Hi - I have a job which runs every hour to collect and report on any/all runs which have failed in the last hour. I am attempting to gather the assets from any job which materializes assets that were affected by the failure and did not successfully materialize. (Yes, I know there's a run failure sensor but it would send an email for each failure during that period while I need a single email containing all failures during that period) To that end, I need to write a unit test which can test fetching context and failure event data from (hopefully mocked) failed assets. Anyone have any suggestions on how I could accomplish this?

Hi Adam, There’s a function called

instance_for_test

that’s part of our public API. We use this extensively internally as well. You use it like this:

from dagster import instance_for_test

with instance_for_test() as instance:
    some_job.execute_in_process(instance=instance)
    ...

You can use this approach to populate the test instance with various failed runs. This should allow you to set it up appropriately to test your job.

That is fantastic! Thank you!

pretty sure it uses sqlite but just in a tmpdir: https://github.com/dagster-io/dagster/blob/master/python_modules/dagster/dagster/_core/instance_for_test.py#L16-L82