sglang/python/sglang/srt/managers/scheduler.py

# 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 dataclasses
import faulthandler
import logging
import math
import os
import signal
import sys
import time
from array import array
from collections import deque
from contextlib import contextmanager, nullcontext
from functools import partial
from http import HTTPStatus
from typing import TYPE_CHECKING, Any, Deque, Dict, List, Optional, Set, Tuple, Union

from sglang.srt.runtime_context import (
    SpawnRanks,
    attention_backends,
    get_context,
    get_device,
    get_disagg,
    get_exec,
    get_lora,
    get_memory,
    get_mm,
    get_model,
    get_observability,
    get_parallel,
    get_schedule,
    get_serving,
    get_spec,
    publish,
    spawn_world_rank,
)

from sglang.srt.utils.common import suppress_noisy_warnings  # isort: skip

suppress_noisy_warnings()

import psutil  # isort: skip
import setproctitle
import torch
import torch.distributed
from torch.distributed import barrier

if TYPE_CHECKING:
    from torch.cuda import Stream as CudaStream

try:
    from sglang.kernels.ops.mamba.triton_ops import (
        initialize_mamba_selective_state_update_backend,
    )
except ImportError:
    initialize_mamba_selective_state_update_backend = None
from sglang.srt.beam_search.coordinator import BeamCoordinator
from sglang.srt.configs.model_config import (
    ModelConfig,
    ModelImpl,
    is_deepseek_v4,
    is_minimax_sparse,
)
from sglang.srt.constrained.grammar_manager import GrammarManager
from sglang.srt.debug_utils.pr_fix_toggle import maybe_revert_pr_fix
from sglang.srt.disaggregation import role_switch
from sglang.srt.disaggregation.checksum import KvChecksumComputer
from sglang.srt.disaggregation.decode import (
    DecodePreallocQueue,
    DecodeTransferQueue,
    SchedulerDisaggregationDecodeMixin,
)
from sglang.srt.disaggregation.decode_kvcache_offload_manager import (
    DecodeKVCacheOffloadManager,
)
from sglang.srt.disaggregation.encoder.receiver import create_mm_receiver
from sglang.srt.disaggregation.prefill import (
    PrefillBootstrapQueue,
    SchedulerDisaggregationPrefillMixin,
    maybe_release_metadata_buffer,
)
from sglang.srt.disaggregation.utils import (
    DisaggregationMode,
    MetadataBuffers,
    ReqToMetadataIdxAllocator,
    TransferBackend,
    get_dsa_seed_metadata_dim,
    prepare_abort,
    unified_memory_disagg_move_gate,
)
from sglang.srt.distributed.parallel_state import (
    abort_distributed_environment,
)
from sglang.srt.distributed.parallel_state_wrapper import ParallelState
from sglang.srt.dllm.mixin.scheduler import SchedulerDllmMixin
from sglang.srt.environ import envs, exportable_env_vars
from sglang.srt.eplb.expert_distribution import get_global_expert_distribution_recorder
from sglang.srt.hardware_backend.mlx.runtime import use_mlx
from sglang.srt.layers.dp_attention import compute_dp_attention_world_info
from sglang.srt.layers.moe import initialize_moe_config
from sglang.srt.layers.quantization.fp4_utils import initialize_fp4_gemm_config
from sglang.srt.layers.quantization.fp8_utils import initialize_fp8_gemm_config
from sglang.srt.layers.quantization.unquant import initialize_bf16_gemm_config
from sglang.srt.lora.lora_drainer import LoRADrainer
from sglang.srt.lora.lora_overlap_loader import LoRAOverlapLoader
from sglang.srt.managers.disagg_service import maybe_create_ascend_config_store
from sglang.srt.managers.hisparse_coordinator import HiSparseCoordinator
from sglang.srt.managers.io_struct import (
    AbortReq,
    ActiveRanksOutput,
    AddExternalCorpusReqInput,
    AddExternalCorpusReqOutput,
    AttachHiCacheStorageReqInput,
    AttachHiCacheStorageReqOutput,
    BatchTokenizedEmbeddingReqInput,
    BatchTokenizedGenerateReqInput,
    CheckWeightsReqInput,
    ClearHiCacheReqInput,
    ClearHiCacheReqOutput,
    CloseSessionReqInput,
    ConfigureLoggingReq,
    ContinueGenerationReqInput,
    DestroyWeightsUpdateGroupReqInput,
    DetachHiCacheStorageReqInput,
    DetachHiCacheStorageReqOutput,
