The temperature sensor is attached to a Raspberry PI. It sendsalerts via SMS if a reading is below a given threshold.
Whenever it is run, the temperature reader script on the RPI will take the current reading of the sensor, store it locally and then also store it in a file in the S3 bucket created for this purpose. The upload to the S3 bucket triggers the execution of aws/process_temp_readings.py as an AWS lambda function. This function does all the heavy lifting, i.e. it consolidates the raw readings into json files by date and writes them to a separate directory in the AWS bucket. It also checks if the latest reading is not too old or if it is below the set threshold and sends alert messages to the configured SMS number.
The same AWS lambda function is also invoked from AWS on a schedule so that it can detect that the RPI has stopped sending updates and alert accordingly.
At this point the code only supports monitoring one temperature sensor.
I created this project as a way to monitor the temperature of a heating pipe in our house that can freeze up during the winter if we are not careful. It uses a Raspberry PI to take readings at regular intervals. The result is pushed into an AWS S3 bucket, which triggers an AWS lambda function that processes the data and sends out any alerts that are considered relevant. Most importantly the Lambda function will send alerts if the measured temperature drops below a configured threshold. However, it will also send alerts if no updates have been received for some time (default is 4 hours).
You will need at least a Raspberry PI and an AWS account plus a temperature sensor. My setup includes a regular PC with Linux (Ubuntu) installed and I will describe the installation process assuming such an environment. If you do not have a Linux computer, you can probably do everything on the PI though I have not tested that.
Unfortunately there are no ready-made temperature sensors for the Raspberry PI that can simply be plugged into the GPIO connector. Instead you will have to make your own by buying the sensor and wiring it up yourself. It is not too hard, I could figure it out, even though I barely know which end of a soldering iron gets hot. Instructions on how to connect a DS18B20 sensor to the PI are here https://www.cl.cam.ac.uk/projects/raspberrypi/tutorials/temperature/ (a bit old by now) or here https://pimylifeup.com/raspberry-pi-temperature-sensor/.
If you do not have an AWS account, you will have to create one. Open http://aws.amazon.com and click on the "Sign Up" button in the top right corner. Note that you will have to supply a credit card tocreate the AWS account. When used with one Raspberry PI, the AWS resource usage should stay within the free tier of AWS.
On the Raspberry PI, you need to install the following items:
- Python 3:
sudo apt-get install python3 python3-pip - pipenv:
sudo pip3 install pipenv
You should also create a dedicated user that will run the temperature reading code. I called it 'tsensor': sudo adduser tsensor.
You need to install the following items:
- Python 3: apt-get install python3 python3-pip
- Pipenv: sudo pip3 install pipenv
- jq: sudo apt-get install jq
- make: sudo apt-get install make
- check out github source (or pull release) FIXME
After unpacking the package, please change into the top level directory and run pipenv install to install the necessary Python packages.
If you have not already done so, please configure the AWS cli as described here: https://docs.aws.amazon.com/cli/latest/userguide/cli-chap-getting-started.html. Note that it is not recommended to use the AWS root account for actual work. The reasoning behind that are outlined here: https://docs.aws.amazon.com/IAM/latest/UserGuide/best-practices.html#lock-away-credentials.Instead create a separate Admin account and use its credentials for the AWS cli configuration.
Aside from the AWS cli configuration, you also need to provide an S3 bucket to house a transient file that is necessary for the AWS configuration. Specifically, we will deposit a zip file there with the backend event detection code. You can use an existing bucket in your account or create a new one.
Unpack the release package to a suitable directory. In a shell window, cd into the root directory and run 'pipenv install'. After that, cd into the aws subdirectory. Open the settings.sh in an editor and provide any missing values and review the rest to make sure they are ok for your purposes. The comments in settings.sh should be sufficient to explain the purpose of each. The PROFILE value will be passed as the aws --profile argument.
For reference I have included the settings.sh file that I am using for my setup: settings-ktsr42.sh. You will not be able to use it though.
Once you are happy with the settings.sh file, save it to disk. Now execute the update_stack.sh script from the aws subdirectory (your cwd should still be this dir). This will create all necessary AWS resources. If you change your mind about a setting in settings.sh, just change it and run update_stack.sh again.
The script should block and wait until the AWS configuraton is fully complete. It creates the configuration file for the temperature reader, which is why it has to be run first.
If you are stil in the aws subdirectory enter cd .. to switch back to the distribution root. Run make dist-reader from the root directory
This will create a temp_reader-<REL>.tar.gz archive file in the root directory. Copy this file to the Raspberry PI and unpack it into a releases subdirectory:
$ whoami
tsensor
$ mkdir releases
$ cd releases
$ tar xzvf ~/temp_reader-<REL>.tar.gz
Create a symbolic link to point to the latest release:
$ ln -snf temp_reader-<REL> prod
Make sure that the temperature sensor is attached to the RPI as described in the guideline you chose to follow and that the path to the temperature reading file under /proc comes back after a reboot.
Create a symbolic link in the home directory of tsensor to point to the actual reading file in /proc. For testing purposes you can point that link instead at releases/prod/sensor.dummy.
Install the required Python packages by running
$ cd ~/releases/prod
$ pipenv install
Create two directories required by the reader script:
$ cd
$ mkdir -p data logs
Execute the temperature reader once for testing purposes:
$ ./releases/prod/run.sh
to ensure the reader is working. It should not produce any output. You can verify that the reader has pushed a new file to the S3 bucket used for capturing by executing:
aws s3 ls gs://<your bucket from settings.sh>/observations
The reader writes its logfile to ~/logs/reader_log. Please review this file if anything seems to be off.
If everything is good, start the regular temperature monitoring by putting the script into cron:
crontab releases/prod/crontab
This will run the temperature reader every 20 minutes. Feel free to edit the crontab file (see man 5 crontab) if you would like to change the frequency. Note that above a certain threshold AWS starts to charge for traffic to S3.