Add tests for data refresh task dependencies

[stacimc]

Fixes

Fixes #4340 by @stacimc

Description

The new data refresh has already been added to the load_sample_data scripts, such that it is run each time a contributor runs ov j api/init and also in CI. This adds a lot of coverage for the "happy path" of the data refresh.

In addition to this, however, it's important to test that certain vital task dependencies are maintained. While developing the data refresh I found that it's surprisingly easy to accidentally break the dependency graph in a manner such that, while a normal data refresh works as expected, an upstream task failure may not cascade properly. TL;DR: it's easy to introduce a change that seems fine, but causes elasticsearch reindexing/promotion to occur even if the copy_data step fails. This is a huge concern because it could result in bad or incomplete data going live.

This PR adds tests to ensure that those vital dependencies are maintained.

Testing Instructions

Tests should pass; take a look and see if there are any other dependencies you feel should be asserted.

Checklist

• My pull request has a descriptive title (not a vague title likeUpdate index.md).
• My pull request targets the default branch of the repository (main) or a parent feature branch.
• My commit messages follow best practices.
• My code follows the established code style of the repository.
• I added or updated tests for the changes I made (if applicable).
• I added or updated documentation (if applicable).
• I tried running the project locally and verified that there are no visible errors.
• I ran the DAG documentation generator (ov just catalog/generate-docs for catalog
  PRs) or the media properties generator (ov just catalog/generate-docs media-props
  for the catalog or ov just api/generate-docs for the API) where applicable.

Developer Certificate of Origin

Developer Certificate of Origin

Developer Certificate of Origin
Version 1.1

Copyright (C) 2004, 2006 The Linux Foundation and its contributors.
1 Letterman Drive
Suite D4700
San Francisco, CA, 94129

Everyone is permitted to copy and distribute verbatim copies of this
license document, but changing it is not allowed.


Developer's Certificate of Origin 1.1

By making a contribution to this project, I certify that:

(a) The contribution was created in whole or in part by me and I
    have the right to submit it under the open source license
    indicated in the file; or

(b) The contribution is based upon previous work that, to the best
    of my knowledge, is covered under an appropriate open source
    license and I have the right under that license to submit that
    work with modifications, whether created in whole or in part
    by me, under the same open source license (unless I am
    permitted to submit under a different license), as indicated
    in the file; or

(c) The contribution was provided directly to me by some other
    person who certified (a), (b) or (c) and I have not modified
    it.

(d) I understand and agree that this project and the contribution
    are public and that a record of the contribution (including all
    personal information I submit with it, including my sign-off) is
    maintained indefinitely and may be redistributed consistent with
    this project or the open source license(s) involved.

[obulat]

Approving because the tests pass, but I'm not sure I understand this:

it's easy to introduce a change that seems fine, but causes elasticsearch reindexing/promotion to occur even if the copy_data step fails. This is a huge concern because it could result in bad or incomplete data going live.

Could you give an example of how it can occur or when it occurred? And why is it not possible to use Airflow to somehow prevent this from happening?

[stacimc]

Could you give an example of how it can occur or when it occurred? And why is it not possible to use Airflow to somehow prevent this from happening?

Sure, here's an example of a fix for one I noticed in the last PR. In that case the issue was that I altered the return statement of a decorated @task_group so that we could access its value in XCOMs; under the covers, though, this also changes the default way that Airflow sets up task dependencies (normally the last task in a task group is set upstream of downstream tasks, but if you have an explicit return, that task and that task alone is set upstream). It's also not possible, to my knowledge, to manually/explicitly set dependencies between tasks inside task groups and outside of task groups. In a DAG as complicated as the data refresh, it's easy to not notice this sort of dependency being broken, particularly because it only "matters" in this case if there's an error.