# Copyright 2023-2024 SGLang Team # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # ============================================================================== """A scheduler that manages a tensor parallel GPU worker.""" import faulthandler import logging import os import signal import sys import threading import time from collections import deque from concurrent import futures from dataclasses import dataclass from http import HTTPStatus from typing import Deque, Dict, List, Optional, Tuple, Union import psutil import setproctitle import torch import torch.distributed import zmq from torch.cuda import Stream as CudaStream from torch.cuda import StreamContext as CudaStreamContext from torch.distributed import barrier from sglang.srt.configs.model_config import ModelConfig from sglang.srt.constrained.base_grammar_backend import ( INVALID_GRAMMAR_OBJ, create_grammar_backend, ) from sglang.srt.disaggregation.decode import ( DecodePreallocQueue, DecodeTransferQueue, SchedulerDisaggregationDecodeMixin, ) from sglang.srt.disaggregation.decode_kvcache_offload_manager import ( DecodeKVCacheOffloadManager, ) from sglang.srt.disaggregation.prefill import ( PrefillBootstrapQueue, SchedulerDisaggregationPrefillMixin, ) from sglang.srt.disaggregation.utils import ( DisaggregationMode, MetadataBuffers, ReqToMetadataIdxAllocator, TransferBackend, prepare_abort, ) from sglang.srt.distributed import get_pp_group, get_world_group from sglang.srt.environ import envs from sglang.srt.eplb.expert_distribution import get_global_expert_distribution_recorder from sglang.srt.layers.dp_attention import compute_dp_attention_world_info from sglang.srt.layers.moe import initialize_moe_config from sglang.srt.managers.io_struct import ( AbortReq, BaseBatchReq, BaseReq, BatchTokenizedEmbeddingReqInput, BatchTokenizedGenerateReqInput, ClearHiCacheReqInput, ClearHiCacheReqOutput, CloseSessionReqInput, DestroyWeightsUpdateGroupReqInput, ExpertDistributionReq, ExpertDistributionReqOutput, ExpertDistributionReqType, FlushCacheReqInput, FlushCacheReqOutput, FreezeGCReq, GetInternalStateReq, GetInternalStateReqOutput, GetLoadReqInput, GetLoadReqOutput, GetWeightsByNameReqInput, HealthCheckOutput, InitWeightsSendGroupForRemoteInstanceReqInput, InitWeightsSendGroupForRemoteInstanceReqOutput, InitWeightsUpdateGroupReqInput, LoadLoRAAdapterReqInput, LoadLoRAAdapterReqOutput, OpenSessionReqInput, OpenSessionReqOutput, ProfileReq, ReleaseMemoryOccupationReqInput, ResumeMemoryOccupationReqInput, RpcReqInput, RpcReqOutput, SendWeightsToRemoteInstanceReqInput, SendWeightsToRemoteInstanceReqOutput, SetInternalStateReq, SetInternalStateReqOutput, SlowDownReqInput, SlowDownReqOutput, TokenizedEmbeddingReqInput, TokenizedGenerateReqInput, UnloadLoRAAdapterReqInput, UnloadLoRAAdapterReqOutput, UpdateWeightFromDiskReqInput, UpdateWeightsFromDistributedReqInput, UpdateWeightsFromIPCReqInput, UpdateWeightsFromTensorReqInput, ) from sglang.srt.managers.mm_utils import init_embedding_cache from sglang.srt.managers.overlap_utils import FutureMap from sglang.srt.managers.schedule_batch import ( FINISH_ABORT, ModelWorkerBatch, MultimodalInputs, Req, RequestStage, ScheduleBatch, ) from sglang.srt.managers.schedule_policy import ( AddReqResult, PrefillAdder, SchedulePolicy, ) from sglang.srt.managers.scheduler_input_blocker import SchedulerInputBlocker from sglang.srt.managers.scheduler_metrics_mixin import ( RECORD_STEP_TIME, SchedulerMetricsMixin, ) from sglang.srt.managers.scheduler_output_processor_mixin import ( SchedulerOutputProcessorMixin, ) from sglang.srt.managers.scheduler_pp_mixin import SchedulerPPMixin from sglang.srt.managers.scheduler_profiler_mixin import SchedulerProfilerMixin from sglang.srt.managers.scheduler_recv_skipper import SchedulerRecvSkipper from sglang.srt.managers.scheduler_runtime_checker_mixin import ( SchedulerRuntimeCheckerMixin, ) from sglang.srt.managers.scheduler_update_weights_mixin import ( SchedulerUpdateWeightsMixin, ) from sglang.srt.managers.session_controller import Session from sglang.srt.managers.utils import GenerationBatchResult, validate_input_length from sglang.srt.mem_cache.chunk_cache import ChunkCache, SWAChunkCache from sglang.srt.mem_cache.hiradix_cache import HiRadixCache from sglang.srt.mem_cache.mamba_radix_cache import MambaRadixCache from sglang.srt.mem_cache.radix_cache import RadixCache from sglang.srt.mem_cache.swa_radix_cache import SWARadixCache from sglang.srt.parser.reasoning_parser import ReasoningParser from sglang.srt.server_args import PortArgs, ServerArgs, get_global_server_args from sglang.srt.speculative.spec_info import SpeculativeAlgorithm from sglang.srt.tracing.trace import ( process_tracing_init, trace_event_batch, trace_set_proc_propagate_context, trace_set_thread_info, trace_slice_batch, trace_slice_end, trace_slice_start, ) from sglang.srt.two_batch_overlap import TboDPAttentionPreparer from sglang.srt.utils import ( DynamicGradMode, broadcast_pyobj, configure_gc_logger, configure_logger, disable_request_logging, freeze_gc, get_available_gpu_memory, get_bool_env_var, get_int_env_var, get_zmq_socket, kill_itself_when_parent_died, numa_bind_to_node, point_to_point_pyobj, pyspy_dump_schedulers, require_mlp_sync, require_mlp_tp_gather, set_gpu_proc_affinity, set_random_seed, suppress_other_loggers, ) from sglang.srt.utils.hf_transformers_utils import ( get_processor, get_tokenizer, get_tokenizer_from_processor, ) from sglang.srt.utils.torch_memory_saver_adapter import TorchMemorySaverAdapter from sglang.utils import TypeBasedDispatcher, get_exception_traceback logger = logging.getLogger(__name__) # Test retract decode for debugging purposes TEST_RETRACT = envs.SGLANG_TEST_RETRACT.get() TEST_RETRACT_INTERVAL = envs.SGLANG_TEST_RETRACT_INTERVAL.get() TEST_RETRACT_NO_PREFILL_BS = envs.SGLANG_TEST_RETRACT_NO_PREFILL_BS.get() GRAMMAR_TIMEOUT = float(os.environ.get("SGLANG_GRAMMAR_TIMEOUT", 300)) @dataclass class EmbeddingBatchResult: embeddings: torch.Tensor class Scheduler( SchedulerOutputProcessorMixin, SchedulerUpdateWeightsMixin, SchedulerProfilerMixin, SchedulerMetricsMixin, SchedulerDisaggregationDecodeMixin, SchedulerDisaggregationPrefillMixin, SchedulerRuntimeCheckerMixin, SchedulerPPMixin, ): """A scheduler that manages a tensor parallel GPU worker.""" def __init__( self, server_args: ServerArgs, port_args: PortArgs, gpu_id: int, tp_rank: int, moe_ep_rank: int, pp_rank: int, dp_rank: Optional[int], ): # Parse args self.server_args = server_args self.tp_rank = tp_rank self.moe_ep_rank = moe_ep_rank self.pp_rank = pp_rank self.dp_rank = dp_rank self.tp_size = server_args.tp_size self.moe_ep_size = server_args.ep_size self.pp_size = server_args.pp_size self.dp_size = server_args.dp_size self.schedule_policy = server_args.schedule_policy self.enable_priority_scheduling = server_args.enable_priority_scheduling self.abort_on_priority_when_disabled = ( server_args.abort_on_priority_when_disabled ) self.schedule_low_priority_values_first = ( server_args.schedule_low_priority_values_first ) self.priority_scheduling_preemption_threshold = ( server_args.priority_scheduling_preemption_threshold ) self.enable_lora = server_args.enable_lora self.max_loras_per_batch = server_args.max_loras_per_batch self.enable_overlap = not server_args.disable_overlap_schedule self.skip_tokenizer_init = server_args.skip_tokenizer_init self.enable_metrics = server_args.enable_metrics self.enable_metrics_for_all_schedulers = ( server_args.enable_metrics_for_all_schedulers ) self.enable_kv_cache_events = bool( server_args.kv_events_config and tp_rank == 0 ) self.enable_trace = server_args.enable_trace self.stream_interval = server_args.stream_interval self.spec_algorithm = SpeculativeAlgorithm.from_string( server_args.speculative_algorithm ) self.gpu_id = gpu_id self.enable_hierarchical_cache = server_args.enable_hierarchical_cache self.enable_hicache_storage = server_args.hicache_storage_backend is not None self.page_size = server_args.page_size self.attn_tp_rank, self.attn_tp_size, self.attn_dp_rank = ( compute_dp_attention_world_info( server_args.enable_dp_attention, self.tp_rank, self.tp_size, self.dp_size, ) ) # Init model config self.model_config = ModelConfig.from_server_args(server_args) # Init inter-process communication self.init_sockets(server_args, port_args) # Init tokenizer self.init_tokenizer() # Init moe config self.init_moe_config() # Set reasoning_parser and think_end_id if --reasoning_parser is enabled if self.server_args.reasoning_parser and self.tokenizer: reasoning_parser = ReasoningParser( model_type=self.server_args.reasoning_parser, stream_reasoning=False ) self.tokenizer.think_end_id = self.tokenizer.encode( reasoning_parser.detector.think_end_token, add_special_tokens=False )[0] # Check whether overlap can be enabled if not self.is_generation: self.enable_overlap = False logger.info("Overlap scheduler is disabled for embedding models.") # Launch a tensor parallel worker from sglang.srt.managers.tp_worker import TpModelWorker self.tp_worker = TpModelWorker( server_args=server_args, gpu_id=gpu_id, tp_rank=tp_rank, moe_ep_rank=moe_ep_rank, pp_rank=pp_rank, dp_rank=dp_rank, nccl_port=port_args.nccl_port, ) # Launch a draft worker for speculative decoding draft_worker_kwargs = dict( gpu_id=gpu_id, tp_rank=tp_rank, moe_ep_rank=moe_ep_rank, server_args=server_args, nccl_port=port_args.nccl_port, target_worker=self.tp_worker, dp_rank=dp_rank, ) if server_args.speculative_draft_load_format is not None: server_args.load_format = server_args.speculative_draft_load_format logger.info( f"Using draft model load_format: '{server_args.speculative_draft_load_format}'" ) # Draft workers are looked up via `SpeculativeAlgorithm` registry; new # algorithms should register their factory instead of patching this code. if self.spec_algorithm.name in {"EAGLE", "EAGLE3"}: draft_worker_kwargs["enable_overlap"] = self.enable_overlap self.draft_worker = self.spec_algorithm.create_draft_worker( **draft_worker_kwargs ) # Dispatch the model worker if self.spec_algorithm.is_none(): self.model_worker = self.tp_worker else: self.model_worker = self.draft_worker # Get token and memory info from the model worker ( self.max_total_num_tokens, self.max_prefill_tokens, self.max_running_requests, self.max_queued_requests, self.max_req_len, self.max_req_input_len, self.random_seed, self.device, _, _, _, ) = self.tp_worker.get_worker_info() if get_global_server_args().pp_max_micro_batch_size is None: get_global_server_args().pp_max_micro_batch_size = max( self.max_running_requests // server_args.pp_size, 1 ) self.tp_group = self.tp_worker.get_tp_group() self.tp_cpu_group = self.tp_group.cpu_group self.attn_tp_group = self.tp_worker.get_attention_tp_group() self.attn_tp_cpu_group = self.tp_worker.get_attention_tp_cpu_group() self.pp_group = get_pp_group() self.world_group = get_world_group() # With DP attention enabled, the entry rank is attn_tp_rank==0; # otherwise the entry rank is TP group local rank 0. # For #11910, use the CPU communication group to broadcast VLM Python objects, # avoiding any coupling