Skip to main content
Applies to BloodHound Enterprise and CE Full collection of MFA factors and application-user assignments has been observed to require millions of API requests in large Okta organizations. Collections of this size can take a long time because Okta applies strict rate limits to different API endpoints as well as limits on concurrent requests. In addition, requests that return large result sets, such as expanded group collection, have been observed to time out repeatedly in large Okta organizations and eventually fail, leaving the affected data missing from the collection.

Limit Collection Scope

By default, the collector collects MFA factors for all users and application-user assignments for all applications. Use the following settings under [sources.source.okta] in config.toml, or their environment variables using the SOURCES__SOURCE__OKTA prefix, to significantly reduce request volume:
Parameter NameEnvironment VariableDescription
limit_user_mfa_collectionCollect MFA factors only for users with role assignments.
limit_app_user_collectionCollect application-user assignments only for the collector’s built-in set of application types.
app_user_collection_app_idsAdditional Okta application IDs to collect when limit_app_user_collection is true.
For example, to enable both limits:
config.toml

Tune Extraction Concurrency

Increasing workers or concurrent requests can improve throughput when capacity is available, but cannot bypass Okta’s API rate limits. Try overriding the default extract parameters for Okta:
config.toml
The collector also applies the following request controls under [sources.source.okta]. Replace {PREFIX} with SOURCES__SOURCE__OKTA in the environment variables below.
Parameter NameEnvironment VariableDefaultDescription
endpoint_concurrency32Maximum simultaneous requests per API endpoint.
org_concurrency32Maximum simultaneous requests across all endpoints.
rate_limit_max_elapsed_seconds900Maximum elapsed retry window for an individual HTTP 429 request.
rate_limit_remaining_reserve1Requests held in reserve when pacing against a rate-limit window.
Keep concurrency within your organization’s endpoint-specific rate limits and concurrency limits, which default to 75 simultaneous transactions, and account for other applications using the API. Retain a client-specific rate limit to avoid consuming the organization’s full quota. Use performance telemetry to see the effective settings and whether a collection is limited by worker capacity or by Okta rate limits. Note that the collector slows down before exhausting rate-limit quotas, so a collection can be quota-bound even without HTTP 429 responses.

Tune Page Sizes

These are advanced settings. The default page sizes should work well for most environments.
Because requests that return large result sets, such as expanded group collection, have been observed to time out repeatedly and eventually fail in large Okta organizations, the collector requests explicit page sizes for these resources. You can tune the default page sizes with the following environment variables using the SOURCES__SOURCE__OKTA prefix:
Environment VariableDefaultDescription
200Applications per page, from 1 through 200.
500Application users per page, from 1 through 500.
200Groups per page, from 1 through 200.
200Application-group assignments per page, from 20 through 200.
1000Group push mappings per page, from 1 through 1,000.
200Identity-provider users per page, from 1 through 200.
If the initial expanded group page repeatedly times out on large Okta tenants, the collector retries that first page with successively halved page sizes before failing. With the default GROUPS_PAGE_SIZE of 200, that fallback sequence is 200, then 100, then 50.