Get Onboard with IoT Edge

Software Prerequisites

Azure ML configuration

Deploy Azure Machine Learning as an IoT Edge module

Create the Azure ML container

**** this par only works with Azure ML Workbench ******

az login

az provider register -n Microsoft.MachineLearningCompute
az provider register -n Microsoft.ContainerRegistry
az provider register -n Microsoft.ContainerService

az provider show -n Microsoft.MachineLearningCompute
az provider show -n Microsoft.MachineLearningCompute -o table

Local deployment

Set up a local environment (creates ACR)

az ml env setup -l [Azure Region, e.g. eastus2] -n [your environment name] [-g [existing resource group]]
az ml env setup -l westEurope -n mlgetonboard -g GetOnboard-rg

Azure Container Registry

The command above creates default ACR in provided resource group. All the containers will use this environment later and containers will be pushed to that registry

To change the resgitry you can use docker command

docker tag --help
example:
docker images
#see all aimges created with associated ACR

docker tag mlcrpacrdab009332683.azurecr.io/mlgob1:3 mlgob.azurecr.io/mlgob1:1
docker push mlgob.azurecr.io/mlgob1

azml env setup command created own registry with funny name . This also can probablyl be changed in aml_config dir in the Azure Workbech workiing directory if you started cmd prompt from Azure Workbench like I am, look at config and yml files inside this dir

Create the Azure ML container and pushes it into Azure Containter Registry

az ml service create realtime --model-file model.pkl -f iot_score.py -n mlgob -r python

check : on portal: go to ACR, it says running

**** this par only works with Azure ML Workbench ******

set up IoT Edge device

on portal: create iot edge service in iot hub cmd: install iotedgectl iotedgectl setup --connection-string "lkajsdfklj" --nopass iotedgectl start

on portal : add iotedge device

Add registry credentials to your Edge device

iotedgectl login --address --username --password

Cleanup installation

iotedgectl stop
iotedgectl uninstall

docker system prune -a -f

iotedgectl setup --connection-string "lkajsdfklj" --nopass

iotedgectl start

docker logs -f edgeAgent

Input Output Routing

Sample:

{
  "routes": {
    "sensorToMachineLearning": "FROM /messages/modules/tempSensor/outputs/temperatureOutput INTO BrokeredEndpoint(\"/modules/mlgob/inputs/amlInput\")",
    "machineLearningToIoTHub": "FROM /messages/modules/mlgob/outputs/amlOutput INTO $upstream"
  }
}

katyamust/ml_on_iot_edge