with CUDA streams/devices. if self.server_args.enable_dp_attention: self.cpu_group = self.attn_tp_cpu_group self.is_entry_rank = self.attn_tp_rank == 0 else: self.cpu_group = self.tp_cpu_group self.is_entry_rank = self.tp_group.rank == 0 self.pad_input_ids_func = self.tp_worker.get_pad_input_ids_func() set_random_seed(self.random_seed) # Hybrid memory pool self.is_hybrid = self.tp_worker.is_hybrid self.is_hybrid_gdn = self.tp_worker.model_runner.hybrid_gdn_config is not None if self.is_hybrid: self.sliding_window_size = self.tp_worker.sliding_window_size self.full_tokens_per_layer, self.swa_tokens_per_layer = ( self.tp_worker.get_tokens_per_layer_info() ) # Print debug info if tp_rank == 0: avail_mem = get_available_gpu_memory( self.device, self.gpu_id, empty_cache=False ) logger.info( f"max_total_num_tokens={self.max_total_num_tokens}, " f"chunked_prefill_size={server_args.chunked_prefill_size}, " f"max_prefill_tokens={self.max_prefill_tokens}, " f"max_running_requests={self.max_running_requests}, " f"context_len={self.model_config.context_len}, " f"{'available_cpu_mem' if self.device == 'cpu' else 'available_gpu_mem'}={avail_mem:.2f} GB" ) # Init memory pool and cache self.init_memory_pool_and_cache() # Init running status self.waiting_queue: List[Req] = [] # The running decoding batch for continuous batching self.running_batch: ScheduleBatch = ScheduleBatch(reqs=[], batch_is_full=False) # The current forward batch self.cur_batch: Optional[ScheduleBatch] = None # The last forward batch self.last_batch: Optional[ScheduleBatch] = None self.forward_ct = 0 self.forward_ct_decode = 0 self.num_generated_tokens = 0 self.last_prefill_tokens = 0 self.last_decode_stats_tic = time.perf_counter() self.last_prefill_stats_tic = time.perf_counter() self.return_health_check_ct = 0 self.num_retracted_reqs: int = 0 self.num_paused_reqs: int = 0 self.kv_transfer_speed_gb_s: float = 0.0 self.kv_transfer_latency_ms: float = 0.0 self.sessions: Dict[str, Session] = {} self.default_stream: CudaStream = torch.get_device_module( self.device ).current_stream() if self.device == "cpu": self.default_stream.synchronize = lambda: None # No-op for CPU self.forward_sleep_time = None # Init chunked prefill self.chunked_prefill_size = server_args.chunked_prefill_size if self.chunked_prefill_size <= 0: # -1 means disable self.chunked_prefill_size = None self.chunked_req = None self.is_mixed_chunk = ( self.chunked_prefill_size is not None and server_args.enable_mixed_chunk ) # Init the grammar backend for constrained generation self.grammar_queue: List[Req] = [] if not server_args.skip_tokenizer_init: self.grammar_backend = create_grammar_backend( server_args, self.tokenizer, self.model_config.vocab_size, self.model_config.hf_eos_token_id, ) else: self.grammar_backend = None # Init schedule policy and new token estimation self.policy = SchedulePolicy( self.schedule_policy, self.tree_cache, self.enable_hierarchical_cache, self.enable_priority_scheduling, self.schedule_low_priority_values_first, ) # Enable preemption for priority scheduling. self.try_preemption = self.enable_priority_scheduling assert ( server_args.schedule_conservativeness >= 0 ), "Invalid schedule_conservativeness" self.init_new_token_ratio = min( envs.SGLANG_INIT_NEW_TOKEN_RATIO.get() * server_args.schedule_conservativeness, 1.0, ) self.min_new_token_ratio = min( self.init_new_token_ratio * envs.SGLANG_MIN_NEW_TOKEN_RATIO_FACTOR.get(), 1.0, ) self.new_token_ratio_decay = ( self.init_new_token_ratio - self.min_new_token_ratio ) / envs.SGLANG_NEW_TOKEN_RATIO_DECAY_STEPS.get() self.new_token_ratio = self.init_new_token_ratio # Init watchdog thread self.watchdog_timeout = server_args.watchdog_timeout t = threading.Thread(target=self.watchdog_thread, daemon=True) t.start() self.parent_process = psutil.Process().parent() # Init memory saver, profiler and metric stats self.memory_saver_adapter = TorchMemorySaverAdapter.create( enable=server_args.enable_memory_saver ) self.offload_tags = set() self.init_profiler() self.recv_skipper = SchedulerRecvSkipper.maybe_create(server_args) self.input_blocker = ( SchedulerInputBlocker(noop=self.attn_tp_rank != 0) if get_bool_env_var("SGLANG_ENABLE_COLOCATED_BATCH_GEN") else None ) # Init metrics stats self.init_metrics(tp_rank, pp_rank, dp_rank) if self.enable_kv_cache_events: self.init_kv_events(server_args.kv_events_config) # Init disaggregation self.disaggregation_mode = DisaggregationMode( self.server_args.disaggregation_mode ) self.init_disaggregation() if envs.SGLANG_LOG_GC.get(): configure_gc_logger() # Init prefill kv split size when deterministic inference is enabled with various attention backends self.init_deterministic_inference_config() # Init overlap self.init_overlap() # Init request dispatcher self._request_dispatcher = TypeBasedDispatcher( [ (TokenizedGenerateReqInput, self.handle_generate_request), (TokenizedEmbeddingReqInput, self.handle_embedding_request), (BatchTokenizedGenerateReqInput, self.handle_batch_generate_request), (BatchTokenizedEmbeddingReqInput, self.handle_batch_embedding_request), (FlushCacheReqInput, self.flush_cache_wrapped), (ClearHiCacheReqInput, self.clear_hicache_storage_wrapped), (AbortReq, self.abort_request), (OpenSessionReqInput, self.open_session), (CloseSessionReqInput, self.close_session), (UpdateWeightFromDiskReqInput, self.update_weights_from_disk), (InitWeightsUpdateGroupReqInput, self.init_weights_update_group), (DestroyWeightsUpdateGroupReqInput, self.destroy_weights_update_group), ( InitWeightsSendGroupForRemoteInstanceReqInput, self.init_weights_send_group_for_remote_instance, ), ( SendWeightsToRemoteInstanceReqInput, self.send_weights_to_remote_instance, ), ( UpdateWeightsFromDistributedReqInput, self.update_weights_from_distributed, ), (UpdateWeightsFromTensorReqInput, self.update_weights_from_tensor), (UpdateWeightsFromIPCReqInput, self.update_weights_from_ipc), (GetWeightsByNameReqInput, self.get_weights_by_name), (ReleaseMemoryOccupationReqInput, self.release_memory_occupation), (ResumeMemoryOccupationReqInput, self.resume_memory_occupation), (SlowDownReqInput, self.slow_down), (ProfileReq, self.profile), (FreezeGCReq, self.handle_freeze_gc), (GetInternalStateReq, self.get_internal_state), (SetInternalStateReq, self.set_internal_state), (RpcReqInput, self.handle_rpc_request), (ExpertDistributionReq, self.expert_distribution_handle), (LoadLoRAAdapterReqInput, self.load_lora_adapter), (UnloadLoRAAdapterReqInput, self.unload_lora_adapter), (GetLoadReqInput, self.get_load), ] ) def init_sockets(self, server_args: ServerArgs, port_args: PortArgs): context = zmq.Context(2) self.idle_sleeper = None class SenderWrapper: def __init__(self, socket: zmq.Socket): self.socket = socket def send_output( self, output: Union[BaseReq, BaseBatchReq], recv_obj: Optional[Union[BaseReq, BaseBatchReq]] = None, ): if self.socket is None: return if ( isinstance(recv_obj, BaseReq) and recv_obj.http_worker_ipc is not None and output.http_worker_ipc is None ): # handle communicator reqs for multi-http worker case output.http_worker_ipc = recv_obj.http_worker_ipc self.socket.send_pyobj(output) if self.pp_rank == 0 and self.attn_tp_rank == 0: self.recv_from_tokenizer = get_zmq_socket( context, zmq.PULL, port_args.scheduler_input_ipc_name, False ) self.recv_from_rpc = get_zmq_socket( context, zmq.DEALER, port_args.rpc_ipc_name, False ) send_to_tokenizer = get_zmq_socket( context, zmq.PUSH, port_args.tokenizer_ipc_name, False ) if server_args.skip_tokenizer_init: # Directly send to the TokenizerManager send_to_detokenizer = get_zmq_socket( context, zmq.PUSH, port_args.tokenizer_ipc_name, False ) else: # Send to the DetokenizerManager send_to_detokenizer = get_zmq_socket( context, zmq.PUSH, port_args.detokenizer_ipc_name, False ) self.send_to_tokenizer = SenderWrapper(send_to_tokenizer) self.send_to_detokenizer = SenderWrapper(send_to_detokenizer) if self.server_args.sleep_on_idle: self.idle_sleeper = IdleSleeper( [ self.recv_from_tokenizer, self.recv_from_rpc, ] ) else: self.recv_from_tokenizer = None self.recv_from_rpc = None self.send_to_tokenizer = SenderWrapper(None) self.send_to_detokenizer = SenderWrapper(None) if self.current_scheduler_metrics_enabled(): self.send_metrics_from_scheduler = get_zmq_socket( context, zmq.PUSH, port_args.metrics_ipc_name, False ) def init_deterministic_inference_config(self): """Initialize deterministic inference configuration for different attention backends.""" if not self.server_args.enable_deterministic_inference: self.truncation_align_size = None return backend_sizes = { "flashinfer": ("SGLANG_FLASHINFER_PREFILL_SPLIT_TILE_SIZE", 4096), "triton": ("SGLANG_TRITON_PREFILL_TRUNCATION_ALIGN_SIZE", 4096), } env_var, default_size = backend_sizes.get( self.server_args.attention_backend, (None, None) ) self.truncation_align_size = ( get_int_env_var(env_var, default_size) if env_var else None ) def init_tokenizer(self): server_args = self.server_args self.is_generation = self.model_config.is_generation if server_args.skip_tokenizer_init: self.tokenizer = self.processor = None else: if self.model_config.is_multimodal: self.processor = get_processor( server_args.tokenizer_path, tokenizer_mode=server_args.tokenizer_mode, trust_remote_code=server_args.trust_remote_code, revision=server_args.revision, use_fast=not server_args.disable_fast_image_processor, ) self.tokenizer = get_tokenizer_from_processor(self.processor) else: self.tokenizer = get_tokenizer( server_args.tokenizer_path, tokenizer_mode=server_args.tokenizer_mode, trust_remote_code=server_args.trust_remote_code, revision=server_args.revision, ) def init_memory_pool_and_cache(self): server_args = self.server_args self.req_to_token_pool, self.token_to_kv_pool_allocator = ( self.tp_worker.get_memory_pool() ) if ( server_args.chunked_prefill_size is not None and server_args.disable_radix_cache ): if self.is_hybrid: ChunkCacheClass = SWAChunkCache else: ChunkCacheClass = ChunkCache self.tree_cache = ChunkCacheClass( req_to_token_pool=self.req_to_token_pool, token_to_kv_pool_allocator=self.token_to_kv_pool_allocator, page_size=self.page_size, ) else: if os.environ.get("SGLANG_EXPERIMENTAL_CPP_RADIX_TREE") == "1": # lazy import to avoid JIT overhead from sglang.srt.mem_cache.radix_cache_cpp import RadixCacheCpp self.tree_cache = RadixCacheCpp( disable=False, use_hicache=self.enable_hierarchical_cache, req_to_token_pool=self.req_to_token_pool, token_to_kv_pool=self.token_to_kv_pool_allocator, tp_cache_group=self.tp_cpu_group, page_size=self.page_size, hicache_ratio=server_args.hicache_ratio, hicache_size=server_args.hicache_size, hicache_write_policy=server_args.hicache_write_policy, enable_kv_cache_events=self.enable_kv_cache_events, ) elif self.enable_hierarchical_cache: self.tree_cache = HiRadixCache( req_to_token_pool=self.req_to_token_pool, token_to_kv_pool_allocator=self.token_to_kv_pool_allocator, tp_cache_group=( self.attn_tp_cpu_group if self.server_args.enable_dp_attention else self.tp_cpu_group ), page_size=self.page_size, eviction_policy=server_args.radix_eviction_policy, hicache_ratio=server_args.hicache_ratio, hicache_size=server_args.hicache_size, hicache_write_policy=server_args.hicache_write_policy, hicache_io_backend=server_args.hicache_io_backend, hicache_mem_layout=server_args.hicache_mem_layout, enable_metrics=self.enable_metrics, hicache_storage_backend=server_args.hicache_storage_backend, hicache_storage_prefetch_policy=server_args.hicache_storage_prefetch_policy, model_name=server_args.served_model_name, storage_backend_extra_config=server_args.hicache_storage_backend_extra_config, is_eagle=self.spec_algorithm.is_eagle(), ) self.tp_worker.register_hicache_layer_transfer_counter( self.tree_cache.cache_controller.layer_done_counter ) elif self.is_hybrid: self.tree_cache = SWARadixCache( req_to_token_pool=self.req_to_token_pool, token_to_kv_pool_allocator=self.token_to_kv_pool_allocator, sliding_window_size=self.sliding_window_size, page_size=self.page_size, disable=server_args.disable_radix_cache, is_eagle=self.spec_algorithm.is_eagle(), ) elif self.is_hybrid_gdn: self.tree_cache = MambaRadixCache( req_to_token_pool=self.req_to_token_pool, token_to_kv_pool_allocator=self.token_to_kv_pool_allocator, page_size=self.page_size, disable=server_args.disable_radix_cache, ) elif server_args.enable_lmcache: from sglang.srt.mem_cache.storage.lmcache.lmc_radix_cache import ( LMCRadixCache, ) self.tree_cache = LMCRadixCache( req_to_token_pool=self.req_to_token_pool, token_to_kv_pool_allocator=self.token_to_kv_pool_allocator, page_size=self.page_size, disable=server_args.disable_radix_cache, model_config=self.model_config, tp_size=self.tp_size, rank=self.tp_rank, tp_group=self.tp_group, eviction_policy=server_args.radix_eviction_policy, ) else: self.tree_cache = RadixCache( req_to_token_pool=self.req_to_token_pool, token_to_kv_pool_allocator=self.token_to_kv_pool_allocator, page_size=self.page_size, disable=server_args.disable_radix_cache, enable_kv_cache_events=self.enable_kv_cache_events, eviction_policy=server_args.radix_eviction_policy, is_eagle=self.spec_algorithm.is_eagle(), ) if ( server_args.disaggregation_mode == "decode" and server_args.disaggregation_decode_enable_offload_kvcache ): self.decode_offload_manager = DecodeKVCacheOffloadManager( req_to_token_pool=self.req_to_token_pool, token_to_kv_pool_allocator=self.token_to_kv_pool_allocator, tp_group=( self.attn_tp_cpu_group if self.server_args.enable_dp_attention else self.tp_cpu_group ), tree_cache=self.tree_cache, server_args=self.server_args, ) else: self.decode_offload_manager = None self.decode_mem_cache_buf_multiplier = ( 1 if self.spec_algorithm.is_none() else ( server_args.speculative_num_draft_tokens + ( (server_args.speculative_eagle_topk or 1) * (server_args.speculative_num_steps or 1) ) ) ) embedding_cache_size = int(os.environ.get("SGLANG_VLM_CACHE_SIZE_MB", "100")) init_embedding_cache(embedding_cache_size * 1024 * 1024) def init_disaggregation(self): self.transfer_backend = TransferBackend( self.server_args.disaggregation_transfer_backend ) if ( self.disaggregation_mode == DisaggregationMode.DECODE ): # *2 for the headroom. buffer_size = (self.req_to_token_pool.size) * 2 self.req_to_metadata_buffer_idx_allocator = ReqToMetadataIdxAllocator( buffer_size ) self.disagg_metadata_buffers = MetadataBuffers( buffer_size, hidden_size=self.model_config.hf_text_config.hidden_size, hidden_states_dtype=self.model_config.dtype, custom_mem_pool=self.token_to_kv_pool_allocator.get_kvcache().maybe_get_custom_mem_pool(), ) # The decode requests polling kv cache self.disagg_decode_transfer_queue = DecodeTransferQueue( gloo_group=self.attn_tp_cpu_group, req_to_metadata_buffer_idx_allocator=self.req_to_metadata_buffer_idx_allocator, tp_rank=self.tp_rank, metadata_buffers=self.disagg_metadata_buffers, scheduler=self, tree_cache=self.tree_cache, ) # The decode requests pending for pre-allocation self.disagg_decode_prealloc_queue = DecodePreallocQueue( req_to_token_pool=self.req_to_token_pool, token_to_kv_pool_allocator=self.token_to_kv_pool_allocator, draft_token_to_kv_pool=( None if self.draft_worker is None or self.spec_algorithm.is_ngram() else self.draft_worker.model_runner.token_to_kv_pool ), req_to_metadata_buffer_idx_allocator=self.req_to_metadata_buffer_idx_allocator, metadata_buffers=self.disagg_metadata_buffers, scheduler=self, transfer_queue=self.disagg_decode_transfer_queue, tree_cache=self.tree_cache, gloo_group=self.attn_tp_cpu_group, tp_rank=self.tp_rank, tp_size=self.tp_size, dp_size=self.server_args.dp_size, gpu_id=self.gpu_id, bootstrap_port=self.server_args.disaggregation_bootstrap_port, max_total_num_tokens=self.max_total_num_tokens, prefill_pp_size=self.server_args.disaggregation_prefill_pp, num_reserved_decode_tokens=self.server_args.num_reserved_decode_tokens, transfer_backend=self.transfer_backend, ) elif self.disaggregation_mode == DisaggregationMode.PREFILL: # *2 for the headroom. buffer_size = self.max_running_requests * 2 self.req_to_metadata_buffer_idx_allocator = ReqToMetadataIdxAllocator( buffer_size ) self.disagg_metadata_buffers = MetadataBuffers( buffer_size, hidden_size=self.model_config.hf_text_config.hidden_size, hidden_states_dtype=self.model_config.dtype, custom_mem_pool=self.token_to_kv_pool_allocator.get_kvcache().maybe_get_custom_mem_pool(), ) self.disagg_prefill_bootstrap_queue = PrefillBootstrapQueue( token_to_kv_pool=self.token_to_kv_pool_allocator.get_kvcache(), draft_token_to_kv_pool=( None if self.draft_worker is None or self.spec_algorithm.is_ngram() else self.draft_worker.model_runner.token_to_kv_pool ), req_to_metadata_buffer_idx_allocator=self.req_to_metadata_buffer_idx_allocator, metadata_buffers=self.disagg_metadata_buffers, tp_rank=self.tp_rank, tp_size=self.tp_size, gpu_id=self.gpu_id, bootstrap_port=self.server_args.disaggregation_bootstrap_port, gloo_group=self.attn_tp_cpu_group, max_total_num_tokens=self.max_total_num_tokens, decode_tp_size=self.server_args.disaggregation_decode_tp, decode_dp_size=self.server_args.disaggregation_decode_dp, scheduler=self, pp_rank=self.pp_rank, pp_size=self.pp_size, transfer_backend=self.transfer_backend, ) # The prefill requests that are in the middle of kv sending self.disagg_prefill_inflight_queue: List[Req] = [] def init_overlap(self): if not self.enable_overlap: return self.forward_stream: CudaStream = torch.get_device_module(self.device).Stream() self.forward_stream_ctx: CudaStreamContext = torch.get_device_module( self.device ).stream(self.forward_stream) self.copy_stream: CudaStream = torch.get_device_module(self.device).Stream() self.copy_stream_ctx: CudaStreamContext = torch.get_device_module( self.device ).stream(self.copy_stream) self.future_map = FutureMap( self.max_running_requests, self.device, self.spec_algorithm ) self.batch_record_buf = [None] * 2 self.batch_record_ct = 0 def record_batch_in_overlap(self, model_worker_batch: ModelWorkerBatch): # FIXME(lsyin): hacky way to keep a reference to avoid GPU tensors being freed by torch GC # NOTE: More Reliable: record all tensors into the forward stream # NOTE: - for all future tensors, we shall always read from future map # - for all non-future tensors (produced only by schedule stream), # we shall keep its reference not being release during all the forwarding pass self.batch_record_ct = (self.batch_record_ct + 1) % 2 self.batch_record_buf[self.batch_record_ct] = model_worker_batch def init_moe_config(self): if hasattr(self.model_config.hf_config, "num_experts_per_tok"): initialize_moe_config(self.server_args) @DynamicGradMode() def event_loop_normal(self): """A normal scheduler loop.""" while True: recv_reqs = self.recv_requests() self.process_input_requests(recv_reqs) batch = self.get_next_batch_to_run() self.cur_batch = batch if batch: result = self.run_batch(batch) self.process_batch_result(batch, result) else: # When the server is idle, do self-check and re-init some states self.self_check_during_idle() self.last_batch = batch @DynamicGradMode() def event_loop_overlap(self): """A scheduler loop that overlaps the CPU processing and GPU computation.""" self.result_queue: Deque[Tuple[ScheduleBatch, GenerationBatchResult]] = deque() while True: recv_reqs = self.recv_requests() self.process_input_requests(recv_reqs) batch = self.get_next_batch_to_run() self.cur_batch = batch batch_result = None if batch: batch_result = self.run_batch(batch) self.result_queue.append((batch.copy(), batch_result)) if self.last_batch: # Process the results of the last batch tmp_batch, tmp_result = self.result_queue.popleft() self.process_batch_result(tmp_batch, tmp_result) elif batch is None: # When the server is idle, do self-check and re-init some states self.self_check_during_idle() self.launch_batch_sample_if_needed(batch_result) self.last_batch = batch if envs.SGLANG_ENABLE_RUNTIME_MEM_LEAK_CHECK.get(): self._check_runtime_mem_leak() def recv_requests(self) -> List[Req]: """Receive results at tp_rank = 0 and broadcast it to all other TP ranks.""" if self.recv_skipper is not None: last_forward_mode = ( self.last_batch.forward_mode if self.last_batch is not None else None ) if not self.recv_skipper.handle(last_forward_mode): return [] if self.pp_rank == 0: if self.attn_tp_rank == 0: recv_reqs = [] while True: try: recv_req = self.recv_from_tokenizer.recv_pyobj(zmq.NOBLOCK) except zmq.ZMQError: break recv_reqs.append(recv_req) while True: try: recv_rpc = self.recv_from_rpc.recv_pyobj(zmq.NOBLOCK) except zmq.ZMQError: break recv_reqs.append(recv_rpc) else: recv_reqs = None else: if self.attn_tp_rank == 0: dp_offset = self.attn_dp_rank * self.attn_tp_size recv_reqs = point_to_point_pyobj( [], self.pp_rank * self.tp_size + dp_offset, self.world_group.device_group, (self.pp_rank - 1) * self.tp_size + dp_offset, self.pp_rank * self.tp_size + dp_offset, ) else: recv_reqs = None if self.input_blocker is not None: recv_reqs = self.input_blocker.handle(recv_reqs) if self.server_args.enable_dp_attention: if self.attn_tp_rank == 0: work_reqs = [ req for req in recv_reqs if isinstance( req, ( TokenizedGenerateReqInput, TokenizedEmbeddingReqInput, BatchTokenizedGenerateReqInput, BatchTokenizedEmbeddingReqInput, ), ) ] control_reqs = [ req for req in recv_reqs if not isinstance( req, ( TokenizedGenerateReqInput, TokenizedEmbeddingReqInput, BatchTokenizedGenerateReqInput, BatchTokenizedEmbeddingReqInput, ), ) ] else: work_reqs = None control_reqs = None if self.attn_tp_size != 1: work_reqs = broadcast_pyobj( work_reqs, self.attn_tp_group.rank, self.attn_tp_cpu_group, src=self.attn_tp_group.ranks[0], ) if self.tp_size != 1: control_reqs = broadcast_pyobj( control_reqs, self.tp_group.rank, self.tp_cpu_group, src=self.tp_group.ranks[0], ) recv_reqs = work_reqs + control_reqs elif self.tp_size != 1: recv_reqs = broadcast_pyobj( recv_reqs, self.tp_group.rank, self.tp_cpu_group, src=self.tp_group.ranks[0], ) if self.enable_trace: for req in recv_reqs: if isinstance( req, (TokenizedGenerateReqInput, TokenizedEmbeddingReqInput) ): trace_set_proc_propagate_context(req.rid, req.trace_context) trace_slice_start("", req.rid, anonymous=True) return recv_reqs def process_input_requests(self, recv_reqs: List): for