azure-iot-sdk-python/sdklab/meantimerecovery

Mean time to recover first phase

Current situation

Currently the only way to run aedes without changing code in the SDK. This is done so that we can pass server authentication with a self signed server verification certificate. In order to do so, a certificate needs to be generated whose subject is equivalent to the hostname of the machine where the aedes server is being run.

Same HOST machine

Run the aedes server (with or without container) in the same host. For connecting to this server the connection string for device client would have HostName=localhost.

• Ensure that the aedes server is loaded with certificates whose common name is localhost.
• Ensure that this certificate is passed as the kwarg server_verification_cert during device client creation.

Different HOST Machine

Run the aedes server (with or without container) in a different host. For example lets say the hostname of the machine or the container is alohomora.net. For connecting to this server the connection string for device client would have HostName=alohomora.net.

• Ensure that the aedes server is loaded with certificates whose common name is alohomora.net.
• Ensure that this certificate is passed as the kwarg server_verification_cert during device client creation.

Script for Mean Time to Recover

The main script for tracking mean time to recover is mean_time_recover_with_docker.py. For this part the python package called docker needs to be installed. pip install docker will do this.

This package is needed so that the python script can access the Docker Engine API. It enables to do anything the docker command does, but from within python apps.

There are some variables that can be changed :-

FACTOR_OF_KEEP_ALIVE : the multiplication factor via which keep alive needs to be modified to calculate the amount of time the MQTT Broker will be down.

Other important variables already having values are :-

KEEP_ALIVE : option for changing the default keep alive for MQTT broker. Currently set at 15 secs. KEEP_RUNNING : the amount of time the server needs to keep running for. dead_duration : the amount of time the MQTT broker will be taken down for. KEEP_DEAD : The amount of time the MQTT broker will be non responsive. MQTT_BROKER_RESTART_COUNT : the count of times server will be stopped and started.

Before running the the mean_time_recover_with_docker.py script make sure your docker engine is running.

Certificate creation

There is another script which will create some self signed certificates for use in the aedes server. For this part we would need a package called cryptography. Prior to running this script please do pip install cryptography in the corresponding python environment . To generate a certificate with a certain common name (possibly localhost) the script needs to be run like python create_self.py "localhost". There are no certificates included in the folder for aedes, so prior to running aedes server, certificates need to be generated.

Debug or Test

If the aedes server or the docker container is run separately then for testing or debugging purposes a simpler script called simple_send_message.py.

MQTT Broker

For this part docker is needed to be installed in your machine. The mqtt broker being used is in the folder aedes. This has been configured to run locally in a docker container. There is no need to run the container as the python script will do so but it is needed to build the image so that the mean_time_recover_with_docker.py can create a container later.

The docker commands to build the appropriate image is : docker build -t mqtt-broker .

The docker command to remove the above image is docker image rm mqtt-broker

Running aedes in a container

For aedes server these are required

• Navigate to aedes folder and run the above image building commands which will build the image with the help of the Dockerfile in that folder.
• Ensure that the package.json file similar to the one inside aedes folder exists.

Running docker container separately

If aedes is run not as the part of the mean time script then the following steps need to be performed prior to that

• node js must be installed locally
• npm install must be done with the package.json in the current folder.
• appropriate certificates must be generated

For running the containerized aedes separately please use this command. docker run -d --publish 8883:8883 --name expecto-patronum mqtt-broker