[PD] Deferred decode-side KV release for aborts mid-transfer (#35049)

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
Shangming Cai
2026-08-18 23:56:33 +08:00
committed by GitHub
co-authored by Claude Opus 4.8
parent c60952d933
commit 97dedd1ce9
8 changed files with 507 additions and 34 deletions
@@ -211,6 +211,7 @@ class TestDecodeQueueCleanup(CustomTestCase):
queue = DecodeTransferQueue.__new__(DecodeTransferQueue)
queue.queue = [decode_req]
queue.enable_staging = False
queue.enable_deferred_kv_release = False
queue.gloo_group = MagicMock()
queue.req_to_metadata_buffer_idx_allocator = MagicMock()
queue.tp_rank = 0