    DumperControlReqInput,
    DumperControlReqOutput,
    ExpertDistributionReq,
    ExpertDistributionReqOutput,
    ExpertDistributionReqType,
    FinishReasonDict,
    FlushCacheReqInput,
    FreezeGCReq,
    GetInternalStateReq,
    GetInternalStateReqOutput,
    GetWeightsByNameReqInput,
    HealthCheckOutput,
    InitWeightsSendGroupForRemoteInstanceReqInput,
    InitWeightsSendGroupForRemoteInstanceReqOutput,
    InitWeightsUpdateGroupReqInput,
    ListExternalCorporaReqInput,
    ListExternalCorporaReqOutput,
    LoadLoRAAdapterFromTensorsReqInput,
    LoadLoRAAdapterFromTensorsReqOutput,
    LoadLoRAAdapterReqInput,
    LoadLoRAAdapterReqOutput,
    MMInputsProcessError,
    OpenSessionReqInput,
    PauseGenerationReqInput,
    PdRoleSwitchReqInput,
    ProfileReq,
    ReleaseMemoryOccupationReqInput,
    RemoveExternalCorpusReqInput,
    RemoveExternalCorpusReqOutput,
    ResumeMemoryOccupationReqInput,
    RpcReqInput,
    RpcReqOutput,
    ScaleElasticEPReqInput,
    ScaleElasticEPReqOutput,
    SendWeightsToRemoteInstanceReqInput,
    SendWeightsToRemoteInstanceReqOutput,
    SetInternalStateReq,
    SetInternalStateReqOutput,
    ShutdownReq,
    SlowDownReqInput,
    SlowDownReqOutput,
    TokenizedEmbeddingReqInput,
    TokenizedGenerateReqInput,
    UnloadLoRAAdapterReqInput,
    UnloadLoRAAdapterReqOutput,
    UpdateWeightFromDiskReqInput,
    UpdateWeightsFromDistributedReqInput,
    UpdateWeightsFromIPCReqInput,
    UpdateWeightsFromTensorReqInput,
    UpdateWeightVersionReqInput,
    UpdateWeightVersionReqOutput,
    sock_send,
)
from sglang.srt.managers.load_snapshot import create_load_snapshot_writer
from sglang.srt.managers.min_free_slots_delayer import (
    MinFreeSlotsDelayer,
    resolve_min_free_slots,
)
from sglang.srt.managers.multimodal_processor import get_mm_processor, import_processors
from sglang.srt.managers.overlap_utils import (
    RelayPayload,
    decide_needs_confidence_relay,
    decide_needs_cpu_seq_lens,
    resolve_forward_inputs,
)
from sglang.srt.managers.prefill_delayer import (
    PrefillDelayer,
    PrefillDelayerSinglePassExecutor,
    RecentPrefillBatchSizeTracker,
)
from sglang.srt.managers.schedule_batch import (
    FINISH_ABORT,
    MultimodalInputs,
    NextBatchPlan,
    Req,
    ScheduleBatch,
    retract_all,
)
from sglang.srt.managers.schedule_policy import (
    AddReqResult,
    PrefillAdder,
    SchedulePolicy,
)
from sglang.srt.managers.scheduler_components.batch_result_processor import (
    SchedulerBatchResultProcessor,
)
from sglang.srt.managers.scheduler_components.dp_attn import SchedulerDPAttnAdapter
from sglang.srt.managers.scheduler_components.dynamic_chunk_sizer import (
    DynamicChunkSizer,
)
from sglang.srt.managers.scheduler_components.flush_wrapper import SchedulerFlushWrapper
from sglang.srt.managers.scheduler_components.idle_sleeper import (
    IdleSleeper,
    RustServerIdleSleeper,
)
from sglang.srt.managers.scheduler_components.invariant_checker import (
    SchedulerInvariantChecker,
    create_scheduler_watchdog,
)
from sglang.srt.managers.scheduler_components.ipc_channels import SchedulerIpcChannels
from sglang.srt.managers.scheduler_components.kv_events_publisher import (
    SchedulerKvEventsPublisher,
)
from sglang.srt.managers.scheduler_components.load_inquirer import SchedulerLoadInquirer
from sglang.srt.managers.scheduler_components.load_publisher import (
    SchedulerLoadPublisher,
)
from sglang.srt.managers.scheduler_components.logprob_result_processor import (
    SchedulerLogprobResultProcessor,
)
from sglang.srt.managers.scheduler_components.memory_usage import (
    build_memory_usage,
    combine_graph_memory_usage,
)
from sglang.srt.managers.scheduler_components.metrics_reporter import (
    RECORD_STEP_TIME,
    PrefillStats,
    SchedulerMetricsReporter,
)
from sglang.srt.managers.scheduler_components.new_token_ratio_tracker import (
    NewTokenRatioTracker,
)
from sglang.srt.managers.scheduler_components.output_streamer import (
    SchedulerOutputStreamer,
)
