Deploy with Helm#

This guide provides step-by-step instructions for deploying the Industrial Edge Insights - Time Series sample application using Helm.

Prerequisites#

System Requirements
K8s installation on single or multi node must be done as prerequisite to continue the following deployment. Note: The Kubernetes cluster is set up with kubeadm, kubectl and kubelet packages on single and multi nodes with v1.30.2. Refer to tutorials such as https://adamtheautomator.com/installing-kubernetes-on-ubuntu and many other online tutorials to setup kubernetes cluster on the web with host OS as Ubuntu 22.04.
For Helm installation, refer to helm website

Note If Ubuntu Desktop is not installed on the target system, follow the instructions from Ubuntu to install Ubuntu desktop. The target system refers to the system where you are installing the application.

Step 1: Generate or download the Helm charts#

Choose one of the following approaches to get the Helm charts:

Wind Turbine Anomaly Detection

Option A: Download the Helm charts

Download Helm chart:

helm pull oci://registry-1.docker.io/intel/wind-turbine-anomaly-detection-sample-app --version 1.1.0-rc2

Extract the Helm chart:

tar -xvzf wind-turbine-anomaly-detection-sample-app-1.1.0-rc2.tgz
cd wind-turbine-anomaly-detection-sample-app

Proceed to Step 2 to configure your values.yaml file present in the current directory.

Option B: Generate Helm charts

Navigate to the source directory:

cd edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series  # path relative to git clone folder

Generate the charts:

make gen_helm_charts app=wind-turbine-anomaly-detection version=1.1.0-rc2
cd helm/

Proceed to Step 2 to configure your values.yaml file present in the current directory.

Weld Anomaly Detection

Option A: Download the Helm charts

Download Helm chart:

helm pull oci://registry-1.docker.io/intel/weld-anomaly-detection-sample-app --version 1.0.0-rc2

Extract the Helm chart:

tar -xvzf weld-anomaly-detection-sample-app-1.0.0-rc2.tgz
cd weld-anomaly-detection-sample-app

Proceed to Step 2 to configure your values.yaml file present in the current directory.

Option B: Generate Helm charts

Navigate to the source directory:

cd edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series  # path relative to git clone folder

Generate the charts:

make gen_helm_charts app=weld-anomaly-detection version=1.0.0-rc2
cd helm/

Proceed to Step 2 to configure your values.yaml file present in the current directory.

Step 2: Configure and update the environment variables#

Update the following fields in values.yaml file of the helm chart

INFLUXDB_USERNAME:
INFLUXDB_PASSWORD:
VISUALIZER_GRAFANA_USER:
VISUALIZER_GRAFANA_PASSWORD:
HTTP_PROXY: # example: http_proxy: http://proxy.example.com:891
HTTPS_PROXY: # example: http_proxy: http://proxy.example.com:891

Step 3: Install Helm charts#

Note:

Uninstall the Helm charts if already installed.

Note the helm install command fails if the above required fields are not populated as per the rules called out in values.yaml file.

Wind Turbine Anomaly Detection

To install Helm charts, use one of the following options:

OPC-UA ingestion flow:

helm install ts-wind-turbine-anomaly --set env.TELEGRAF_INPUT_PLUGIN=opcua . -n ts-sample-app --create-namespace

MQTT ingestion flow:

helm install ts-wind-turbine-anomaly --set env.TELEGRAF_INPUT_PLUGIN=mqtt_consumer . -n ts-sample-app --create-namespace

Note: To deploy with GPU support for inferencing, use the following command:
helm install ts-wind-turbine-anomaly \
  --set privileged_access_required=true \
  --set env.TELEGRAF_INPUT_PLUGIN=<input_plugin> \
  . -n ts-sample-app --create-namespace
The privileged_access_required=true setting enables Time Series Analytics Microservice access to GPU device through /dev/dri.

Weld Anomaly Detection

To install Helm charts, run the following command:

helm install ts-weld-anomaly . -n ts-sample-app --create-namespace

Step 4: Copy the udf package for helm deployment to Time Series Analytics Microservice#

Wind Turbine Anomaly Detection

To copy your own or existing model into Time Series Analytics Microservice in order to run this sample application in Kubernetes environment:

The following udf package is placed in the repository under edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series/apps/wind-turbine-anomaly-detection/time-series-analytics-config.

