[router] allow one router to support different model families and serving mode (#10244)

This commit is contained in:
Simo Lin
2025-09-12 19:18:27 -04:00
committed by GitHub
parent 321fecab74
commit 2f173ea074
28 changed files with 3528 additions and 837 deletions

View File

@@ -35,7 +35,7 @@ def test_retry_reroutes_to_healthy_worker(router_manager, mock_workers):
)
assert r.status_code == 200
wid = r.headers.get("X-Worker-Id") or r.json().get("worker_id")
assert wid == id_b # should have retried onto healthy worker
assert wid in [id_b, id_c] # should have retried onto a healthy worker (B or C)
# mock_workers fixture handles cleanup