recv_req in recv_reqs: # If it is a health check generation request and there are running requests, ignore it. if is_health_check_generate_req(recv_req) and ( self.chunked_req is not None or not self.running_batch.is_empty() or len(self.offload_tags) > 0 ): self.return_health_check_ct += 1 continue output = self._request_dispatcher(recv_req) if output is not None: if isinstance(output, RpcReqOutput): if self.recv_from_rpc is not None: self.recv_from_rpc.send_pyobj(output) else: self.send_to_tokenizer.send_output(output, recv_req) def init_req_max_new_tokens(self, req): req.sampling_params.max_new_tokens = min( ( req.sampling_params.max_new_tokens if req.sampling_params.max_new_tokens is not None else 1 << 30 ), self.max_req_len - len(req.origin_input_ids) - 1, ) def _process_and_broadcast_mm_inputs( self, raw_mm_inputs: Optional[dict], ): """Materialize MultimodalInputs once on the entry rank and broadcast to others. Entry rank: - constructs MultimodalInputs.from_dict(raw_mm_inputs) once - broadcasts to other ranks in self.cpu_group (if world_size > 1) Non-entry ranks: - receive the object via broadcast (if world_size > 1) - otherwise (single-rank / no group) fall back to local from_dict Returns: MultimodalInputs | None """ if raw_mm_inputs is None: return None group_world_size = 1 try: if ( torch.distributed.is_available() and torch.distributed.is_initialized() and self.cpu_group is not None ): group_world_size = torch.distributed.get_world_size( group=self.cpu_group ) except Exception as e: logger.warning( f"Failed to get world size in mm_inputs handling with {e}, fallback to 1." ) # In case tp size > 1, all the Scheduler TP ranks runs the duplicated computing # process in CPU which occupies the main thread CPU cycle. This computing logic # merely needs to be run on TP0 and be broadcast to other TP ranks. # Since the Scheduler is single-threaded, any large CPU cost will impact # handling of other messages. For example, CPU hits 99.9% can significantly # increase the CUDA kernel launch time. if self.is_entry_rank: # Only the entry rank materializes once from dict. image_inputs = MultimodalInputs.from_dict(raw_mm_inputs) # Broadcast to other TP ranks (use src=0 within the group). if group_world_size > 1: obj_list = [image_inputs] torch.distributed.broadcast_object_list( obj_list, src=0, group=self.cpu_group ) image_inputs = obj_list[0] else: # Non-entry ranks: receive if group size > 1; otherwise materialize locally. if group_world_size > 1: obj_list = [None] torch.distributed.broadcast_object_list( obj_list, src=0, group=self.cpu_group ) image_inputs = obj_list[0] else: image_inputs = MultimodalInputs.from_dict(raw_mm_inputs) return image_inputs def handle_generate_request( self, recv_req: TokenizedGenerateReqInput, ): # Create a new request if ( recv_req.session_params is None or recv_req.session_params.id is None or recv_req.session_params.id not in self.sessions ): if recv_req.input_embeds is not None: # Generate fake input_ids based on the length of input_embeds seq_length = len(recv_req.input_embeds) fake_input_ids = [1] * seq_length recv_req.input_ids = fake_input_ids if recv_req.bootstrap_port is None: # Use default bootstrap port recv_req.bootstrap_port = self.server_args.disaggregation_bootstrap_port req = Req( recv_req.rid, recv_req.input_text, recv_req.input_ids, recv_req.sampling_params, return_logprob=recv_req.return_logprob, top_logprobs_num=recv_req.top_logprobs_num, token_ids_logprob=recv_req.token_ids_logprob, stream=recv_req.stream, lora_id=recv_req.lora_id, input_embeds=recv_req.input_embeds, custom_logit_processor=recv_req.custom_logit_processor, return_hidden_states=recv_req.return_hidden_states, eos_token_ids=self.model_config.hf_eos_token_id, bootstrap_host=recv_req.bootstrap_host, bootstrap_port=recv_req.bootstrap_port, bootstrap_room=recv_req.bootstrap_room, disagg_mode=self.disaggregation_mode, data_parallel_rank=recv_req.data_parallel_rank, vocab_size=self.model_config.vocab_size, priority=recv_req.priority, metrics_collector=( self.metrics_collector if self.enable_metrics else None ), http_worker_ipc=recv_req.http_worker_ipc, ) req.tokenizer = self.tokenizer if self.disaggregation_mode != DisaggregationMode.NULL: # Invalid request for disaggregated mode if recv_req.bootstrap_room is None: error_msg = ( f"Invalid request: Disaggregated request received without " f"boostrap room id. {req.rid=}" ) logger.error(error_msg) prepare_abort(req, error_msg, status_code=HTTPStatus.BAD_REQUEST) self.stream_output([req], req.return_logprob) return if ( recv_req.session_params is not None and recv_req.session_params.id is not None ): req.set_finish_with_abort( f"Invalid request: session id {recv_req.session_params.id} does not exist" ) self.init_req_max_new_tokens(req) self._add_request_to_queue(req) return else: # Create a new request from a previous session session = self.sessions[recv_req.session_params.id] req = session.create_req(recv_req, self.tokenizer) if isinstance(req.finished_reason, FINISH_ABORT): self.init_req_max_new_tokens(req) self._add_request_to_queue(req) return # Handle multimodal inputs if recv_req.mm_inputs is not None: image_inputs = self._process_and_broadcast_mm_inputs(recv_req.mm_inputs) # The following steps are already fast, execute locally on each rank. # Expand a single image token into multiple dummy tokens for receiving image embeddings req.origin_input_ids = self.pad_input_ids_func( req.origin_input_ids, image_inputs ) req.extend_image_inputs(image_inputs) if len(req.origin_input_ids) >= self.max_req_input_len: req.set_finish_with_abort( error_msg=( "Multimodal prompt is too long after expanding multimodal tokens. " f"After expanding {len(req.origin_input_ids_unpadded)=} => {len(req.origin_input_ids)} >= {self.max_req_input_len}." ) ) self.init_req_max_new_tokens(req) self._add_request_to_queue(req) return # initialize before returning self.init_req_max_new_tokens(req) # Validate prompt length error_msg = validate_input_length( req, self.max_req_input_len, self.server_args.allow_auto_truncate, ) if error_msg: req.set_finish_with_abort(error_msg) self._add_request_to_queue(req) return # Copy more attributes if recv_req.logprob_start_len == -1 or not recv_req.return_logprob: # By default, only return the logprobs for output tokens # For prefill-only requests with logprob_start_len == -1, set logprob_start_len beyond input sequence # to skip input logprob computation entirely if req.is_prefill_only: req.logprob_start_len = len(req.origin_input_ids) else: # TODO: For text generation, evaluate setting logprob_start_len to len(req.origin_input_ids) as well req.logprob_start_len = len(req.origin_input_ids) - 1 else: req.logprob_start_len = recv_req.logprob_start_len if not req.is_prefill_only and req.logprob_start_len >= len( req.origin_input_ids ): error_msg = f"{req.logprob_start_len=} is higher than the number of input tokens {len(req.origin_input_ids)=}. Please use a smaller logprob_start_len." req.logprob_start_len = len(req.origin_input_ids) - 1 req.set_finish_with_abort(error_msg) self._add_request_to_queue(req) return # Init grammar cache for this request add_to_grammar_queue = False if ( req.sampling_params.json_schema is not None or req.sampling_params.regex is not None or req.sampling_params.ebnf is not None or req.sampling_params.structural_tag is not None ): if self.grammar_backend is None: error_msg = "Grammar-based generation (json_schema, regex, ebnf, structural_tag) is not supported when the server is launched with --grammar-backend none" req.set_finish_with_abort(error_msg) else: if req.sampling_params.json_schema is not None: key = ("json", req.sampling_params.json_schema) elif req.sampling_params.regex is not None: key = ("regex", req.sampling_params.regex) elif req.sampling_params.ebnf is not None: key = ("ebnf", req.sampling_params.ebnf) elif req.sampling_params.structural_tag: key = ("structural_tag", req.sampling_params.structural_tag) value, cache_hit = self.grammar_backend.get_cached_or_future_value(key) req.grammar = value if not cache_hit: req.grammar_key = key add_to_grammar_queue = True else: if value is INVALID_GRAMMAR_OBJ: # We hit a cached invalid grammar. error_msg = f"Invalid grammar request with cache hit: {key=}" req.set_finish_with_abort(error_msg) if add_to_grammar_queue: self.grammar_queue.append(req) else: self._add_request_to_queue(req) def handle_batch_generate_request( self, recv_req: BatchTokenizedGenerateReqInput, ): """Handle optimized batch generate request.""" logger.debug(f"Processing batch generate request with {len(recv_req)} requests") # Process each request in the batch for tokenized_req in recv_req: self.handle_generate_request(tokenized_req) def _prefetch_kvcache(self, req: Req): if self.enable_hicache_storage: req.init_next_round_input(self.tree_cache) if req.last_node.backuped: # only to initiate the prefetch if the last node is backuped # otherwise, the allocated GPU memory must be locked for integrity last_hash = req.last_host_node.get_last_hash_value() matched_len = len(req.prefix_indices) + req.host_hit_length new_input_tokens = req.fill_ids[matched_len:] prefix_keys = ( req.last_node.get_prefix_hash_values(req.last_node.parent) if self.tree_cache.hicache_storage_pass_prefix_keys else None ) self.tree_cache.prefetch_from_storage( req.rid, req.last_host_node, new_input_tokens, last_hash, prefix_keys, ) def _add_request_to_queue(self, req: Req, is_retracted: bool = False): if self.disaggregation_mode == DisaggregationMode.NULL: self._set_or_validate_priority(req) if self._abort_on_queued_limit(req): return self._prefetch_kvcache(req) self.waiting_queue.append(req) req.time_stats.wait_queue_entry_time = time.perf_counter() trace_slice_end(RequestStage.REQUEST_PROCESS, req.rid, auto_next_anon=True) elif self.disaggregation_mode == DisaggregationMode.PREFILL: self._prefetch_kvcache(req) self.disagg_prefill_bootstrap_queue.add( req, self.model_config.num_key_value_heads ) req.time_stats.prefill_bootstrap_queue_entry_time = time.perf_counter() elif self.disaggregation_mode == DisaggregationMode.DECODE: self.disagg_decode_prealloc_queue.add(req, is_retracted=is_retracted) if not is_retracted: req.time_stats.decode_prealloc_queue_entry_time = time.perf_counter() else: raise ValueError(f"Invalid {self.disaggregation_mode=}") def _set_or_validate_priority(self, req: Req): """Set the default priority value, or abort the request based on the priority scheduling mode.""" if self.enable_priority_scheduling and req.priority is None: if self.schedule_low_priority_values_first: req.priority = sys.maxsize else: req.priority = -sys.maxsize - 1 elif ( not self.enable_priority_scheduling and req.priority is not None and self.abort_on_priority_when_disabled ): abort_req = AbortReq( finished_reason={ "type": "abort", "status_code": HTTPStatus.SERVICE_UNAVAILABLE, "message": "Using priority is disabled for this server. Please send a new request without a priority.", }, rid=req.rid, ) self.send_to_tokenizer.send_output(abort_req, req) def _abort_on_queued_limit(self, recv_req: Req) -> bool: """Abort an incoming