from sglang.srt.managers.scheduler_components.pool_stats_observer import (
    SchedulerPoolStatsObserver,
)
from sglang.srt.managers.scheduler_components.profiler_manager import (
    SchedulerProfilerManager,
)
from sglang.srt.managers.scheduler_components.recv_skipper import (
    SchedulerRecvSkipper,
)
from sglang.srt.managers.scheduler_components.request_receiver import (
    SchedulerRequestReceiver,
)
from sglang.srt.managers.scheduler_components.weight_updater import (
    SchedulerWeightUpdaterManager,
)
from sglang.srt.managers.scheduler_input_blocker import SchedulerInputBlocker
from sglang.srt.managers.scheduler_pp_mixin import SchedulerPPMixin
from sglang.srt.managers.utils import (
    EmbeddingBatchResult,
    GenerationBatchResult,
    allocate_distinct_stream,
    is_health_check_generate_req,
    validate_input_length,
)
from sglang.srt.mem_cache import kv_cache_builder
from sglang.srt.mem_cache.base_prefix_cache import CacheRequestOutcome
from sglang.srt.mem_cache.common import (
    maybe_cache_unfinished_req,
    release_kv_cache,
    retraction_discard,
)
from sglang.srt.model_executor.forward_batch_info import PPProxyTensors
from sglang.srt.model_executor.runner_utils.pool import prewarm_graph_pool_borrow
from sglang.srt.model_loader.utils import get_resolved_model_impl
from sglang.srt.multiplex.multiplexing_mixin import SchedulerMultiplexMixin
from sglang.srt.observability.metrics_collector import SchedulerMetricsCollector
from sglang.srt.observability.req_time_stats import (
    flush_trace_batch,
    set_schedule_time_batch,
    set_time_batch,
)
from sglang.srt.observability.scheduler_stage_metrics import (
    SCHEDULER_STAGE_GET_NEXT_BATCH,
    SCHEDULER_STAGE_IDLE,
    SCHEDULER_STAGE_PROCESS_BATCH_RESULT,
    SCHEDULER_STAGE_PROCESS_REQUESTS,
    SCHEDULER_STAGE_RUN_BATCH,
    SCHEDULER_STAGE_SANITY_CHECK_CACHE,
    scheduler_stage_method,
)
from sglang.srt.observability.startup_time import build_scheduler_startup_time
from sglang.srt.observability.trace import process_tracing_init, trace_set_thread_info
from sglang.srt.parser.reasoning_parser import ReasoningParser
from sglang.srt.platforms import current_platform
from sglang.srt.plugins import load_plugins
from sglang.srt.rust_server.server import RustServer
from sglang.srt.sampling.sampling_batch_info import SamplingBatchInfo
from sglang.srt.server_args import PortArgs, ServerArgs, compute_world_size
from sglang.srt.session.session_controller import SessionController
from sglang.srt.speculative.base_spec_worker import BaseSpecWorker
from sglang.srt.speculative.dflash_utils import validate_dflash_request
from sglang.srt.speculative.eagle_utils import (
    get_draft_recurrent_hidden_state_spec_from_config,
)
from sglang.srt.speculative.spec_info import SpeculativeAlgorithm
from sglang.srt.speculative.uno_validation import validate_uno_request
from sglang.srt.state_capturer.indexer_topk import destroy_global_indexer_capturer
from sglang.srt.state_capturer.routed_experts import destroy_global_experts_capturer
from sglang.srt.utils import (
    DynamicGradMode,
    configure_gc_logger,
    configure_logger,
    freeze_gc,
    get_available_gpu_memory,
    get_bool_env_var,
    get_int_env_var,
    is_cuda,
    is_hip,
    is_mps,
    kill_itself_when_parent_died,
    rank_consensus_checker,
    require_mlp_sync,
    set_gpu_proc_affinity,
    set_random_seed,
    suppress_other_loggers,
    triton_load_watch,
)
from sglang.srt.utils.common import is_npu
from sglang.srt.utils.hf_transformers_utils import (
    get_processor,
    get_tokenizer,
    get_tokenizer_from_processor,
    resolve_image_processor_backend,
)
from sglang.srt.utils.msgspec_utils import msgspec_to_builtins
from sglang.srt.utils.numa_utils import get_numa_node_if_available, numa_bind_to_node
from sglang.srt.utils.torch_memory_saver_adapter import TorchMemorySaverAdapter
from sglang.srt.utils.weight_versions import (
    compute_weight_version_spans,
    record_weight_version_events,
)
from sglang.utils import TypeBasedDispatcher, get_exception_traceback

