Configuring Azure Monitor

Add Telemetry Endpoint

Note

If you would like to use an existing Application Insights and Log Analytics Workspace, navigate to Overview, then JSON View to capture the connection string.

Azure Monitor Setup

1. Create a new Application Insights Instance.

Log into Microsoft Azure. Click Monitor, then Application Insights.

2. Under Application Insights, click View, and then +Create.

Choose the appropriate values for Subscription, Resource Group, and Log Analytics Workspace as defined in your tenant.

3. Click Review + create.

4. Navigate to Overview, then JSON View and then capture the ConnectionString Value

NVCF Create Telemetry Endpoint

Additional documentation for creating a Telemetry Endpoint can be found here.

1. Using NGC, navigate to Cloud Functions, and then click Settings.

2. Under Telemetry Endpoints, click + Add Endpoint.

3. Provide an appropriate Name under Endpoint Details. (The example below uses azure-monitor-endpoint.)

4. Click Azure Monitor.

5. From the copied connection string value from Azure, paste the following values:

Endpoint (IngestionEndpoint)	https://xxxx-x.in.applicationinsights.azure.com/
Instrumentation Key	xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
Live Endpoint	https://xxxx.livediagnostics.monitor.azure.com/
Application ID	xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx

6. Click Logs, then Metrics under Telemetry Type.

7. Select HTTP for the communication protocol.

8. Click Save Configuration.

9. Using the CLI, run the following command:

curl -s --location --request POST
'https://api.ngc.nvidia.com/v2/nvcf/telemetries' \
--header 'Content-Type: application/json' \
--header 'Authorization: Bearer '$NVCF_TOKEN \
--data '{
    "endpoint": "YOUR_AZURE_MONITOR_ENDPOINT",
    "protocol": "HTTP",
    "provider": "AZURE_MONITOR",
    "types": [
        "LOGS",
        "METRICS"
    ],
    "secret": {
        "name": "YOUR_NVCF_TELEMETRY_NAME",
        "value": {
            "instrumentationKey": "YOUR_INSTRUMENTATION_KEY",
            "liveEndpoint": "YOUR_LIVE_ENDPOINT",
            "applicationId": "YOUR_APPLICATION_ID"
        }
    }
}'

Get Telemetry ID

Once the telemetry endpoint is created, we need to capture the telemetryId.

Note

This step is not required for creating the function using the NVCF UI.

echo NVCF_TOKEN="nvapi-xxxxxxxxxxxxxxxxxxxxxx"

1. Run the following command to get the telemetryId of the created Azure Monitor endpoint:

curl -s --location --request GET
'https://api.ngc.nvidia.com/v2/nvcf/telemetries' \
 --header 'Content-Type: application/json' \
 --header 'Authorization: Bearer '$NVCF_TOKEN'' | jq

2. Copy the telemetryId field for the created azure-monitor-endpoint:

"telemetryId": "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx",
"name": "azure-monitor-endpoint",
"endpoint": xxx
.
.
"createdAt":xxx

3. Store the value in a variable called Telemetry_ID:

export TELEMETRY_ID="xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"

Environment Variables

Observability implementation uses environment variables to control Vector behavior and configuration. These variables are set during NVCF function deployment and control how the container processes logs.

Environment Variable	Possible Values	Function
VECTOR_OTEL_ACTIVE	TRUE ; FALSE / Not set	When TRUE: Container uses Vector for log processing and forwarding to NVCF collector. When FALSE or unset: Container bypasses Vector and runs the Kit App directly using /entrypoint.sh
VECTOR_CONF_B64	Base64 encoded string	Provides custom Vector configuration via Base64-encoded string

• If you provide your VECTOR_CONF_B64 value, the entrypoint decodes and uses your custom Vector configuration.

• When not provided, it uses the default configuration from vector.toml which is copied to the path /opt/vector/static_config.toml inside the container.

To Base64 encode, use the following command:

base64 -w 0 vector.toml

Container to Function Flow

• Using the CLI:

curl -s -v --location --request POST 'https://api.ngc.nvidia.com/v2/nvcf/functions' \
--header 'Content-Type: application/json' \
--header 'Authorization: Bearer '$NVCF_TOKEN'' \
--data '{
    "name": "'${STREAMING_FUNCTION_NAME:-usd-composer}'",
    "inferenceUrl": "'${STREAMING_START_ENDPOINT:-/sign_in}'",
    "inferencePort": '${STREAMING_SERVER_PORT:-49100}',
    "health": {
        "protocol": "HTTP",
        "uri": "/v1/streaming/ready",
        "port": '${CONTROL_SERVER_PORT:-8111}',
        "timeout": "PT10S",
        "expectedStatusCode": 200
},
"containerImage": "'$STREAMING_CONTAINER_IMAGE'",
"apiBodyFormat": "CUSTOM",
"description": "'${STREAMING_FUNCTION_NAME:-usd-composer}'",
"functionType": "STREAMING",
"containerEnvironment": [
    {"key": "NVDA_KIT_NUCLEUS", "value": "'$NUCLEUS_SERVER'"},
    {"key": "OMNI_JWT_ENABLED", "value": "1"},
    {"key": "VECTOR_OTEL_ACTIVE", "value": "TRUE"},
    {"key": "NVDA_KIT_ARGS", "value":
"--/app/livestream/nvcf/sessionResumeTimeoutSeconds=300"}
  ],
  “telemetries”: {
    “logsTelemetryId”: “'$TELEMETRY_ID'”,
    "metricsTelemetryId": "'$TELEMETRY_ID'"
}

• Using the UI:

• Or if Base64 is used:

Verify Telemetry Appears within Azure Monitor

Sample KQL queries:

For Logs:

AppTraces
| where Properties.function_id == "xxxxxxxxxxxx"

For Metrics:

AppMetrics
| where Properties.function_id == "xxxxxxxxxxxx"