or existing request if the waiting queue is full. Returns True if the incoming request is aborted.""" if ( self.max_queued_requests is None or len(self.waiting_queue) + 1 <= self.max_queued_requests ): return False # Reject the incoming request by default. req_to_abort = recv_req message = "The request queue is full." if self.enable_priority_scheduling: # With priority scheduling, consider aboritng an existing request based on the priority. # direction = 1 => smaller number = higher priority; -1 => larger number = higher priority. # max(...) + (direction * priority, queue_time_start) picks the least-preferred request. # Tie: later queue_time_start (newer) is evicted first. Preempt only if strictly better. direction = 1 if self.schedule_low_priority_values_first else -1 key_fn = lambda item: ( direction * item[1].priority, item[1].time_stats.wait_queue_entry_time, ) idx, candidate_req = max(enumerate(self.waiting_queue), key=key_fn) abort_existing_req = ( direction * recv_req.priority < direction * candidate_req.priority ) if abort_existing_req: self.waiting_queue.pop(idx) req_to_abort = candidate_req message = "The request is aborted by a higher priority request." self.send_to_tokenizer.send_output( AbortReq( finished_reason={ "type": "abort", "status_code": HTTPStatus.SERVICE_UNAVAILABLE, "message": message, }, rid=req_to_abort.rid, ), req_to_abort, ) return req_to_abort.rid == recv_req.rid def handle_embedding_request( self, recv_req: TokenizedEmbeddingReqInput, ): req = Req( recv_req.rid, recv_req.input_text, recv_req.input_ids, recv_req.sampling_params, token_type_ids=recv_req.token_type_ids, priority=recv_req.priority, dimensions=recv_req.dimensions, http_worker_ipc=recv_req.http_worker_ipc, ) req.tokenizer = self.tokenizer # Handle multimodal inputs if recv_req.image_inputs is not None: image_inputs = self._process_and_broadcast_mm_inputs(recv_req.image_inputs) # Expand a single image token into multiple dummy tokens for receiving image embeddings req.origin_input_ids = self.pad_input_ids_func( req.origin_input_ids, image_inputs ) req.extend_image_inputs(image_inputs) if len(req.origin_input_ids) >= self.max_req_input_len: req.set_finish_with_abort( error_msg=( "Multimodal prompt is too long after expanding multimodal tokens. " f"After expanding {len(req.origin_input_ids_unpadded)=} => {len(req.origin_input_ids)} >= {self.max_req_input_len}." ) ) self._add_request_to_queue(req) return # Validate prompts length error_msg = validate_input_length( req, self.max_req_input_len, self.server_args.allow_auto_truncate, ) if error_msg: self._add_request_to_queue(req) return # Copy more attributes req.logprob_start_len = len(req.origin_input_ids) - 1 self._add_request_to_queue(req) def handle_batch_embedding_request( self, recv_req: BatchTokenizedEmbeddingReqInput, ): """Handle optimized batch embedding request.""" logger.debug( f"Processing batch embedding request with {len(recv_req)} requests" ) # Process each request in the batch for tokenized_req in recv_req: self.handle_embedding_request(tokenized_req) def _get_token_info(self): available_size = self.token_to_kv_pool_allocator.available_size() evictable_size = self.tree_cache.evictable_size() num_used = self.max_total_num_tokens - (available_size + evictable_size) token_usage = num_used / self.max_total_num_tokens return num_used, token_usage, available_size, evictable_size def _get_mamba_token_info(self): is_radix_tree = isinstance(self.tree_cache, MambaRadixCache) full_available_size = self.token_to_kv_pool_allocator.available_size() full_evictable_size = ( self.tree_cache.full_evictable_size() if is_radix_tree else 0 ) mamba_available_size = self.req_to_token_pool.mamba_pool.available_size() mamba_evictable_size = ( self.tree_cache.mamba_evictable_size() if is_radix_tree else 0 ) full_num_used = self.token_to_kv_pool_allocator.size - ( full_available_size + full_evictable_size ) mamba_num_used = self.req_to_token_pool.mamba_pool.size - ( mamba_available_size + mamba_evictable_size ) full_token_usage = full_num_used / self.token_to_kv_pool_allocator.size mamba_usage = mamba_num_used / self.req_to_token_pool.mamba_pool.size return ( full_num_used, mamba_num_used, full_token_usage, mamba_usage, full_available_size, full_evictable_size, mamba_available_size, mamba_evictable_size, ) def _get_swa_token_info(self): full_available_size = self.token_to_kv_pool_allocator.full_available_size() full_evictable_size = self.tree_cache.full_evictable_size() swa_available_size = self.token_to_kv_pool_allocator.swa_available_size() swa_evictable_size = self.tree_cache.swa_evictable_size() full_num_used = self.full_tokens_per_layer - ( full_available_size + full_evictable_size ) swa_num_used = self.swa_tokens_per_layer - ( swa_available_size + swa_evictable_size ) full_token_usage = full_num_used / self.full_tokens_per_layer swa_token_usage = swa_num_used / self.swa_tokens_per_layer return ( full_num_used, swa_num_used, full_token_usage, swa_token_usage, full_available_size, full_evictable_size, swa_available_size, swa_evictable_size, ) def get_next_batch_to_run(self) -> Optional[ScheduleBatch]: # Merge the prefill batch into the running batch chunked_req_to_exclude = set() if self.chunked_req: # Move the chunked request out of the batch so that we can merge # only finished requests to running_batch. chunked_req_to_exclude.add(self.chunked_req) self.tree_cache.cache_unfinished_req(self.chunked_req, chunked=True) # chunked request keeps its rid but will get a new req_pool_idx if self.tp_worker.model_runner.mambaish_config is not None: self.req_to_token_pool.free( self.chunked_req.req_pool_idx, free_mamba_cache=False ) else: self.req_to_token_pool.free(self.chunked_req.req_pool_idx) if self.last_batch and self.last_batch.forward_mode.is_extend(): if self.last_batch.chunked_req is not None: # In the context pipeline parallelism, after the last chunk, the current microbatch still track outdated chunked_req. # We need to discard it. chunked_req_to_exclude.add(self.last_batch.chunked_req) # Filter batch last_bs = self.last_batch.batch_size() self.last_batch.filter_batch( chunked_req_to_exclude=list(chunked_req_to_exclude) ) if self.last_batch.batch_size() < last_bs: self.running_batch.batch_is_full = False # Merge the new batch into the running batch. # For prefill-only batch, we can avoid going through decoding step. if not self.last_batch.is_empty() and not self.last_batch.is_prefill_only: if self.running_batch.is_empty(): self.running_batch = self.last_batch else: # Merge running_batch with prefill batch self.running_batch.merge_batch(self.last_batch) new_batch = self.get_new_batch_prefill() need_dp_attn_preparation = require_mlp_sync(self.server_args) if need_dp_attn_preparation and not self.spec_algorithm.is_none(): # In speculative decoding, prefill batches and decode batches cannot be processed in the same DP attention group. # We prepare idle batches in advance to skip preparing decode batches when there are prefill batches in the group. new_batch = self.prepare_mlp_sync_batch(new_batch) need_dp_attn_preparation = new_batch is None if new_batch is not None: # Run prefill first if possible ret = new_batch else: # Run decode if not self.running_batch.is_empty(): self.running_batch = self.update_running_batch(self.running_batch) ret = self.running_batch if not self.running_batch.is_empty() else None else: ret = None # Handle DP attention if need_dp_attn_preparation: ret = self.prepare_mlp_sync_batch(ret) if ret: attrs = {"bid": hex(id(ret)), "batch_size": ret.batch_size()} trace_event_batch("schedule", ret.reqs, attrs=attrs) return ret def get_num_allocatable_reqs(self, running_bs): res = get_global_server_args().pp_max_micro_batch_size - running_bs if self.pp_size > 1: res = min(res, self.req_to_token_pool.available_size()) return res def get_new_batch_prefill(self) -> Optional[ScheduleBatch]: # Check if the grammar is ready in the grammar queue if self.grammar_queue: self.move_ready_grammar_requests() if self.try_preemption: # Reset batch_is_full to try preemption with a prefill adder. self.running_batch.batch_is_full = False # Handle the cases where prefill is not allowed if ( self.running_batch.batch_is_full or len(self.waiting_queue) == 0 ) and self.chunked_req is None: return None running_bs = len(self.running_batch.reqs) # Ignore the check if self.chunked_req is not None. # In the non-PP case, when self.chunked_req is not None, num_allocatable_reqs should always be greater than 0, # as the space for the chunked request has just been released. # In PP case, a chunked req can start in one microbatch and end in another microbatch, so the max_running_requests per microbatch should not be strict. # Instead, we should always allow chunked request to be added, otherwise, there will be a memory leak. if ( self.get_num_allocatable_reqs(running_bs) <= 0 and not self.chunked_req and not self.try_preemption ): self.running_batch.batch_is_full = True return None if self.enable_hierarchical_cache: self.tree_cache.check_hicache_events() # Get priority queue self.policy.calc_priority(self.waiting_queue) if TEST_RETRACT and running_bs > TEST_RETRACT_NO_PREFILL_BS: # If we are testing retraction and the running batch size exceeds # TEST_RETRACT_NO_PREFILL_BS, we skip the prefill to keep the requests # in the waiting queue. return None # Prefill policy adder = PrefillAdder( self.page_size, self.tree_cache, self.token_to_kv_pool_allocator, self.running_batch, self.new_token_ratio, self.max_prefill_tokens, self.chunked_prefill_size, running_bs if self.is_mixed_chunk else 0, self.priority_scheduling_preemption_threshold, ) if self.chunked_req is not None: self.chunked_req.init_next_round_input() self.chunked_req = adder.add_chunked_req(self.chunked_req) if self.enable_lora: lora_set = set([req.lora_id for req in self.running_batch.reqs]) # Get requests from the waiting queue to a new prefill batch for req in self.waiting_queue: if self.enable_lora and not self.tp_worker.can_run_lora_batch( lora_set | set([req.lora_id for req in adder.can_run_list]) | set([req.lora_id]) ): self.running_batch.batch_is_full = True break running_bs = len(self.running_batch.reqs) - len(adder.preempt_list) if len(adder.can_run_list) >= self.get_num_allocatable_reqs(running_bs): self.running_batch.batch_is_full = True if self.disaggregation_mode == DisaggregationMode.PREFILL: # In prefill mode, prealloc queue and transfer queue can also take memory, # so we need to check if the available size for the actual available size. if len(adder.can_run_list) >= self.req_to_token_pool.available_size(): self.running_batch.batch_is_full = True if self.running_batch.batch_is_full: if not self.try_preemption: break if not adder.preempt_to_schedule(req, self.server_args): break if self.enable_hicache_storage: prefetch_done = self.tree_cache.check_prefetch_progress(req.rid) if not prefetch_done: # skip staging requests that are ongoing prefetch continue req.init_next_round_input(self.tree_cache) res = adder.add_one_req( req, has_chunked_req=(self.chunked_req is not None), truncation_align_size=self.truncation_align_size, ) if