if is_mps():
    CudaStreamContext = nullcontext
    from sglang.srt.hardware_backend.mlx.scheduler_mixin import SchedulerMlxOverlapMixin
else:
    from torch.cuda import StreamContext as CudaStreamContext

    class SchedulerMlxOverlapMixin:
        pass


logger = logging.getLogger(__name__)


def _prewarm_hccl_group(device, group, device_module):
    warmup_tensor = torch.zeros(1, dtype=torch.int32, device=device)
    torch.distributed.all_reduce(warmup_tensor, group=group)
    device_module.synchronize()


# 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()


STEP_MAX_US = 2_000_000

# Min wall-clock between load publishes on the stalled no-batch path, which
# spins on_idle without sleeping. Bounds the O(queue) get_loads for both the
# DP-balancing writer and the router-facing socket.
LOAD_STALL_REFRESH_S = 0.05


@dataclasses.dataclass(frozen=True)
class _MultimodalInputBroadcast:
    inputs: Optional[MultimodalInputs] = None
    error: Optional[str] = None


class _MultimodalInputProcessingError(RuntimeError):
    pass


def _accumulate_decode_moment(
    totals: list[float],
    batch_size: int,
    step_us: int,
    generated: int,
) -> None:
    if batch_size <= 0 or step_us <= 0:
        return
    b = float(batch_size)
    t = float(step_us)
    g = float(generated)
    totals[0] += 1.0
    totals[1] += b
    totals[2] += t
    totals[3] += b * b
    totals[4] += b * t
    totals[5] += g


_is_npu = is_npu()
_is_hip = is_hip()


class Scheduler(
    SchedulerDisaggregationDecodeMixin,
    SchedulerDisaggregationPrefillMixin,
    SchedulerMultiplexMixin,
    SchedulerPPMixin,
    SchedulerDllmMixin,
    SchedulerMlxOverlapMixin,
):
    """A scheduler that manages a tensor parallel GPU worker."""

    # Class-level default so on_idle's stall gate works even if a fork
    # overrides init_load_publisher (which would otherwise not set it).
    _last_stall_publish_ts: float = float("-inf")
    kv_checksum_computer: Optional[KvChecksumComputer] = None

    def __init__(
        self,
        server_args: ServerArgs,
        port_args: PortArgs,
        gpu_id: int,
        tp_rank: int,
        moe_ep_rank: int,
        pp_rank: int,
        attn_cp_rank: int,
        moe_dp_rank: int,
        dp_rank: Optional[int],
    ):
        # NOTE: KEEP THE FOLLOWING CODE STYLE for this function:
        # Keep __init__ as an orchestrator: sequence init_* and maybe_init_* calls
        # with minimal glue. Move substantial component-specific logic into
        # dedicated methods instead of adding inline blocks here.
        self.init_startup_timing_begin()
        self.is_initializing = True
        # init_soft_watchdog starts a daemon thread that reads these on its first tick.
        self.forward_ct: int = 0
        # Prefill tokens processed so far; used as the aging axis for the HRRN scheduling policy. Reqs snapshot this at waiting_queue entry.
        self.processed_tokens_counter: int = 0
        self.cur_batch_for_debug: Optional[ScheduleBatch] = None
        self.init_soft_watchdog()

