feat: search edge identities by trait (WIP)

[matthewelwell]

Changes

This is a WIP implementation of #4016

How did you test this code?

N/a (so far)

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

3 Skipped Deployments

Name	Status	Preview	Comments	Updated (UTC)
docs	⬜️ Ignored (Inspect)	Visit Preview		Aug 15, 2024 0:43am
flagsmith-frontend-preview	⬜️ Ignored (Inspect)	Visit Preview		Aug 15, 2024 0:43am
flagsmith-frontend-staging	⬜️ Ignored (Inspect)	Visit Preview		Aug 15, 2024 0:43am

[github-actions]

Uffizzi Preview deployment-55286 was deleted.

[matthewelwell]

I've done a quick performance test of this implementation and, I think, it's not particularly useable as it is here. The test created 5M identities, each with 5 traits, and searched for a particular trait key / value combination. On my local machine, this query took 20s to return.

Here is the (crude) test I used:

def test_search_edge_api_identity_meta(project: Project) -> None:
    # Given
    num_environments = 5
    num_identities = 1_000_000
    num_traits_per_identity = 5

    for _environment in (
        Environment.objects.create(name=f"environment_{i}", project=project)
        for i in range(1, num_environments + 1)
    ):
        to_create = []
        start = time.time()
        for j in range(1, num_identities + 1):
            edge_identity_meta = EdgeIdentityMeta(
                identifier=f"identity_{j}",
                edge_identity_uuid=str(uuid.uuid4()),
                environment=_environment,
            )
            traits = [
                TraitModel(
                    trait_key=f"{_environment.name}_trait_{j}_{k}",
                    trait_value=f"value_{j}_{k}"
                )
                for k in range(1, num_traits_per_identity + 1)
            ]
            edge_identity_meta.update_searchable_traits(traits)
            to_create.append(edge_identity_meta)
        EdgeIdentityMeta.objects.bulk_create(to_create)

        logger.info(f"Created EdgeIdentityMeta objects for environment {_environment.name} in {time.time() - start} seconds")

    environment_id_to_search_for = random.choice(list(Environment.objects.values_list("id", flat=True)))
    environment_name_to_search_for = Environment.objects.get(id=environment_id_to_search_for).name
    identity_id_to_search_for = random.randint(1, num_identities)
    trait_id_to_search_for = random.randint(1, num_traits_per_identity)
    trait_key_to_search_for = f"environment_{environment_name_to_search_for}_trait_{identity_id_to_search_for}_{trait_id_to_search_for}"
    trait_value_to_search_for = f"value_{identity_id_to_search_for}_{trait_id_to_search_for}"
    search_string = f"{trait_key_to_search_for}{EdgeIdentityMeta._EQUALITY_CHARS}{trait_value_to_search_for}"

    # When
    start = time.time()
    result_exists = EdgeIdentityMeta.objects.filter(
        environment_id=environment_id_to_search_for,
        searchable_traits__search=search_string
    ).exists()
    end = time.time()

    # Then
    assert result_exists, search_string

    query_time_ms = (end - start) * 1000
    logger.info("query time = %sms", query_time_ms)
    assert query_time_ms < 1

[matthewelwell]

By adding the index as per this commit, I've brought the query time down to ~500ms, but even that seems like a lot here.

[codecov]

Codecov Report

Attention: Patch coverage is 97.22222% with 1 line in your changes missing coverage. Please review.

Project coverage is 96.93%. Comparing base (bf82b9d) to head (362346a).
Report is 10 commits behind head on main.

Files	Patch %	Lines
api/edge_api/identities/models.py	93.75%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4493      +/-   ##
==========================================
+ Coverage   96.91%   96.93%   +0.02%     
==========================================
  Files        1176     1180       +4     
  Lines       39349    39526     +177     
==========================================
+ Hits        38135    38316     +181     
+ Misses       1214     1210       -4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.