res != AddReqResult.CONTINUE: if res == AddReqResult.NO_TOKEN: if self.enable_hierarchical_cache: # Set batch_is_full after making sure there are requests that can be served self.running_batch.batch_is_full = len( adder.can_run_list ) > 0 or (not self.running_batch.is_empty()) else: self.running_batch.batch_is_full = True break # Update waiting queue can_run_list: List[Req] = adder.can_run_list if len(can_run_list) == 0: return None if self.enable_metrics: # only record queue time when enable_metrics is True to avoid overhead for req in can_run_list: req.add_latency(RequestStage.PREFILL_WAITING) self.waiting_queue = [ x for x in self.waiting_queue if x not in set(can_run_list) ] if adder.preempt_list: for req in adder.preempt_list: self._add_request_to_queue(req) if adder.new_chunked_req is not None: assert self.chunked_req is None self.chunked_req = adder.new_chunked_req if self.chunked_req: self.chunked_req.is_chunked += 1 # Print stats if self.current_scheduler_metrics_enabled(): self.log_prefill_stats(adder, can_run_list, running_bs, 0) for req in can_run_list: if req.time_stats.forward_entry_time == 0: # Avoid update chunked request many times req.time_stats.forward_entry_time = time.perf_counter() if self.enable_metrics: self.metrics_collector.observe_queue_time( req.time_stats.get_queueing_time(), ) # Create a new batch new_batch = ScheduleBatch.init_new( can_run_list, self.req_to_token_pool, self.token_to_kv_pool_allocator, self.tree_cache, self.model_config, self.enable_overlap, self.spec_algorithm, chunked_req=self.chunked_req, ) if self.enable_hierarchical_cache: # todo (zhiqiang): disable cuda graph execution if hicache loading triggered new_batch.hicache_consumer_index = ( self.tree_cache.ready_to_load_host_cache() ) new_batch.prepare_for_extend() # Mixed-style chunked prefill if ( self.is_mixed_chunk and not self.running_batch.is_empty() and not (new_batch.return_logprob or self.running_batch.return_logprob) ): # TODO (lianmin): support return_logprob + mixed chunked prefill self.running_batch.filter_batch() if not self.running_batch.is_empty(): self.running_batch.prepare_for_decode() new_batch.mix_with_running(self.running_batch) new_batch.decoding_reqs = self.running_batch.reqs self.running_batch = ScheduleBatch( reqs=[], batch_is_full=self.running_batch.batch_is_full ) else: new_batch.decoding_reqs = None return new_batch def update_running_batch(self, batch: ScheduleBatch) -> Optional[ScheduleBatch]: """Update the current running decoding batch.""" initial_bs = batch.batch_size() batch.filter_batch() if batch.is_empty(): batch.batch_is_full = False return batch # Check if decode out of memory if not batch.check_decode_mem(self.decode_mem_cache_buf_multiplier) or ( TEST_RETRACT and self.forward_ct % TEST_RETRACT_INTERVAL == 0 ): old_ratio = self.new_token_ratio retracted_reqs, new_token_ratio, reqs_to_abort = batch.retract_decode( self.server_args ) self.num_retracted_reqs = len(retracted_reqs) self.new_token_ratio = new_token_ratio for req in reqs_to_abort: abort_reason: FINISH_ABORT = req.to_finish self.send_to_tokenizer.send_output( AbortReq(abort_message=abort_reason.message, rid=req.rid), req ) logger.info( "KV cache pool is full. Retract requests. " f"#retracted_reqs: {len(retracted_reqs)}, " f"#new_token_ratio: {old_ratio:.4f} -> {new_token_ratio:.4f}" ) for req in retracted_reqs: self._add_request_to_queue(req, is_retracted=True) else: self.new_token_ratio = max( self.new_token_ratio - self.new_token_ratio_decay, self.min_new_token_ratio, ) if batch.batch_size() < initial_bs: batch.batch_is_full = False # Update batch tensors batch.prepare_for_decode() return batch # placeholder for override def update_cache_from_scheduler( self, schedule_batch: ScheduleBatch, batch_result: GenerationBatchResult ): pass def run_batch( self, batch: ScheduleBatch ) -> Union[GenerationBatchResult, EmbeddingBatchResult]: """Run a batch.""" self.forward_ct += 1 # Whether to run the profiler self._profile_batch_predicate(batch) if self.forward_sleep_time is not None: logger.info(f"Scheduler.run_batch sleep {self.forward_sleep_time}s") time.sleep(self.forward_sleep_time) # Run forward if self.is_generation: batch_or_worker_batch = batch if self.enable_overlap or self.spec_algorithm.is_none(): # FIXME(lsyin): remove this if and finally unify the abstraction batch_or_worker_batch = batch.get_model_worker_batch() if self.enable_overlap: # FIXME: remove this assert assert isinstance(batch_or_worker_batch, ModelWorkerBatch) model_worker_batch = batch_or_worker_batch self.record_batch_in_overlap(model_worker_batch) # Sampling info will be modified during forward model_worker_batch.sampling_info = ( model_worker_batch.sampling_info.copy_for_forward() ) bs = len(model_worker_batch.seq_lens) future_indices = self.future_map.alloc_future_indices(bs) with self.forward_stream_ctx: self.forward_stream.wait_stream(self.default_stream) self.future_map.resolve_future(model_worker_batch) batch_result = self.model_worker.forward_batch_generation( model_worker_batch ) # FIXME(lsyin): maybe move this to forward_batch_generation batch_result.copy_done = torch.get_device_module( self.device ).Event() if batch_result.delay_sample_func is None: self.future_map.store_to_map(future_indices, batch_result) batch_result.copy_to_cpu() else: batch_result.future_indices = future_indices # FIXME(lsyin): move this assignment elsewhere future_indices_or_next_token_ids = -future_indices.indices if batch.is_v2_eagle: # FIXME(lsyin): tmp code for eagle v2 # We only keep future indices for next draft input batch.spec_info = batch_result.next_draft_input batch.spec_info.future_indices = future_indices # batch.spec_info = EagleDraftInput( # future_indices=future_indices, # verify_done=batch_result.next_draft_input.verify_done, # # FIXME(lsyin): remove the allocate_lens in EagleDraftInput # allocate_lens=batch_result.next_draft_input.allocate_lens, # ) # The future value, usually for next batch preparation # Current implementation strictly synchronizes the seq_lens batch.seq_lens = batch_result.next_draft_input.new_seq_lens else: batch_result = self.model_worker.forward_batch_generation( batch_or_worker_batch ) future_indices_or_next_token_ids = batch_result.next_token_ids self.update_cache_from_scheduler(batch, batch_result) # NOTE: future_indices_or_next_token_ids is used in ScheduleBatch, # which can probably be replaced by future_indices later [TODO(lsyin)]. # we shall still keep the original outputs, e.g. next_token_ids # in the GenerationBatchOutput for processing after copy_done. batch.output_ids = future_indices_or_next_token_ids # These 2 values are needed for processing the output, but the values can be # modified by overlap schedule. So we have to copy them here so that # we can use the correct values in output processing. if batch.return_logprob or self.spec_algorithm.is_eagle(): extend_input_len_per_req = [req.extend_input_len for req in batch.reqs] else: extend_input_len_per_req = None if batch.return_logprob: extend_logprob_start_len_per_req = [ req.extend_logprob_start_len for req in batch.reqs ] else: extend_logprob_start_len_per_req = None batch_result.extend_input_len_per_req = extend_input_len_per_req batch_result.extend_logprob_start_len_per_req = ( extend_logprob_start_len_per_req ) return batch_result else: # embedding or reward model model_worker_batch = batch.get_model_worker_batch() embeddings = self.tp_worker.forward_batch_embedding(model_worker_batch) ret = EmbeddingBatchResult(embeddings=embeddings) return ret def launch_batch_sample_if_needed( self, batch_result: GenerationBatchResult ) -> Union[GenerationBatchResult, EmbeddingBatchResult]: # TODO(lsyin): make the delayed sample a default behavior after # unifying the forward_batch_generation interface (related to spec V2). if batch_result is None or batch_result.delay_sample_func is None: return with self.forward_stream_ctx: self.forward_stream.wait_stream(self.default_stream) _batch_result = batch_result.delay_sample_func() assert _batch_result is batch_result self.future_map.store_to_map(batch_result.future_indices, batch_result) batch_result.copy_to_cpu() def process_batch_result( self, batch: ScheduleBatch, result: Union[GenerationBatchResult, EmbeddingBatchResult], ): if batch.forward_mode.is_decode(): self.process_batch_result_decode(batch, result) trace_slice_batch(RequestStage.DECODE_LOOP, batch.reqs) elif batch.forward_mode.is_extend(): self.process_batch_result_prefill(batch, result) elif batch.forward_mode.is_idle(): if self.enable_overlap: if result.copy_done is not None: result.copy_done.synchronize() self.maybe_send_health_check_signal() def maybe_send_health_check_signal(self): if self.return_health_check_ct: # Return some signal for the health check. # This is used to prevent the health check signal being blocked by long context prefill. # However, one minor issue is that this code path does not check the status of detokenizer manager. self.return_health_check_ct -= 1 self.send_to_tokenizer.send_output(HealthCheckOutput()) def prepare_mlp_sync_batch(self, local_batch: ScheduleBatch): return self.prepare_mlp_sync_batch_raw( local_batch, dp_size=self.server_args.dp_size, attn_tp_size=self.attn_tp_size, tp_group=self.tp_group, get_idle_batch=self.get_idle_batch, disable_cuda_graph=self.server_args.disable_cuda_graph, spec_algorithm=self.spec_algorithm, speculative_num_draft_tokens=self.server_args.speculative_num_draft_tokens, require_mlp_tp_gather=require_mlp_tp_gather(self.server_args), disable_overlap_schedule=self.server_args.disable_overlap_schedule, offload_tags=self.offload_tags, ) @staticmethod def prepare_mlp_sync_batch_raw( local_batch: ScheduleBatch, dp_size, attn_tp_size: int, tp_group, get_idle_batch, disable_cuda_graph: bool, spec_algorithm, speculative_num_draft_tokens, require_mlp_tp_gather: bool, disable_overlap_schedule: bool, offload_tags: set[str], ): # Check if other DP workers have running batches if local_batch is None: num_tokens = 0 num_tokens_for_logprob = 0 elif local_batch.forward_mode.is_decode(): num_tokens = local_batch.batch_size() num_tokens_for_logprob = num_tokens else: num_tokens = local_batch.extend_num_tokens if local_batch.return_logprob: num_tokens_for_logprob = sum( # We should have at least 1 token for sample in every case. max(extend_len - logprob_start_len, 1) for logprob_start_len, extend_len in zip( local_batch.extend_logprob_start_lens, local_batch.extend_lens, ) ) else: # When return_logprob = False, only need last token per request num_tokens_for_logprob = local_batch.batch_size() if local_batch is None or local_batch.forward_mode.is_decode_or_idle(): can_cuda_graph = 1 else: can_cuda_graph = 0 is_extend_in_batch = ( local_batch.forward_mode.is_extend() if local_batch else False ) tbo_preparer = TboDPAttentionPreparer() if len(offload_tags) == 0 and disable_overlap_schedule: group = tp_group.device_group device = tp_group.device else: group = tp_group.cpu_group device = "cpu" local_info = torch.tensor( [ num_tokens, can_cuda_graph, num_tokens_for_logprob, is_extend_in_batch, *tbo_preparer.prepare_all_gather( local_batch, ), ], dtype=torch.int64, device=device, ) global_info = torch.empty( (dp_size, attn_tp_size, 6), dtype=torch.int64, device=device, ) torch.distributed.all_gather_into_tensor( global_info.flatten(), local_info, group=group, ) global_num_tokens = global_info[:, 0, 0].tolist() can_cuda_graph = min(global_info[:, 0, 1].tolist()) global_num_tokens_for_logprob = global_info[:, 0, 2].tolist() is_extend_in_batch = global_info[:, 0, 3].tolist() tbo_split_seq_index, global_forward_mode = tbo_preparer.compute_output( global_info[:, :, 4:6] ) if local_batch is None and max(global_num_tokens) > 0: local_batch = get_idle_batch() if local_batch is not None: # TODO: handle the case when moe_dense_tp_size != 1 if not require_mlp_tp_gather: local_batch.global_num_tokens = [num_tokens] local_batch.global_num_tokens_for_logprob = [num_tokens_for_logprob] else: local_batch.global_num_tokens = global_num_tokens local_batch.global_num_tokens_for_logprob = ( global_num_tokens_for_logprob ) local_batch.is_extend_in_batch = any(is_extend_in_batch) local_batch.tbo_split_seq_index = tbo_split_seq_index local_batch.global_forward_mode = global_forward_mode # Check forward mode for cuda graph if not disable_cuda_graph: local_batch.can_run_dp_cuda_graph = can_cuda_graph return local_batch def get_idle_batch(self): idle_batch = ScheduleBatch.init_new( [], self.req_to_token_pool, self.token_to_kv_pool_allocator, self.tree_cache, self.model_config, self.enable_overlap, self.spec_algorithm, ) idle_batch.prepare_for_idle() return idle_batch def move_ready_grammar_requests(self): """Move requests whose grammar objects are ready from grammar_queue to waiting_queue.""" num_ready_reqs = 0 num_timeout_reqs = 0 for req in self.grammar_queue: try: if req.finished(): # It is aborted by AbortReq num_ready_reqs += 1 continue req.grammar = req.grammar.result(timeout=0.03) self.grammar_backend.set_cache(req.grammar_key, req.grammar.copy()) if req.grammar is INVALID_GRAMMAR_OBJ: error_msg = f"Invalid grammar request: {req.grammar_key=}" req.set_finish_with_abort(error_msg) num_ready_reqs += 1 except futures._base.TimeoutError: req.grammar_wait_ct += 1 # NOTE(lianmin): this timeout is the waiting time of the above line. It is # not the waiting time from it enters the grammar queue. if req.grammar_wait_ct > GRAMMAR_TIMEOUT / 0.03: num_timeout_reqs = 1 break if self.server_args.enable_dp_attention: tp_size = self.attn_tp_size tp_group = self.attn_tp_cpu_group else: tp_size = self.tp_size tp_group = self.tp_cpu_group if tp_size > 1: # Sync across TP ranks to make sure they have the same number of ready requests tensor = torch.tensor([num_ready_reqs, num_timeout_reqs], dtype=torch.int32) torch.distributed.all_reduce( tensor, op=torch.distributed.ReduceOp.MAX, group=tp_group ) num_ready_reqs_max, num_timeout_reqs_max = tensor.tolist() for i in range(num_ready_reqs, num_ready_reqs_max): req = self.grammar_queue[i] if req.finished(): # It is aborted by AbortReq continue req.grammar = req.grammar.result() self.grammar_backend.set_cache(req.grammar_key, req.grammar.copy()) if req.grammar is INVALID_GRAMMAR_OBJ: error_msg = f"Invalid grammar request: {req.grammar_key=}" req.set_finish_with_abort(error_msg) else: num_ready_reqs_max = num_ready_reqs num_timeout_reqs_max = num_timeout_reqs for i in range(num_ready_reqs, num_ready_reqs + num_timeout_reqs_max): req = self.grammar_queue[i] req.grammar.cancel() self.grammar_backend.set_cache(req.grammar_key, INVALID_GRAMMAR_OBJ) error_msg = f"Grammar preprocessing timed out for {req.grammar_key=}" req.set_finish_with_abort(error_msg) num_ready_reqs = num_ready_reqs_max + num_timeout_reqs_max for req in self.grammar_queue[:num_ready_reqs]: self._add_request_to_queue(req) self.grammar_queue = self.grammar_queue[num_ready_reqs:] def watchdog_thread(self): """A watch dog thread that will try to kill the server itself if one forward batch takes too long.""" self.watchdog_last_forward_ct = 0 self.watchdog_last_time = time.perf_counter() while True: current = time.perf_counter() if self.cur_batch is not None: if self.watchdog_last_forward_ct == self.forward_ct: if current > self.watchdog_last_time + self.watchdog_timeout: break else: self.watchdog_last_forward_ct = self.forward_ct self.watchdog_last_time = current time.sleep(self.watchdog_timeout // 2) if not disable_request_logging(): # Print batch size and memory pool info to check whether there are de-sync issues. if self.is_hybrid: ( _, _, _, _, full_available_size, full_evictable_size, swa_available_size, swa_evictable_size, ) = self._get_swa_token_info() info_msg = ( f"{full_available_size=}, " f"{full_evictable_size=}, " f"{swa_available_size=}, " f"{swa_evictable_size=}, " ) elif self.is_hybrid_gdn and isinstance(self.tree_cache, MambaRadixCache): ( _, _, _, _, full_available_size, full_evictable_size, mamba_available_size, mamba_evictable_size, ) = self._get_mamba_token_info() info_msg = ( f"{full_available_size=}, " f"{full_evictable_size=}, " f"{mamba_available_size=}, " f"{mamba_evictable_size=}, " ) else: _, _, available_size, evictable_size = self._get_token_info() info_msg = f"{available_size=}, " f"{evictable_size=}, " logger.error( f"{self.cur_batch.batch_size()=}, " f"{self.cur_batch.reqs=}, " f"{info_msg}" ) pyspy_dump_schedulers() logger.error(f"Watchdog timeout ({self.watchdog_timeout=})") print(file=sys.stderr, flush=True) print(file=sys.stdout, flush=True) # Wait for some time so that the parent process can print the error. time.sleep(5) self.parent_process.send_signal(signal.SIGQUIT) def flush_cache_wrapped(self, recv_req: FlushCacheReqInput): success = self.flush_cache() return FlushCacheReqOutput(success=success) def clear_hicache_storage_wrapped(self, recv_req: ClearHiCacheReqInput): if self.enable_hierarchical_cache: self.tree_cache.clear_storage_backend() logger.info("Hierarchical cache cleared successfully!") if_success = True else: logging.warning("Hierarchical cache is not enabled.") if_success = False return ClearHiCacheReqOutput(success=if_success) def _is_no_request(self): no_request = ( len(self.waiting_queue) == 0 and self.running_batch.is_empty() and (self.last_batch is None or self.last_batch.is_empty()) and (self.cur_batch is None or self.cur_batch.is_empty()) and (not self.enable_overlap or len(self.result_queue) == 0) and (self.pp_size == 1 or all(x.is_empty() for x in self.running_mbs)) ) if self.disaggregation_mode == DisaggregationMode.PREFILL: no_request &= ( len(self.disagg_prefill_bootstrap_queue.queue) == 0 and len(self.disagg_prefill_inflight_queue) == 0 ) if self.disaggregation_mode == DisaggregationMode.DECODE: no_request &= ( len(self.disagg_decode_prealloc_queue.queue) == 0 and len(self.disagg_decode_transfer_queue.queue) == 0 ) return no_request def flush_cache(self): """Flush the memory pool and cache.""" if self._is_no_request(): self.cur_batch = None self.last_batch = None self.tree_cache.reset() if self.grammar_backend: self.grammar_backend.reset() self.req_to_token_pool.clear() self.token_to_kv_pool_allocator.clear() if self.draft_worker: self.draft_worker.clear_cache_pool() self.num_generated_tokens = 0 self.forward_ct_decode = 0 self.spec_num_accepted_tokens = 0 self.spec_num_forward_ct = 0 self.spec_total_num_accepted_tokens = 0 self.spec_total_num_forward_ct = 0 torch.cuda.empty_cache() logger.info("Cache flushed successfully!") if_success = True else: logging.warning( f"Cache not flushed because there are pending requests. " f"#queue-req: {len(self.waiting_queue)}, " f"#running-req: {len(self.running_batch.reqs)}" ) if_success = False return if_success def get_load(self, recv_req: GetLoadReqInput = None) -> GetLoadReqOutput: # TODO(lsyin): use dynamically maintained num_waiting_tokens if self.is_hybrid: num_tokens_full = ( self.full_tokens_per_layer - self.token_to_kv_pool_allocator.full_available_size() - self.tree_cache.full_evictable_size() ) num_tokens_swa = ( self.swa_tokens_per_layer - self.token_to_kv_pool_allocator.swa_available_size() - self.tree_cache.swa_evictable_size() ) num_tokens = max(num_tokens_full, num_tokens_swa) else: num_tokens = ( self.max_total_num_tokens - self.token_to_kv_pool_allocator.available_size() - self.tree_cache.evictable_size() ) # Tokens in waiting queue, bootstrap queue, prealloc queue num_tokens += sum(len(req.origin_input_ids) for req in self.waiting_queue) num_waiting_reqs = len(self.waiting_queue) if self.disaggregation_mode == DisaggregationMode.PREFILL: num_tokens += sum( len(req.origin_input_ids) for req in self.disagg_prefill_bootstrap_queue.queue ) num_waiting_reqs += len(self.disagg_prefill_bootstrap_queue.queue) elif self.disaggregation_mode == DisaggregationMode.DECODE: num_tokens += sum( len(req.req.origin_input_ids) for req in self.disagg_decode_prealloc_queue.queue ) num_waiting_reqs += len(self.disagg_decode_prealloc_queue.queue) return GetLoadReqOutput( dp_rank=self.dp_rank, num_reqs=len(self.running_batch.reqs) + num_waiting_reqs, num_waiting_reqs=num_waiting_reqs, num_tokens=num_tokens, ) def get_internal_state(self, recv_req: GetInternalStateReq): ret = vars(get_global_server_args()) ret["last_gen_throughput"] = self.last_gen_throughput ret["memory_usage"] = { "weight": round(self.tp_worker.model_runner.weight_load_mem_usage, 2), "kvcache": round( self.token_to_kv_pool_allocator.get_kvcache().mem_usage, 2 ), "token_capacity": int(self.max_total_num_tokens), } ret["memory_usage"]["graph"] = round( self.tp_worker.model_runner.graph_mem_usage, 2 ) if not self.spec_algorithm.is_none() and self.spec_total_num_forward_ct > 0: ret["avg_spec_accept_length"] = ( self.spec_total_num_accepted_tokens / self.spec_total_num_forward_ct ) if RECORD_STEP_TIME: ret["step_time_dict"] = self.step_time_dict # This field is not serializable. ret.pop("model_config", None) return GetInternalStateReqOutput(internal_state=ret) def set_internal_state(self, recv_req: SetInternalStateReq): server_args_dict = recv_req.server_args args_allow_update = set( [ "pp_max_micro_batch_size", "speculative_accept_threshold_single", "speculative_accept_threshold_acc", ] ) if_success = True for k, v in server_args_dict.items(): if k not in args_allow_update: logging.warning(f"Updating {k} is not supported.") if_success = False break elif k == "pp_max_micro_batch_size" and ( v > self.max_running_requests // self.pp_size or v < 1 ): logging.warning( f"Updating {k} to {v} is rejected because it is out of the valid range [1, {self.max_running_requests // self.pp_size}]." ) if_success = False break if if_success: if not self.spec_algorithm.is_none() and self.spec_total_num_forward_ct > 0: avg_spec_accept_length = ( self.spec_total_num_accepted_tokens / self.spec_total_num_forward_ct ) logger.info(f"{avg_spec_accept_length=}") self.spec_total_num_accepted_tokens = self.spec_total_num_forward_ct = 0 for k, v in server_args_dict.items(): setattr(get_global_server_args(), k, v) logger.info(f"Global server args updated! {get_global_server_args()=}") return SetInternalStateReqOutput( updated=True, server_args=vars(get_global_server_args()), ) def