Azure Monitor

Overview

Azure Monitor sends logs, metrics and traces to Azure Application Insights as envelope telemetry. Log records land in the traces table, spans in requests and dependencies, and metrics in customMetrics.

Supported platforms

• Linux: Logs, Metrics, Traces
• Windows: Logs, Metrics, Traces
• macOS: Logs, Metrics, Traces

Authentication

Attach a Connection String credential holding the Application Insights connection string:

InstrumentationKey=<guid>;IngestionEndpoint=https://<region>.in.applicationinsights.azure.com/

Copy it from the Application Insights resource Overview page in the Azure portal. The instrumentation key inside it both identifies the resource and authorizes the write, so no other credential is needed.

Note: The credential type is shared with other Azure services, so the form does not say which one it is for. A connection string for a different service is rejected when the pipeline is saved.

Basic Configuration

Parameter	Type	Default	Required	Description
maxbatchsize	integer	1024	No	Telemetry items buffered before a batch is sent.
maxbatchinterval	string	10s	No	Longest wait before sending a partial batch.

Note: The two batch settings race — whichever is reached first sends the batch.

Telemetry Configuration

These route log records to a different Application Insights table. A record is only re-routed when the setting is on and it carries the matching attribute.

Parameter	Type	Default	Required	Description
spaneventsenabled	boolean	false	No	Export non-exception span events. Exception events are always exported.
custom_events_enabled	boolean	false	No	Records with a microsoft.custom_event.name attribute go to customEvents.
exception_events_enabled	boolean	false	No	Records with exception.type or exception.message go to exceptions.

Advanced Configuration

Parameter	Type	Default	Required	Description
timeout	string	10s	No	Maximum time to wait for an ingestion request.
shutdown_timeout	string	1s	No	How long to wait for queued telemetry to flush on shutdown.
trace_http_success_policy	string	none	No	Set to otel to mark 4xx HTTP server spans as Success. Traces only.
trace_http_additional_success_status_codes	array	[]	No	Extra HTTP status codes treated as Success, e.g. 404, 409. Traces only.
cloud_role_instance	array	[]	No	Sources for the ai.cloud.roleInstance tag, in order. Empty uses the default.
application_version	array	[]	No	Sources for the ai.application.ver tag, in order. Empty uses the default.
sending_queue	object	—	No	Backpressure queue size, consumers, and optional disk backing.

Tag Sources

cloud_role_instance and application_version take an ordered list of sources; the first that resolves to a non-empty value wins. An entry containing a dot is read as a resource attribute (host.name); an entry without one is a literal fallback (unknown-instance), so it must come last.

Defaults are service.instance.id and service.version respectively.

Example Configuration

{
  "maxbatchsize": 1024,              // default: 1024
  "maxbatchinterval": "10s",         // default: "10s"

  "spaneventsenabled": false,        // default: false
  "custom_events_enabled": false,    // default: false
  "exception_events_enabled": false, // default: false

  "advanced": {
    "timeout": "10s",                // default: "10s"
    "shutdown_timeout": "1s",        // default: "1s"
    "cloud_role_instance": ["host.name", "service.instance.id", "unknown-instance"],
    "sending_queue": {
      "queue_size": 1000,            // default: 1000
      "num_consumers": 10,           // default: 10
      "enable_disk_backed_queue": true,
    },
  },
}

Where the data lands

Signal	Application Insights table	Log Analytics workspace table
Logs	traces	AppTraces
Logs with custom events enabled	customEvents	AppEvents
Logs with exception events enabled	exceptions	AppExceptions
Traces	requests, dependencies	AppRequests, AppDependencies
Metrics	customMetrics	AppMetrics

TraceId is mapped to the operation_id column and SpanId to operation_parentId.

Note: The table name depends on where you query from. The Application Insights resource uses the classic names; the Log Analytics workspace backing it uses the App-prefixed names.