Cloud Trace API - Class Google::Cloud::Trace::Middleware (v0.45.0)

Reference documentation and code samples for the Cloud Trace API class Google::Cloud::Trace::Middleware.

Trace Middleware

A Rack middleware that manages trace context and captures a trace of the request. Specifically, it:

  • Reads the trace context from the request headers, if present. Otherwise, generates a new trace context.
  • Makes a sampling decision if one is not already specified.
  • Records a span measuring the entire handling of the request, annotated with a set of standard request data.
  • Makes the trace context available so downstream middlewares and the app can add further spans to the trace.
  • Sends the completed trace to the Stackdriver service.

Installing

To use this middleware, simply install it in your middleware stack. Here is an example Sinatra application that includes the Trace middleware:

```ruby

Simple sinatra application

require "sinatra" require "google/cloud/trace"

use Google::Cloud::Trace::Middleware

get "/" do "Hello World!" end ```

Here is an example config.ru file for a web application that uses the standard Rack configuration mechanism.

```ruby

config.ru for simple Rack application

require "google/cloud/trace" use Google::Cloud::Trace::Middleware

run MyApp ```

If your application uses Ruby On Rails, you may also use the provided Railtie for close integration with Rails and ActiveRecord.

Custom measurements

By default, this middleware creates traces that measure just the http request handling as a whole. If you want to provide more detailed measurements of smaller processes, use the classes provided in this library. Below is a Sinatra example to get you started.

```ruby

Simple sinatra application

require "sinatra" require "google/cloud/trace"

use Google::Cloud::Trace::Middleware

get "/" do Google::Cloud::Trace.in_span "Sleeping on the job!" do sleep rand end "Hello World!" end ```

Error handling

An error encountered during the reporting of traces by the middleware can be handled using a Proc set in the on_error configuration. (See configure.) The Proc must take the error object as the single argument.

```ruby

Configure error handling

require "sinatra" require "google/cloud/trace" require "google/cloud/error_reporting"

Google::Cloud::Trace.configure do |config| config.on_error = lambda do |error| Google::Cloud::ErrorReporting.report error end end

use Google::Cloud::Trace::Middleware

get "/" do Google::Cloud::Trace.in_span "Sleeping on the job!" do sleep rand end "Hello World!" end ```

Sampling and blacklisting

A sampler makes the decision whether to record a trace for each request (if the decision was not made by the context, e.g. by providing a request header). By default, this sampler is the default TimeSampler, which enforces a maximum QPS per process, and blacklists a small number of request paths such as health checks sent by Google App Engine. You may adjust this behavior by providing an alternate sampler. See TimeSampler.

Inherits

  • Object

Methods

#call

def call(env) -> Rack::Response

Implementation of the trace middleware. Creates a trace for this request, populates it with a root span for the entire request, and ensures it is reported back to Stackdriver.

Parameter
  • env (Hash) — Rack environment hash
Returns
  • (Rack::Response) — The response from downstream Rack app

#initialize

def initialize(app, service: nil, **kwargs) -> Middleware

Create a new Middleware for traces

Parameters
  • app (Rack Application) — Rack application
  • service (Google::Cloud::Trace::Service, AsyncReporter) (defaults to: nil) — The service object to update traces. Optional if running on GCE.
  • kwargs (Hash) — Hash of configuration settings. Used for backward API compatibility. See the Instrumentation Guide and Configuration Guide for the prefered way to set configuration parameters.
Returns

Constants

AGENT_NAME

value: "ruby #VERSION".freeze
The name of this trace agent as reported to the Stackdriver backend.