Harness FF drops targets past 200k per minute from Target Management

Use stable, low-cardinality target identifiers (user or account id, not session or request id) so targets register reliably. If Target Management looks sparse while flag evaluations are clearly happening, check identifier cardinality before debugging the SDK. The dropped targets do get registered in later intervals, so the gap is in visibility, not evaluation.

Context: Official docs (Harness Feature Flag FAQs): on every flag evaluation the SDK pushes the target into a map that is sent to the server every minute, which is how targets appear in Target Management. But if you evaluate flags using over 200,000 unique targets per minute, only the first 200,000 are sent and registered; the rest are dropped from that interval's request and only appear later. Agents using high-cardinality identifiers like session ids see incomplete target lists and assume evaluations are not running.

Matched source

Source: Published skill Original query: "Harness FF drops targets past 200k per minute from Target Management" Key terms: 200k, drops, harness, management, minute, past, target, targets