        # Parse args
        self.server_args = server_args
        self.nccl_port = port_args.nccl_port
        self.schedule_policy = get_schedule().schedule_policy
        self.enable_priority_scheduling = get_schedule().enable_priority_scheduling
        self.abort_on_priority_when_disabled = (
            get_schedule().abort_on_priority_when_disabled
        )
        self.schedule_low_priority_values_first = (
            get_schedule().schedule_low_priority_values_first
        )
        self.priority_scheduling_preemption_threshold = (
            get_schedule().priority_scheduling_preemption_threshold
        )
        self.enable_lora = get_lora().enable_lora
        self.enable_lora_overlap_loading = get_lora().enable_lora_overlap_loading
        self.max_loras_per_batch = get_lora().max_loras_per_batch
        self.enable_overlap = (
            not get_schedule().disable_overlap_schedule and not use_mlx()
        )
        self.enable_overlap_mlx = (
            not get_schedule().disable_overlap_schedule and use_mlx()
        )
        self.enable_pdmux = get_disagg().enable_pdmux
        self.skip_tokenizer_init = get_serving().skip_tokenizer_init
        self.stream_interval = get_serving().stream_interval
        self.spec_algorithm = SpeculativeAlgorithm.from_string(
            get_spec().speculative_algorithm
        )
        self.page_size = get_schedule().page_size
        self.enable_hierarchical_cache = get_memory().enable_hierarchical_cache
        self.enable_session_radix_cache = get_memory().enable_session_radix_cache
        self.enable_hicache_storage = get_memory().hicache_storage_backend is not None
        self.enable_unified_cache_external_linker = (
            get_memory().enable_unified_cache_external_linker
        )
        self.enable_decode_hicache = (
            get_disagg().disaggregation_decode_enable_radix_cache
            and self.enable_hierarchical_cache
        )
        self.max_recv_per_poll = envs.SGLANG_SCHEDULER_MAX_RECV_PER_POLL.get()
        self.max_new_tokens_limit = envs.SGLANG_MAX_NEW_TOKENS_LIMIT.get()
        self.enable_hisparse = get_memory().enable_hisparse
        self.enable_dp_attention = get_parallel().enable_dp_attention
        self.enable_unified_memory = get_memory().enable_unified_memory

        # Distributed rank info
        attn_tp_rank, attn_tp_size, attn_dp_rank, attn_dp_size = (
            compute_dp_attention_world_info(
                get_parallel().enable_dp_attention,
                tp_rank,
                get_parallel().tp_size,
                get_parallel().dp_size,
                get_parallel().attn_cp_size,
            )
        )
        self.ps = ParallelState(
            tp_rank=tp_rank,
            tp_size=get_parallel().tp_size,
            pp_rank=pp_rank,
            pp_size=get_parallel().pp_size,
            dp_rank=dp_rank,
            dp_size=get_parallel().dp_size,
            attn_tp_rank=attn_tp_rank,
            attn_tp_size=attn_tp_size,
            attn_cp_rank=attn_cp_rank,
            attn_cp_size=get_parallel().attn_cp_size,
            attn_dcp_rank=tp_rank % get_parallel().dcp_size,
            attn_dcp_size=get_parallel().dcp_size,
            attn_dp_rank=attn_dp_rank,
            attn_dp_size=attn_dp_size,
            moe_ep_rank=moe_ep_rank,
            moe_ep_size=get_parallel().ep_size,
            moe_dp_rank=moe_dp_rank,
            moe_dp_size=get_parallel().moe_dp_size,
            gpu_id=gpu_id,
        )

        # Init model configs
        self.init_model_config()

        # Init metrics stats
        self.init_metrics_collector(tp_rank, pp_rank, dp_rank)

        # Init inter-process communication
        self.init_ipc_channels(port_args)
        self.init_idle_sleeper()

        # Init ZBAL, switch allocator should before any torch alloc action
        self.init_zbal_on_npu()

        # Init PD-multiplexing context
        if self.enable_pdmux:
            self.init_pdmux()

        # Init tokenizer
        self.init_tokenizer()

        # Init moe config and GEMM config (FP8 GEMM, etc.)
        self.init_moe_gemm_config()

        # Init mamba backend
        self.init_mamba_backend()

        # Must precede init_model_worker: revert targets like _init_pools run during it,
        # so patching them afterwards is a no-op.
        maybe_revert_pr_fix()

        # Launch a model worker and draft model worker if using speculative decoding
        self.init_model_worker()

        if (t := envs.SGLANG_TEST_STUCK_SCHEDULER_INIT.get()) > 0:
            time.sleep(t)

