"(Tesla.Mock.Error) There is no mock set for process" despite defining a mock

Hi, I’m getting the following error message in my tests, and I’m not sure where they come from. I can’t see any difference to Tesla.Mock — tesla v1.8.0, except for async: true, but setting that to false doesn’t change anything.

I have other tests in the same project that also use Tesla.Mock, which don’t have this problem, and I couldn’t find a difference there either. Mocking works when I use mock_global and async: false, but that breaks a lot of other testcases, so that’s not an option.

I have 2 tests that fail because of this. They both reference the same process #PID<0.902.0> in the error message

Any help would be appreciated. Thanks

above_mock: #PID<0.914.0>
above_post: #PID<0.914.0>

  1) test ... (...)
     ** (Tesla.Mock.Error) There is no mock set for process #PID<0.902.0>.
     Use Tesla.Mock.mock/1 to mock HTTP requests.

     See https://github.com/teamon/tesla#testing

       (tesla 1.8.0) lib/tesla/mock.ex:206: Tesla.Mock.call/2
       (tesla 1.8.0) lib/tesla/middleware/timeout.ex:59: anonymous fn/1 in Tesla.Middleware.Timeout.safe_async/2
       (elixir 1.15.7) lib/task/supervised.ex:101: Task.Supervised.invoke_mfa/2
       (elixir 1.15.7) lib/task/supervised.ex:36: Task.Supervised.reply/4

# config
config :tesla, adapter: Tesla.Mock

# Test file
def MyApp.RedactedTest do
  use MyApp.DataCase, async: true
  import Tesla.Mock

  setup :general_setup

  describe "describe" do
    test "..." do
      IO.inspect(self(), label: :above_mock)
      mock(fn _ -> json(%{translations: []}) end)

# Implementation

defmodule MyApp.Client do
  def client(opts) do
     middlewares = [...]

  def call_service(client, body) do
    IO.inspect(self(), label: :above_post) 
    Tesla.post(client, "path", body)

# call_service is called like this
client = Client.client([...])
body = "body here"
Client.call_service(client, body)

Is the code under test performing the http request from a Task or from a GenServer? Those process pids will not have the Tesla mocked call available to them.

I have solved this with a combo of mock_global and async: false in the test file. This means it doesn’t step on my async tests. You will need to use both mock_global and async: false on each test file where this error appears if you are unable (or don’t want to due to requirements) refactor the http call to happen in a different process

There is no Task or GenServer involved. The first 2 lines of my code block show that the mock and the http request happen in the same process. It’s deeply nested, but it happens immediately after.

above_mock: #PID<0.914.0>
above_post: #PID<0.914.0>

Can you up your stack trace length / post the full stack trace? It seems like the mock call is coming from a task:

(elixir 1.15.7) lib/task/supervised.ex:101: Task.Supervised.invoke_mfa/2
(elixir 1.15.7) lib/task/supervised.ex:36: Task.Supervised.reply/4

Unfortunately, this already is the full stack track provided by the test. I didn’t touch the default values of the length, so it should already be the whole stack trace