lightstep-tracer

lightstep-tracer

LightStep distributed tracing library for Node.js and the browser.

Installation

npm install --save lightstep-tracer opentracing

All modern browsers and Node versions >= 8 are supported.

Getting started

To use LightStep as the OpenTracing binding, initialize the global Tracer with the LightStep implementation:

var opentracing = require('opentracing');
var lightstep   = require('lightstep-tracer');

opentracing.initGlobalTracer(new lightstep.Tracer({
    access_token   : '{your_access_token}',
    component_name : '{your_service_or_app_name}',
}));

The LightStep JavaScript Tracing Cookbook is a good next stop for information on how to quickly instrument your system. If you want to try something out quickly in your browser code, see the browser quick start example.

• For more information about using the OpenTracing API, see http://opentracing.io/
• See examples/browser for a complete JavaScript browser example
• See examples/node for a complete Node.js server-side example

The browser version of the code can be explicitly included using the following, which can be helpful in some browserify (or similar) setups:

var lightstep = require('lightstep-tracer/browser');

LightStep-specific API

The OpenTracing standard JavaScript API is documented here. The following describes LightStep-specific options and methods.

LightStep

---

new Tracer(options)

Required options

• access_token string required - the project access token
• component_name string required - the string identifier for the application, service, or process

Standard options

• verbosity number optional, default=1 - controls the level of logging to the console
  • 0 - the client library will never log to the console
  • 1 - error reporting will be throttled to the first error per minute
  • 2 - all errors are logged to the console
  • 3 - all errors, warnings, and info statements are logged to the console
  • 4 - all log statements, including debugging details
• collector_host string optional - custom collector hostname
• collector_port number optional - custom collector port
• collector_path string optional - custom collector base path (if served behind a reverse proxy)
• collector_encryption string optional, default='tls'
  • tls - use HTTPS encrypted connections
  • none - use HTTP plain-text connections

Browser-specific initialization options

• instrument_page_load bool - creates a long-lived single span for the entire page view and is recommended for short-page visits in a multi-page website. For a single-page web app, this behavior may be undesirable. Defaults to false. This must be set at initialization, changes after initialization will have no effect.

• xhr_instrumentation bool - if false, disables automatic instrumentation of XMLHttpRequests (XHRs). This must be set at initialization; changes after initialization will have no effect. Defaults to false.

• xhr_url_inclusion_patterns RegExp[] - an array of regular expressions used to whitelist URLs for XMLHttpRequest auto-instrumentation. The default value is wildcard matching all strings. For a given URL to be instrumented, it must match at least one regular expression in xhr_url_inclusion_patterns and not match any regular expressions in xhr_url_exclusion_patterns.

• xhr_url_exclusion_patterns RegExp[] - an array of regular expressions used to exclude URLs from XMLHttpRequest auto-instrumentation. The default value is an empty array. For a given URL to be instrumented, it must match at least one regular expression in xhr_url_inclusion_patterns and not match any regular expressions in xhr_url_exclusion_patterns.

• fetch_instrumentation bool - if false, disables automatic instrumentation of window.fetch. This must be set at initialization; changes after initialization will have no effect. Defaults to false.

• fetch_url_inclusion_patterns RegExp[] - an array of regular expressions used to whitelist URLs for window.fetch auto-instrumentation. The default value is wildcard matching all strings. For a given URL to be instrumented, it must match at least one regular expression in fetch_url_inclusion_patterns and not match any regular expressions in fetch_url_exclusion_patterns.

• fetch_url_exclusion_patterns RegExp[] - an array of regular expressions used to exclude URLs from window.fetch auto-instrumentation. The default value is an empty array. For a given URL to be instrumented, it must match at least one regular expression in fetch_url_inclusion_patterns and not match any regular expressions in fetch_url_exclusion_patterns.

• include_cookies bool - if true, includes cookies in the span logs for both window.fetch and XMLHttpRequest. Defaults to true.

node-specific initialization options

• nodejs_instrumentation bool - if false, disables automatic instrumentation of node requests. This must be set at initialization; changes after initialization will have no effect. Defaults to false.

• nodejs_url_inclusion_patterns RegExp[] - an array of regular expressions used to whitelist URLs for the http and https modules auto-instrumentation. The default value is wildcard matching all strings. For a given URL to be instrumented, it must match at least one regular expression in nodejs_url_inclusion_patterns and not match any regular expressions in nodejs_url_exclusion_patterns.

• nodejs_url_exclusion_patterns RegExp[] - an array of regular expressions used to exclude URLs from the http and https modules auto-instrumentation. The default value is an empty array. For a given URL to be instrumented, it must match at least one regular expression in nodejs_url_inclusion_patterns and not match any regular expressions in nodejs_url_exclusion_patterns.

Non-standard options

NOTE: Future API compatibility on non-standard options is not guaranteed.

• disable_reporting_loop bool optional, default=false - if true, the timer that automatically sends reports to the collector will be disabled. This option is independent of disable_report_on_exit.
• disable_report_on_exit bool optional, default=false - if true, the final report that is automatically sent at process exit in Node or page unload in the browser will not be sent.
• report_timeout_millis number optional, default=30000 - the default timeout value, in milliseconds, for reports to the LightStep collector
• gzip_json_requests bool optional, default=true - if true, the reports will be gzipped before sent to the collector.
• default_span_tags string optional - an associative array of tags to add to every span started by the tracer (e.g., the active user id in a browser client)
• delay_initial_report_millis int optional, default=1000 - maximum additional delay of the initial report in addition to the normal reporting interval. A value between zero and this maximum will be selected as the actual delay. This can be useful when concurrently launching a large number of new processes and there is a desire to distribute the initial reports over a window of time.
• error_throttle_millis int optional, default=60000 - when verbosity is set to 1, this the minimum time between logged errors.
• transport string optional, default=proto - when transport is set to thrift, the Tracer will use Thrift as its transport instead of Proto over HTTP.
• logger function(level: string, message: string, payload: any): void optional - specify a custom logger function. Possible level values are debug, info, warn and error. By default messages will be logged to the console.
• disable_meta_event_reporting bool optional, default=false - when disable_meta_event_reporting is set to true, the tracer will disable meta event reporting even if requested by the Satellite.
• propagators dictionary optional, defaults={ [opentracing.FORMAT_HTTP_HEADERS]: new lightstep.LightStepPropagator(), [opentracing.FORMAT_TEXT_MAP]: new lightstep.LightStepPropagator(), [opentracing.FORMAT_BINARY]: new lightstep.UnsupportedPropagator() }: Allows inject/extract to use custom propagators for different formats. This package includes B3Propagator that supports B3 headers on text maps and http headers. DDPropagator supports DataDog trace headers.
• clear_span_buffer_consecutive_errors number optional, default=null - each consecutive buffer flush error will check to see if clear_span_buffer_consecutive_errors has been reached. If reached, the span buffer will be emptied. This is useful for auto-recovering from errors based on request size constraints, particularly max payload size on intermediate load balancers.

An example configuration using custom propagators might look like:

const tracer = new lightstep.Tracer({
  propagators: {
    [opentracing.FORMAT_HTTP_HEADERS]: new lightstep.DDPropagator(),
    [opentracing.FORMAT_TEXT_MAP]: new lightstep.B3Propagator()
  }
})

SpanImp

---

generateTraceURL()

Returns an absolute URL to the LightStep application for the trace containing this span. It is safe to call this method after finish().

...
span.finish();

var url = span.generateTraceURL())
console.log('View the trace for this span at:', url);

License

The MIT License.

Copyright (c) 2016, LightStep