Just come to this thread after a Google search. I can share some practice with resetting the read store.
The recommended testing approach by Commanded community conflicts with Phoenix’s sandbox one.
I believe sandbox is the better way because of concurrent testing, so I also spent some time trying to make the read store repo run in sandbox mode. It turns out that it’s easier than I thought since it already is. What I did was to remove the manual resetting (truncates).
Here’s what testing support files look like:
# file: test/support/storage.ex
defmodule MyAp.Storage do
@moduledoc """
Clear the event store and read store databases
"""
alias EventStore.Storage.Initializer
def reset! do
reset_eventstore()
# For read store, it is expected to be automatically
# rolled back by the transactional tests, as we are
# using sandbox mode of the pool.
end
defp reset_eventstore do
config = Investracker.EventStore.config()
{:ok, conn} = Postgrex.start_link(config)
Initializer.reset!(conn, config)
end
end
# file: test/support/data_case.ex
defmodule MyApp.DataCase do
@moduledoc """
Test case template with data cleaning up.
"""
...
using do
quote do
import Ecto
import Ecto.Changeset
import Ecto.Query
import Commanded.Assertions.EventAssertions
import unquote(__MODULE__)
end
end
setup tags do
{:ok, _} = Application.ensure_all_started(:my_app)
setup_sandbox(tags)
:ok
end
@doc """
Sets up the sandbox based on the test tags.
"""
def setup_sandbox(tags) do
pid =
Ecto.Adapters.SQL.Sandbox.start_owner!(
Investracker.Repo,
shared: not tags[:async]
)
on_exit(fn ->
:ok = Application.stop(:my_app)
# clean event store
MyApp.Storage.reset!()
# clean read store with the help of sandbox rollback
Ecto.Adapters.SQL.Sandbox.stop_owner(pid)
end)
end
end
Ideally we can also employ sandbox to reset event store, too, but I haven’t run into any issue yet, so I leave it as it is.
After a reset of the read store with MIX_ENV=test mix ecto.reset, everything seems working now.
