ApolloTracing adds data to your GraphQL query response so that an Apollo Engine can provide insights into your Absinthe-based GraphQL service.

• Performance Tracing
• Response Caching

Add :apollo_tracing to your deps

def deps do [ {:apollo_tracing, "~> 0.4.0"} ] end

ApolloTracing uses the Absinthe's middleware functionality to track field-level resolution times. In order to register our custom middleware, you have a few options:

Add use ApolloTracing to your schema file:

def MyApp.Schema do use Absinthe.Schema use ApolloTracing end

If you have a custom middleware stack, add the apollo tracing middlewares to the beginning of your middleware stack:

def middleware(middleware, _field, _object), do: [ApolloTracing.Middleware.Tracing, ApolloTracing.Middleware.Caching] ++ [...your other middlewares]

If you prefer to only add tracing to some fields, you can selectively add tracing information:

field :selected_field, :string do middleware ApolloTracing.Middleware # Has to be the first middleware resolve fn _, _ -> {:ok, "this field is now added to be traced"} end end

ApolloTracing currently requires you to use a custom Pipeline in order to register 'Phases' in the correct order during resolution. Phases are used for measuring overall query times as well as appending the custom data to the response (including cache hints).

Specify the pipeline in your Absinthe.Plug endpoint:

forward "/graphql", Absinthe.Plug, schema: MyApp.Schema, pipeline: {ApolloTracing.Pipeline, :plug}

If you have your own pipeline function, you can add the phases directly:

def my_pipeline_creator(config, pipeline_opts) do config.schema_mod |> Absinthe.Pipeline.for_document(pipeline_opts) |> add_my_phases() # w.e your custom phases are |> ApolloTracing.Pipeline.add_phases() # Add apollo at the end end

When you want to just call run a query with tracing, but without going through a Plug endpoint:

def custom_absinthe_runner(query, opts \\ []) do pipeline = ApolloTracing.Pipeline.default(YourSchema, opts) case Absinthe.Pipeline.run(query, pipeline) do {:ok, %{result: result}, _} -> {:ok, result} {:error, err, _} -> {:ok, err} end end """  query {  fielda  fieldb  } """ |> custom_absinthe_runner()

You can configure caching by adding metadata to your Absinthe objects:

object :user do meta :cache, max_age: 30 end # or object :user, meta: [cache: [max_age: 30]] do # ... end

To ensure that the object is not cached across users, you can mark it as private:

object :user do meta :cache, max_age: 30, scope: :private end

See the Apollo docs for more information about cache scope.