3 people 4e582c5b54 [PD][Bugfix] Rebase KV lease deadlines onto worker clock (#50326) ...

Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: esmeetu <jasonailu87@gmail.com>
Co-authored-by: Yifan Qiao <yifanqiao@inferact.ai>

2026-07-30 01:25:42 +00:00

..

offloading_connector

[Bugfix][KV Offload] Keep Mamba block span unscaled under DCP (#49964)

2026-07-28 17:41:55 +03:00

__init__.py

…

test_bidirectional_kv_transfer.py

[NIXL] Avoid reading expired blocks in bidirectional turn-2 read (#47021)

2026-07-14 13:03:41 +00:00

test_cache_pollution_prevention.py

[P/D] KV Load Failure Recovery/Abort Configuration (#26813)

2025-12-10 11:00:52 -08:00

test_config.py

[KV Connector] Update lmcache kv_offloading_backend to use LMCacheMPConnector (#42865)

2026-06-03 19:23:55 -07:00

test_decode_bench_connector.py

[Spec Decode] Support hybrid attention models in extract_hidden_states (#39949)

2026-05-13 10:45:53 -07:00

test_error_propagation.py

[Perf] Optimize scheduler overhead for PD disaggregation, around 5% E2E perf improvement (#35781)

2026-03-10 21:25:30 -07:00

test_example_connector.py

[Mypy] Better fixes for the mypy issues in vllm/config (#37902)

2026-03-25 06:14:43 -07:00

test_flexkv_connector.py

[KV Connector] Support using FlexKV as KV Cache Offloading option. (#34328)

2026-03-12 00:46:20 -07:00

test_handshake_pp_aggregation.py

[Model] Add DiffusionGemma Support (#45163)

2026-06-11 22:17:35 -07:00

test_hf3fs_client.py

[KVConnector] Support 3FS KVConnector (#37636)

2026-04-07 15:46:00 +00:00

test_hf3fs_connector.py

Revert "[Platform] Replace torch.cuda.Event with torch.Event (#47140)" (#47668)

2026-07-06 14:18:33 +01:00

test_hf3fs_metadata_server.py

[KVConnector] Support 3FS KVConnector (#37636)

2026-04-07 15:46:00 +00:00

test_hidden_states_connector.py

[Spec Decode] Fix hidden-state extraction block size for hybrid verifiers (#46301)

2026-06-30 08:19:51 -07:00

test_hma_auto_config.py

[KV Transfer] Enable HMA by default for connectors that support it (#41847)

2026-05-26 12:28:51 +02:00

test_invalid_blocks_correctness.py

[Perf] Optimize scheduler overhead for PD disaggregation, around 5% E2E perf improvement (#35781)

2026-03-10 21:25:30 -07:00

test_kv_cache_layout.py

[XPU][CI]fix xpu kv cache layout test (#46429)

2026-06-23 03:43:29 +00:00

test_kv_connector_lifecycle.py

[Spec Decode] Support hybrid attention models in extract_hidden_states (#39949)

2026-05-13 10:45:53 -07:00

test_kv_load_failure_recovery.py

[Perf] Optimize scheduler overhead for PD disaggregation, around 5% E2E perf improvement (#35781)

2026-03-10 21:25:30 -07:00

test_lmcache_connector.py

[Prefix Cache] Include lora_name in BlockStored event for deterministic KV-cache reconstruction (#27577)

2025-12-30 00:17:16 +00:00

test_lmcache_integration.py

[V0 Deprecation] Deprecate virtual engine (#37195)

2026-03-18 14:30:14 -07:00

test_mooncake_connector_hma.py

PD disagg with Mooncake Connector: GDN support (Qwen3.5) and MLA support (Deepseek-V4-Flash) (#46807)

2026-06-29 23:29:04 -07:00

test_mooncake_connector_hybrid_mamba.py

PD disagg with Mooncake Connector: GDN support (Qwen3.5) and MLA support (Deepseek-V4-Flash) (#46807)

2026-06-29 23:29:04 -07:00

test_mooncake_connector.py

[2/N][KV-Cache Layout Refactor] Pack K/V into the content dim across attention backends (#44455)

2026-07-11 11:11:16 -04:00

test_mooncake_stats.py

[P/D][Mooncake] Add KVConnectorStats for transfer observability (#40414)

2026-05-05 02:17:38 -07:00

test_mooncake_store_connector.py

[MooncakeStore] Re-derive full external hits on stored boundaries (#49481)

2026-07-23 18:12:28 +08:00

test_mooncake_store_coordinator.py

[Test] Regression test for hybrid-Mamba eagle cache-peek in Mooncake connector (#43559) (#48361)