        # Init cache and memory pool
        result = kv_cache_builder.build_kv_cache(
            server_args=self.server_args,
            model_config=self.model_config,
            tp_worker=self.tp_worker,
            page_size=self.page_size,
            spec_algorithm=self.spec_algorithm,
            attn_tp_cpu_group=self.attn_tp_cpu_group,
            tp_cpu_group=self.tp_cpu_group,
            attn_cp_cpu_group=self.attn_cp_cpu_group,
            enable_metrics=get_observability().enable_metrics,
            enable_kv_cache_events=bool(
                get_observability().kv_events_config
                and get_parallel().pp_rank == 0
                and get_parallel().attn_tp_rank == 0
                and get_parallel().attn_cp_rank == 0
            ),
            tp_group=self.tp_group,
            pp_group=self.pp_group,
            enable_hierarchical_cache=self.enable_hierarchical_cache,
            hicache_draft_plan=(
                self.draft_worker.hicache_draft_plan
                if self.draft_worker is not None
                else None
            ),
        )
        self.is_hybrid_swa = result.is_hybrid_swa
        self.is_hybrid_ssm = result.is_hybrid_ssm
        self.sliding_window_size = result.sliding_window_size
        self.full_tokens_per_layer = result.full_tokens_per_layer
        self.swa_tokens_per_layer = result.swa_tokens_per_layer
        self.req_to_token_pool = result.req_to_token_pool
        self.token_to_kv_pool_allocator = result.token_to_kv_pool_allocator
        self.disable_radix_cache = result.disable_radix_cache
        self.tree_cache = result.tree_cache
        if self.enable_hierarchical_cache:
            cache_controller = self.tree_cache.cache_controller
            if cache_controller is not None:
                cache_controller.load_fence_stream = (
                    self.tp_worker.model_runner.forward_stream
                )
        self.emit_metrics_constants()
        self.maybe_init_hccl_dp_prewarm()

        if (c := self.tp_worker.model_runner.canary_manager) is not None:
            c.attach_radix_cache(self.tree_cache)

        self.init_hisparse_coordinator()

        if (
            get_disagg().disaggregation_mode == "decode"
            and get_disagg().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.enable_dp_attention
                    else self.tp_cpu_group
                ),
                tree_cache=self.tree_cache,
            )
        else:
            self.decode_offload_manager = None

        # Init running status
        self.init_running_status()

        # Init chunked prefill
        self.init_chunked_prefill()
        self.maybe_init_dynamic_chunk_sizer()

        # Init diffusion LLM
        self.init_diffusion_llm()

        self.init_metrics_reporter(tp_rank, pp_rank, dp_rank)
        self.scheduler_stage_metrics = self.metrics_reporter.scheduler_stage_metrics

        # Init schedule policy and new token estimation
        self.init_schedule_policy()

        # Init watchdog, memory saver, input blocker and recv skipper
        self.init_watch_dog_memory_saver_input_blocker()

        # Init profiler
        self.init_profiler()

        # Start the embedded Rust frontend (rank 0). Must precede
        # init_disaggregation: on PD prefill the rust api listener also serves
        # the KV bootstrap registry, and the KVManagers built there register to
        # it synchronously. (The listener is bound synchronously inside launch,
        # so the registry is accepting once this returns.) Must also precede
        # the request receiver, which reads self.recv_from_tokenizer to pick
        # its ingress transport.
        self.maybe_init_rust_server()

        # Init prefill-decodedisaggregation
        self.init_disaggregation()

        # Init overlap schedule
        self.init_overlap()

        # Init Ngram Embedding
        self.maybe_init_ngram_embedding()

        # Init prefill kv split size when deterministic inference is enabled with various attention backends
        self.init_deterministic_inference_config()
        self.init_dsa_kpool_truncation_align()

        self.init_weight_updater()

        # Init request dispatcher
        self.init_request_dispatcher()

        # Init LoRA drainer for fair scheduling
        self.init_lora_drainer()

        # Init LoRA overlap loader
        self.init_lora_overlap_loader()

        # Init the grammar backend for constrained generation
        self.init_grammar_manager()

        self.maybe_init_scripted_scheduler_hook()

        self.init_request_receiver()

        self.init_dp_attn_adapter()

        self.init_pool_stats_observer()

        self.init_invariant_checker()

        self.init_kv_events_publisher()

        self.init_load_publisher()

        self.init_load_inquirer()

        self.init_output_streamer()

        self.init_batch_result_processor()

        self.init_rank_consensus_checker()

        self.is_initializing = False
        self.init_startup_timing_summary()

