Overview

We need to test the refresh. And we probably need a good set of unit tests to run on the database to check integrity. They can be ran after the refresh from datasets, refresh from objects, regularly, or ad hoc.

Will need to think through this, but basically we should run a set of tests on every table/view.

Two approaches

Testing with existing data

1. Make concept set TermHub Test Static -- just one version
2. Make sure it's not already in the database (make the test agnostic to using live or test databases)
3. Update modified_at for container and version (by updating a field like notes or limitations using actions update api)
4. Set archived = false
5. Launch db-refresh
6. Wait a little
7. Check for presence of expected data
   a. concept_set_container, code_sets, concept_set_members, concept_set version_item
   b. also derived table all_csets, cset_members_items, ... (not sure what else)
8. Check front end for presence of this concept set
9. Delete data from database
   a. concept set container, codesets, conceptset version item, concept set members
   a. rerun ddl tables (done in refresh.py)
10. Update archived = true on the container

Testing with new data

1. Have one concept set that we keep adding new versions to: TermHub Test New Version
2. Upload new version, include datetime in some metadata field
3. Update archived = false on the container
4. Launch db-refresh
5. Wait a little
6. Check for presence of expected data
   a. concept_set_container, code_sets, concept_set version_item -- NOT concept_set_members, which won't be ready for a while
   b. keep checking until concept_set_members records for new codeset_id show up
   c. also check derived table all_csets, cset_members_items, ... (not sure what else)
7. Check front end for presence of this concept set
8. Delete data from database
9. Update archived = true on the container

Solving this will solve :

• Clear cache: When DB updates #396

Specific things to test

• Make sure derived tables have new data