handle_rpc_request(self, recv_req: RpcReqInput): # Handle RPC requests logger.info( f"handle_rpc_request: {recv_req.method}, param: {recv_req.parameters}" ) success = True exec = None try: func = getattr(self, recv_req.method) func(recv_req.parameters) except Exception as e: success = False exec = e logger.error(f"Failed to call rpc {recv_req.method}: {str(e)}") barrier() return RpcReqOutput(success, "" if not exec else str(exec)) def abort_request(self, recv_req: AbortReq): # Delete requests in the waiting queue to_del = [] for i, req in enumerate(self.waiting_queue): if recv_req.abort_all or req.rid.startswith(recv_req.rid): to_del.append(i) # Sort in reverse order to avoid index issues when deleting for i in reversed(to_del): # Abort method 1: directly pop from the queue # This only works for requests that have not started anything. # We still need to send something back to TokenizerManager to clean up the state. req = self.waiting_queue.pop(i) if self.enable_hicache_storage: # to release prefetch events associated with the request self.tree_cache.release_aborted_request(req.rid) self.send_to_tokenizer.send_output(AbortReq(rid=req.rid), req) # For disaggregation decode mode, the request in the waiting queue has KV cache allocated. if self.disaggregation_mode == DisaggregationMode.DECODE: self.tree_cache.cache_finished_req(req) logger.debug(f"Abort queued request. {req.rid=}") # Delete the requests in the grammar queue for req in self.grammar_queue: # Abort method 2: call `set_finish_with_abort` # The request will still run one prefill forward pass. # In this case, we change the input_ids to be only one token to make this prefill cheap. if recv_req.abort_all or req.rid.startswith(recv_req.rid): logger.debug(f"Abort grammar queue request. {req.rid=}") if req.grammar: req.grammar.cancel() req.set_finish_with_abort("Aborted by AbortReq.") # Delete requests not in the waiting queue when PD disaggregation is enabled if self.disaggregation_mode == DisaggregationMode.PREFILL: # Abort requests that have not yet been bootstrapped for req in self.disagg_prefill_bootstrap_queue.queue: if recv_req.abort_all or req.rid.startswith(recv_req.rid): logger.debug(f"Abort bootstrap queue request. {req.rid=}") if hasattr(req.disagg_kv_sender, "abort"): req.disagg_kv_sender.abort() # Abort in-flight requests for req in self.disagg_prefill_inflight_queue: if recv_req.abort_all or req.rid.startswith(recv_req.rid): logger.debug(f"Abort inflight queue request. {req.rid=}") if hasattr(req.disagg_kv_sender, "abort"): req.disagg_kv_sender.abort() elif self.disaggregation_mode == DisaggregationMode.DECODE: # Abort requests that have not yet finished preallocation for decode_req in self.disagg_decode_prealloc_queue.queue: if recv_req.abort_all or decode_req.req.rid.startswith(recv_req.rid): logger.debug(f"Abort prealloc queue request. {decode_req.req.rid=}") if hasattr(decode_req.kv_receiver, "abort"): decode_req.kv_receiver.abort() # Abort requests waiting for kvcache to release tree cache for decode_req in self.disagg_decode_transfer_queue.queue: if recv_req.abort_all or decode_req.req.rid.startswith(recv_req.rid): logger.debug(f"Abort transfer queue request. {decode_req.req.rid=}") if hasattr(decode_req.kv_receiver, "abort"): decode_req.kv_receiver.abort() # Delete requests in the running batch if self.cur_batch is self.running_batch or self.cur_batch is None: reqs = self.running_batch.reqs else: reqs = self.running_batch.reqs + self.cur_batch.reqs for req in reqs: if not req.finished() and ( recv_req.abort_all or req.rid.startswith(recv_req.rid) ): # Abort method 3: set `to_finish` # The request will still run one decode forward pass. # Then we reuse all existing code to clean up the KV cache allocation. logger.debug(f"Abort running request. {req.rid=}") req.to_finish = FINISH_ABORT() def _pause_engine(self) -> Tuple[List[Req], int]: raise NotImplementedError() def load_lora_adapter( self, recv_req: LoadLoRAAdapterReqInput ) -> LoadLoRAAdapterReqOutput: """In-place loading a new lora adapter from disk or huggingface.""" result = self.tp_worker.load_lora_adapter(recv_req) return result def unload_lora_adapter( self, recv_req: UnloadLoRAAdapterReqInput ) -> UnloadLoRAAdapterReqOutput: """Unload the lora adapter.""" result = self.tp_worker.unload_lora_adapter(recv_req) return result def init_weights_send_group_for_remote_instance( self, recv_req: InitWeightsSendGroupForRemoteInstanceReqInput ): """Init the seed and client instance communication group.""" success, message = self.tp_worker.init_weights_send_group_for_remote_instance( recv_req ) return InitWeightsSendGroupForRemoteInstanceReqOutput(success, message) def send_weights_to_remote_instance( self, recv_req: SendWeightsToRemoteInstanceReqInput ): """Send the seed instance weights to the destination instance.""" success, message = self.tp_worker.send_weights_to_remote_instance(recv_req) return SendWeightsToRemoteInstanceReqOutput(success, message) def slow_down(self, recv_req: SlowDownReqInput): t = recv_req.forward_sleep_time if t is not None and t <= 0: t = None self.forward_sleep_time = t return SlowDownReqOutput() def expert_distribution_handle(self, recv_req: ExpertDistributionReq): action = recv_req.action if action == ExpertDistributionReqType.START_RECORD: get_global_expert_distribution_recorder().start_record() elif action == ExpertDistributionReqType.STOP_RECORD: get_global_expert_distribution_recorder().stop_record() elif action == ExpertDistributionReqType.DUMP_RECORD: get_global_expert_distribution_recorder().dump_record() else: raise ValueError(f"Unrecognized ExpertDistributionReq value: {recv_req=}") return ExpertDistributionReqOutput() def open_session(self, recv_req: OpenSessionReqInput): # handle error session_id = recv_req.session_id if session_id in self.sessions: logger.warning(f"session id {session_id} already exist, cannot open.") return OpenSessionReqOutput(session_id, False) elif session_id is None: logger.warning("session id is None, cannot open.") return OpenSessionReqOutput(session_id, False) else: self.sessions[session_id] = Session( recv_req.capacity_of_str_len, session_id ) return OpenSessionReqOutput(session_id, True) def close_session(self, recv_req: CloseSessionReqInput): # handle error session_id = recv_req.session_id if session_id not in self.sessions: logger.warning(f"session id {session_id} does not exist, cannot delete.") else: del self.sessions[session_id] def get_print_prefix(self): prefix = "" if self.attn_dp_rank is not None: prefix += f" DP{self.attn_dp_rank}" if self.server_args.tp_size > 1: prefix += f" TP{self.tp_rank}" if self.pp_size > 1: prefix += f" PP{self.pp_rank}" return prefix def current_scheduler_metrics_enabled(self): return self.attn_tp_rank == 0 or self.enable_metrics_for_all_schedulers def maybe_sleep_on_idle(self): if self.idle_sleeper is not None: self.idle_sleeper.maybe_sleep() def handle_freeze_gc(self, recv_req: FreezeGCReq): """Handle freeze_gc request: freeze scheduler's GC and forward to detokenizer.""" freeze_gc("Scheduler") self.send_to_detokenizer.send_output(recv_req, recv_req) return None class IdleSleeper: """ In setups which have long inactivity periods it is desirable to reduce system power consumption when sglang does nothing. This would lead not only to power savings, but also to more CPU thermal headroom when a request eventually comes. This is important in cases when multiple GPUs are connected as each GPU would otherwise pin one thread at 100% CPU usage. The simplest solution is to use zmq.Poller on all sockets that may receive data that needs handling immediately. """ def __init__(self, sockets): self.poller = zmq.Poller() self.last_empty_time = time.time() for s in sockets: self.poller.register(s, zmq.POLLIN) self.empty_cache_interval = envs.SGLANG_EMPTY_CACHE_INTERVAL.get() def maybe_sleep(self): self.poller.poll(1000) if ( self.empty_cache_interval > 0 and time.time() - self.last_empty_time > self.empty_cache_interval ): self.last_empty_time = time.time() torch.cuda.empty_cache() def is_health_check_generate_req(recv_req): rid = getattr(recv_req, "rid", None) return rid is not None and rid.startswith("HEALTH_CHECK") def is_work_request(recv_req): return isinstance( recv_req, ( TokenizedGenerateReqInput, TokenizedEmbeddingReqInput, BatchTokenizedGenerateReqInput, BatchTokenizedEmbeddingReqInput, ), ) def run_scheduler_process( server_args: ServerArgs, port_args: PortArgs, gpu_id: int, tp_rank: int, moe_ep_rank: int, pp_rank: int, dp_rank: Optional[int], pipe_writer, ): # Generate the logger prefix prefix = "" if dp_rank is None and "SGLANG_DP_RANK" in os.environ: # [For Router] if env var "SGLANG_DP_RANK" exist, set dp_rank to the value of the env var dp_rank = int(os.environ["SGLANG_DP_RANK"]) if dp_rank is not None: prefix += f" DP{dp_rank}" if server_args.tp_size > 1: prefix += f" TP{tp_rank}" if server_args.ep_size > 1: prefix += f" EP{moe_ep_rank}" if server_args.pp_size > 1: prefix += f" PP{pp_rank}" # Config the process setproctitle.setproctitle(f"sglang::scheduler{prefix.replace(' ', '_')}") faulthandler.enable() kill_itself_when_parent_died() parent_process = psutil.Process().parent() # Configure the logger configure_logger(server_args, prefix=prefix) suppress_other_loggers() # Set cpu affinity to this gpu process if get_bool_env_var("SGLANG_SET_CPU_AFFINITY"): set_gpu_proc_affinity( server_args.pp_size, server_args.tp_size, server_args.nnodes, gpu_id ) if (numa_node := server_args.numa_node) is not None: numa_bind_to_node(numa_node[gpu_id]) # Set up tracing if server_args.enable_trace: process_tracing_init(server_args.otlp_traces_endpoint, "sglang") thread_label = "Scheduler" if server_args.disaggregation_mode == "prefill": thread_label = "Prefill Scheduler" elif server_args.disaggregation_mode == "decode": thread_label = "Decode Scheduler" trace_set_thread_info(thread_label, tp_rank, dp_rank) # Create a scheduler and run the event loop try: scheduler = Scheduler( server_args, port_args, gpu_id, tp_rank, moe_ep_rank, pp_rank, dp_rank, ) pipe_writer.send( { "status": "ready", "max_total_num_tokens": scheduler.max_total_num_tokens, "max_req_input_len": scheduler.max_req_input_len, } ) disaggregation_mode: DisaggregationMode = scheduler.disaggregation_mode if disaggregation_mode == DisaggregationMode.NULL: if server_args.pp_size > 1: scheduler.event_loop_pp() elif scheduler.enable_overlap: scheduler.event_loop_overlap() else: scheduler.event_loop_normal() elif disaggregation_mode == DisaggregationMode.PREFILL: if scheduler.enable_overlap: scheduler.event_loop_overlap_disagg_prefill() else: if server_args.pp_size > 1: scheduler.event_loop_pp_disagg_prefill() else: scheduler.event_loop_normal_disagg_prefill() elif disaggregation_mode == DisaggregationMode.DECODE: if scheduler.enable_overlap: scheduler.event_loop_overlap_disagg_decode() else: scheduler.event_loop_normal_disagg_decode() except Exception: traceback = get_exception_traceback() logger.error(f"Scheduler hit an exception: {traceback}") parent_process.send_signal(signal.SIGQUIT)