- time-series-analytics-config/
    - models/
        - windturbine_anomaly_detector.pkl
    - tick_scripts/
        - windturbine_anomaly_detector.tick
    - udfs/
        - requirements.txt
        - windturbine_anomaly_detector.py

Copy your new UDF package (using the wind turbine anomaly detection UDF package as an example) to the time-series-analytics-microservice pod:

export SAMPLE_APP="wind-turbine-anomaly-detection"
cd edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series/apps/wind-turbine-anomaly-detection/time-series-analytics-config # path relative to git clone folder
mkdir -p $SAMPLE_APP
cp -r models tick_scripts udfs $SAMPLE_APP/.

POD_NAME=$(kubectl get pods -n ts-sample-app -o jsonpath='{.items[*].metadata.name}' | tr ' ' '\n' | grep deployment-time-series-analytics-microservice | head -n 1)

kubectl cp $SAMPLE_APP $POD_NAME:/tmp/ -n ts-sample-app

Weld Anomaly Detection

To copy your own or existing model into Time Series Analytics Microservice in order to run this sample application in Kubernetes environment:

The following udf package is placed in the repository under edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series/apps/weld-anomaly-detection/time-series-analytics-config.

- time-series-analytics-config/
    - models/
        - weld_anomaly_detector.cb
    - tick_scripts/
        - weld_anomaly_detector.tick
    - udfs/
        - requirements.txt
        - weld_anomaly_detector.py

Copy your new UDF package (using the weld anomaly detection UDF package as an example) to the time-series-analytics-microservice pod:

export SAMPLE_APP="weld-anomaly-detection"
cd edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series/apps/weld-anomaly-detection/time-series-analytics-config # path relative to git clone folder
mkdir -p $SAMPLE_APP
cp -r models tick_scripts udfs $SAMPLE_APP/.

POD_NAME=$(kubectl get pods -n ts-sample-app -o jsonpath='{.items[*].metadata.name}' | tr ' ' '\n' | grep deployment-time-series-analytics-microservice | head -n 1)

kubectl cp $SAMPLE_APP $POD_NAME:/tmp/ -n ts-sample-app

Note: Run the commands only after performing the Helm install.

Step 5: Activate the New UDF Deployment Package#

NOTE: To activate the UDF inferencing on GPU, additionally run the following command as a prerequisite before activating the UDF deployment package:
curl -k -X 'POST' \
'https://<HOST_IP>:30001/ts-api/config' \
-H 'accept: application/json' \
-H 'Content-Type: application/json' \
-d '<Add contents of edge-ai-suites/manufacturing-ai-suite/industrial-edge-insights-time-series/apps/wind-turbine-anomaly-detection/time-series-analytics-config/config.json with device
    value updated to gpu from cpu>'
GPU Inferencing is supported only for Wind Turbine Anomaly Detection sample app

Run the following command to activate the UDF deployment package:

curl -k -X 'GET' \
  'https://<HOST_IP>:30001/ts-api/config?restart=true' \
  -H 'accept: application/json'

Step 6: Verify the Results#

Follow the verification steps in the Get Started guide.

Uninstall Helm Charts#

Wind Turbine Anomaly Detection

helm uninstall ts-wind-turbine-anomaly -n ts-sample-app
kubectl get all -n ts-sample-app # It may take a few minutes for all application resources to be cleaned up.

Weld Anomaly Detection

helm uninstall ts-weld-anomaly -n ts-sample-app
kubectl get all -n ts-sample-app # It may take a few minutes for all application resources to be cleaned up.

Configure Alerts in Time Series Analytics Microservice#

To configure alerts in Time Series Analytics Microservice, follow the steps in Time Series Analytics Helm Deployment.

Deploy the Application with a Custom UDF#

To deploy the application with a custom UDF, follow the steps in Custom UDF Helm Deployment.

Troubleshooting#

Check pod details or container logs to diagnose failures:

kubectl get pods -n ts-sample-app
kubectl describe pod <pod_name> -n ts-sample-app # Shows details of the pod
kubectl logs -f <pod_name> -n ts-sample-app # Shows logs of the container in the pod

Deploy with Helm#

Prerequisites#

Step 1: Generate or download the Helm charts#

Step 2: Configure and update the environment variables#

Step 3: Install Helm charts#

Step 4: Copy the udf package for helm deployment to Time Series Analytics Microservice#

Step 5: Activate the New UDF Deployment Package#

Step 6: Verify the Results#

Uninstall Helm Charts#

Configure Alerts in Time Series Analytics Microservice#

Deploy the Application with a Custom UDF#

Troubleshooting#

This Page