Modifier

Partager via


Telemetry logger

This article covers the telemetry logger in Dynamics 365 Commerce.

The Dynamics 365 Commerce online software development kit (SDK) includes a custom telemetry logger that you can use to log to multiple resources at various levels while also maintaining a unified context on both the server and the client.

Access the telemetry logger

By default, the telemetry logger is available to all react components. It can be accessed with the this.props.telemetry property from the module view code file.

Sometimes you want to access the telemetry logger in a shared component instead of having to pass it down through the properties of every component in your module. The SDK repository includes a WithContext() utility that lets you inject the telemetry logger directly into your component.

When the telemetry logger runs in a development environment, all logs are also shown in the console.

Trace logging

The messages that are logged can range in importance from debug statements to critical errors. This logging is available through the trace logging feature of the telemetry logger. The trace logger uses the following logging methods and levels.

enum LogLevel {
    Trace = 'trace',
    Debug = 'debug',
    Information = 'information',
    Warning = 'warning',
    Error = 'error',
    Critical = 'critical',
    None = 'none'
}

Trace logging APIs

The following example shows the primary application programming interface (API) that is used to log a trace message.

telemetry.log(
    logLevel: LogLevel,                 // One of the log levels listed in the enum above
    messageTemplate: string,            // A string that follows a message template format (see below for more info)
    logOptions?: TelemetryLogOptions    // Object wrapping up the optional parameters for the log statement
)

The trace logging system uses message templates to enforce a structured logging system for telemetry. Therefore, the logs can be captured and rendered in both human-friendly and machine-friendly formats. The messageTemplate argument of the .log call uses what are known as named holes for any event data. When you're writing trace log messages, you can write human-readable messages that include event data. To include event data, mark the place in the message template by using {<name>}, where <name> is the name that you want to use for the event data. The actual value that fills that named hole is provided in the logOptions.values parameter. The logOptions parameter contains any optional parameters that should be included for a trace log. Here's an example.

type TelemetryLogOptions = {
    values?: unknown[];       // Holds any arguments that are meant for placeholders in the message template
    customTags?: string[];    // Array of custom tags to add to log. Custom tags can be used to group message in the telemetry back-end
    exception?: Error;        // Exception that can be attached to the log. Will contain details like stack trace info
};

The benefit of this structured logging system is that when log messages are rendered so that a human can read them, the system can replace the named holes with values that you provide. However, when the messages are sent and stored, or when a machine processes the telemetry, the event data can be kept separate. For example, event data can be used to aggregate or filter specific messages without requiring any string parsing. For more information about structured logging and message templates, see Message Templates.

The following examples show some trace log calls.

telemetry.log(LogLevel.Debug, "{user} says {word}", {values: ["Bill", "Hi!"]});
// Output: "Bill says Hi!"

telemetry.log(LogLevel.Debug, "Customer {customer} purchased item {productID}", {values: [12345, 321]});
// Output: "Customer 12345 purchased item 321

telemetry.log(LogLevel.Debug, "Module {id} threw error {code} while rendering", {values: [123, 404], customTags: ["Module Error"], exception: error});
// Output: "Module 123 threw error 404 while rendering
// The customTags and exception will be attached to the log call, and can be viewed in the telemetry back-end

There are also wrapper methods that simplify the logging calls. However, these wrapper methods can log only strings. There's one wrapper method for each log level, as the following example shows.

this.props.telemetry.trace("This will log at log level 'trace'");
this.props.telemetry.debug("This will log at log level 'debug'");
this.props.telemetry.information("This will log at log level 'information'");
this.props.telemetry.warning("This will log at log level 'warning'");
this.props.telemetry.error("This will log at log level 'error");
this.props.telemetry.critical("This will log at log level 'critical");

If you want to include additional strings or other objects in the wrapper methods, you can add them to the message by passing them as arguments at the end of the call. All additional arguments are converted to strings that are joined to the first string message and separated by a comma, as shown in the following example.

this.props.telemetry.trace("This is the first message", "This is the second message", {some object})

// Output will be: "This is the first message,This is the second message,{.toString result of {some object}}"

The ?debug=true query string controls the console logs.

Exception logging

Logging of Error objects (and classes that inherit from them) can be done through the .exception(Error) API, aa shown in the following example.

this.props.telemetry.exception(new Error("Something is broken!"));

error() API vs. exception() API

You might be confused about when you should use the error() API to log an error in your application, and when you should use the exception() API. This confusion can arise because the names are similar, and because you can use the error() API to log Error objects by passing the Error objects as additional parameters.

The best guidance is to use the exception() API to log actual Error objects and the error() API to log string messages that state that an error occurred in the business logic. Generally, exception() API logs are more easily correlated with issues and allow for faster debugging when real issues arise. The messages from the error() API are treated as another trace log, and more detailed analysis might be required to find the issue than if you use the exception() API. Therefore, it can take more time to recognize that an issue occurred. In addition, the exception() API allows for better tracking across different requests. Therefore, it supports features such as automatic alerting when an issue begins to affect many requests.

Expose telemetry data in Azure Application Insights

To connect telemetry events to your Azure Application Insights subscription, follow the instructions in Platform settings file.

Additional resources

Online channel extensibility overview