[V1] [P/D] Add Support for KV Load Failure Recovery (#19330)

Signed-off-by: David Ben-David <davidb@pliops.com> Co-authored-by: David Ben-David <davidb@pliops.com>
2025-10-01 00:57:08 +03:00
parent 67f3fb0844
commit 9a9f48dff7
24 changed files with 1039 additions and 86 deletions
--- a/tests/v1/worker/test_gpu_model_runner.py
+++ b/tests/v1/worker/test_gpu_model_runner.py
@ -250,6 +250,7 @@ def test_update_states_request_resumed(model_runner, dist_init):
        new_token_ids=[[]],
        new_block_ids=([[0]], ),
        num_computed_tokens=[0],
+        num_output_tokens=[0],
    )

    scheduler_output = SchedulerOutput(