2026-07-27 19:03:58 +00:00

test_mooncake_store_hma_e2e.py

[3/N][Core][KV Connector] Support reliable partial-tail KV offload for sub-block prompts (#49502)

2026-07-26 23:22:17 -07:00

test_mooncake_store_prepare_values.py

[Perf][KVConnector][Mooncake] Vectorize prepare_value on the KV load path (#48531)

2026-07-23 18:12:00 +08:00

test_mooncake_store_scheduler.py

[3/N][Core][KV Connector] Support reliable partial-tail KV offload for sub-block prompts (#49502)

2026-07-26 23:22:17 -07:00

test_mooncake_store_worker.py

[3/N][Core][KV Connector] Support reliable partial-tail KV offload for sub-block prompts (#49502)

2026-07-26 23:22:17 -07:00

test_moriio_connector.py

[ROCm][P/D] Support MoRIIO heterogeneous TP fan-in (#46332)

2026-06-23 10:33:23 +00:00

test_moriio_kv_layout.py

[Bugfix][KV-transfer] MoRIIO: retry RDMA send-queue-full backpressure instead of failing the read (#47495)

2026-07-16 20:02:27 +00:00

test_moriio_proxy_routing.py

[Bugfix] MoRIIO toy P/D proxy: fix DP-rank index aliasing + harden for high-concurrency bursts (#46115)

2026-07-17 12:35:04 -07:00

test_moriio_routing_fairness.py

[Core][KV-transfer] MoRIIO: heterogeneous TP<->DP prefill/decode read routing (#46116)

2026-07-27 02:10:00 +00:00

test_moriio_toy_proxy_server.py

[Bugfix] MoRIIO toy P/D proxy: add /health (#45222)

2026-07-14 22:56:46 +00:00

test_moriio_tp_ack.py

[Bugfix][KV-transfer] MoRIIO: retry RDMA send-queue-full backpressure instead of failing the read (#47495)

2026-07-16 20:02:27 +00:00

test_multi_connector.py

fix flaky multi example connector consistency (#48206)

2026-07-15 09:20:34 +02:00

test_nixl_connector_hma.py

[KV Connector] Fix NIXL mamba state pairing for multi-slot block tables (#50153)

2026-07-29 09:45:53 -07:00

test_nixl_connector.py

[PD][Bugfix] Rebase KV lease deadlines onto worker clock (#50326)

2026-07-30 01:25:42 +00:00

test_nixl_desc_geometry.py

[KV Connector] Support NIXL P/D for hybrid MLA+SSM models (#49762)

2026-07-29 11:50:30 +08:00

test_nixl_heartbeat.py

[Nixl][PD] Lease renewal TTL KV blocks on P (#41383)

2026-05-11 09:27:30 +00:00

test_nixl_push_connector.py

[KV Connector] Fix NIXL mamba state pairing for multi-slot block tables (#50153)

2026-07-29 09:45:53 -07:00

test_nixl_rocm_gpu_mem_diag.py

[ROCm] Upgrade NIXL and UCX (#49251)

2026-07-21 22:30:02 -05:00

test_nixl_simple_cpu_offload.py

[KV Connector]: Support KV push from Prefill to Decode node using Nixl KV Connector (#35264)

2026-06-12 10:38:41 +00:00

test_offloading_connector.py

[KV Offload] Define clean backend configuration boundary (#48150)

2026-07-16 13:27:05 +03:00

test_output_aggregator.py

[Core] Simplify async KV output aggregation (#28327)

2025-11-09 09:44:13 -08:00

test_remote_decode_lifecycle.py

[Core][KVConnector] Support HMA+NixlConnector (#35758)

2026-03-06 08:51:21 +01:00

test_remote_prefill_lifecycle.py

[ModelRunner V2] Enable by default for all dense models (#44443)

2026-07-02 18:48:57 +08:00

test_scheduler_kv_connector_override.py

[KV Connector] Remove compat support for pre-v0.12.0 constructor signatures without KVCacheConfig (#39832)

2026-05-09 23:39:46 +00:00

test_tp_mapping.py

[KV Connector] Support NIXL heterogeneous P/D block sizes for hybrid models (#49612)

2026-07-28 06:45:02 -07:00

test_transfer_topology_sharded.py

[2/N][KV-Cache Layout Refactor] Pack K/V into the content dim across attention backends (#44455)

2026-07-11 11:11:16 -04:00

utils.py

[Bugfix] Prefix-cache metrics double-counted when a KV connector defers requests (#48860)

2026-07-21 15:02:03 +01:00