    def init_startup_timing_begin(self) -> None:
        self.scheduler_startup_begin = time.perf_counter()

    def init_startup_timing_summary(self) -> None:
        self.startup_time = build_scheduler_startup_time(
            target_load_weight=self.tp_worker.weight_load_time,
            draft_load_weight=(
                0.0 if self.draft_worker is None else self.draft_worker.weight_load_time
            ),
            kv_cache_allocation=self.kv_cache_allocation_time,
            scheduler_e2e=time.perf_counter() - self.scheduler_startup_begin,
            target_cuda_graph=self.tp_worker.graph_time_usage,
            draft_cuda_graph=(
                None
                if self.draft_worker is None
                else self.draft_worker.graph_time_usage
            ),
        )

    def maybe_init_hccl_dp_prewarm(self) -> None:
        if not (
            _is_npu
            and is_deepseek_v4(self.tp_worker.model_runner.model_config.hf_config)
        ):
            return

        rank = (
            get_parallel().dp_rank
            if get_parallel().dp_rank is not None
            else self.tp_group.rank_in_group
        )
        logger.info("HCCL DP prewarm start: rank=%s", rank)
        _prewarm_hccl_group(
            device=self.tp_group.device,
            group=self.tp_group.device_group,
            device_module=self.tp_group.device_module,
        )
        logger.info("HCCL DP prewarm done: rank=%s", rank)

    def init_zbal_on_npu(self):
        if _is_npu:
            from sglang.srt.hardware_backend.npu.utils import init_zbal

            if get_parallel().pp_size > 1:
                logger.error("only zbal mix mode support pp_size > 1!")
            init_zbal(
                get_parallel().tp_size, self.ps.gpu_id, get_parallel().tp_rank
            )  # only switch allocator if is mix mode

    def init_model_config(self):
        self.model_config = ModelConfig.from_server_args(self.server_args)
        if _is_npu:
            # make sure the page size is not larger than block_size and chunked_prefill_size on NPU backend
            # the npu backend request the defined page size to be no larger than block_size and chunked_prefill_size
            from sglang.srt.dllm.config import DllmConfig

            self.dllm_config = (  # For diffusion LLM
                DllmConfig.from_server_args(self.server_args)
                if get_exec().dllm.dllm_algorithm is not None
                else None
            )

    def init_metrics_collector(
        self, tp_rank: int, pp_rank: int, dp_rank: Optional[int]
    ) -> None:
        self.metrics_collector_context = SchedulerMetricsCollector.init_new(
            server_args=self.server_args,
            tp_rank=tp_rank,
            pp_rank=pp_rank,
            dp_rank=dp_rank,
            enable_priority_scheduling=self.enable_priority_scheduling,
            enable_lora=self.enable_lora,
            enable_hierarchical_cache=self.enable_hierarchical_cache,
        )
        self.metrics_collector = self.metrics_collector_context.collector

    def init_ipc_channels(self, port_args: PortArgs):
        is_rank_zero = (
            get_parallel().pp_rank == 0
            and get_parallel().attn_tp_rank == 0
            and get_parallel().attn_cp_rank == 0
        )
        self.ipc_channels = SchedulerIpcChannels.create(
            port_args=port_args,
            is_rank_zero=is_rank_zero,
            # The snapshot taken at construction, not a second bag read: this
            # scheduler gates its tokenizer init on the same value, and the two
            # must not be able to disagree.
            skip_tokenizer_init=self.skip_tokenizer_init,
            metrics_enabled=get_observability().enable_metrics
            and (
                get_parallel().attn_tp_rank == 0
                or get_observability().enable_metrics_for_all_schedulers
            ),
            enable_scripted_runtime=envs.SGLANG_TEST_SCRIPTED_RUNTIME.get(),
        )

        self.load_snapshot_writer = None
        self.recv_from_tokenizer = None

        if not is_rank_zero:
            return

        self.recv_from_tokenizer = self.ipc_channels.recv_from_tokenizer
        dp_rank = get_parallel().dp_rank if get_parallel().dp_rank is not None else 0
        try:
            self.load_snapshot_writer = create_load_snapshot_writer(
                port_args,
                get_parallel().dp_size,
                dp_rank,
                publish_interval=get_observability().load_snapshot_publish_interval,
            )
        except Exception as e:
            logger.warning("load snapshot writer init failed: %s", e)

    def init_idle_sleeper(self) -> None:
        if (
            get_parallel().pp_rank == 0
            and get_parallel().attn_tp_rank == 0
            and get_parallel().attn_cp_rank == 0
            and get_device().sleep_on_idle
        ):
            self.idle_sleeper = IdleSleeper(
                sockets=[
                    self.ipc_channels.recv_from_tokenizer,
                    self.ipc_channels.recv_from_rpc,
                ],
            )
        else:
            self.idle_sleeper = None

    def publish_load_snapshot(self, force: bool = False):
        """Returns the LoadSnapshot it published, or None when disabled,
        throttled, or failed — so co-located sinks (the router-facing load
        publisher) can reuse it instead of walking the queues again."""
        writer = self.load_snapshot_writer
        if writer is None:
            return None
        if not force:
            writer.publish_counter += 1
            if writer.publish_counter < writer.publish_interval:
                return None
        writer.publish_counter = 0
        try:
            load = self.load_inquirer.get_loads()
            writer.write(load)
            return load
        except Exception as e:
            logger.warning("load snapshot publish failed: %s", e)
            return None

    def init_tokenizer(self):
        server_args = self.server_args
        self.is_generation = self.model_config.is_generation

        if self.skip_tokenizer_init:
            self.tokenizer = self.processor = None
        else:
            if self.model_config.is_multimodal:
                self.processor = get_processor(
                    get_serving().tokenizer_path,
                    tokenizer_mode=get_serving().tokenizer_mode,
                    trust_remote_code=get_model().trust_remote_code,
                    revision=get_model().revision,
                    image_processor_backend=resolve_image_processor_backend(get_mm()),
                    tokenizer_backend=get_serving().tokenizer_backend,
                    model_name=get_model().model_path,
                )
                self.tokenizer = get_tokenizer_from_processor(self.processor)
            else:
                self.tokenizer = get_tokenizer(
                    get_serving().tokenizer_path,
                    tokenizer_mode=get_serving().tokenizer_mode,
                    trust_remote_code=get_model().trust_remote_code,
                    revision=get_model().revision,
                    tokenizer_backend=get_serving().tokenizer_backend,
                )

        # Load multimodal processor for M-RoPE fallback computation.
        self._mm_processor = None
        if (
            self.model_config.is_multimodal
            and self.processor is not None
            and not get_disagg().language_model_only
        ):
            try:
                import_processors("sglang.srt.multimodal.processors")
                self._mm_processor = get_mm_processor(
                    self.model_config.hf_config,
                    server_args,
                    self.processor,
                    "default",
                    skip_mm_pool=True,
                )
            except Exception:
                logger.warning(
                    "Failed to load multimodal processor in scheduler; "
                    "M-RoPE fallback will not be available."
                )

        if get_serving().reasoning_parser and self.tokenizer:
            reasoning_parser = ReasoningParser(
                model_type=get_serving().reasoning_parser,
                stream_reasoning=False,
                tokenizer=self.tokenizer,
            )
            think_end_ids = self.tokenizer.encode(
                reasoning_parser.detector.think_end_token, add_special_tokens=False
            )
            if think_end_ids:
                self.model_config.think_end_ids = think_end_ids
            else:
                logger.warning(
                    "Reasoning parser think_end_token %r could not be encoded; "
                    "grammar-gated reasoning is disabled.",
                    reasoning_parser.detector.think_end_token,
                )

            selectable_tokens = getattr(
                reasoning_parser.detector,
                "request_selectable_think_end_tokens",
                (),
            )
            if selectable_tokens:
                selectable_sequences = []
                for end_token in selectable_tokens:
                    token_ids = self.tokenizer.encode(
                        end_token, add_special_tokens=False
                    )
                    if not token_ids:
                        raise ValueError(
                            f"Request-selectable reasoning terminator {end_token!r} "
                            "could not be encoded"
                        )
                    selectable_sequences.append(token_ids)
                self.model_config.request_selectable_think_end_id_sequences = [
                    list(sequence)
                    for sequence in dict.fromkeys(
                        tuple(sequence) for sequence in selectable_sequences
                    )
                ]

    def init_mamba_backend(self) -> None:
        if initialize_mamba_selective_state_update_backend is not None:
            initialize_mamba_selective_state_update_backend(self.server_args)

    def init_moe_gemm_config(self):
        # Use the language config already normalized by ModelConfig. Multimodal
        # wrappers expose it under different attributes (for example,
        # ``text_config`` or ``llm_config``).
        config_to_check = self.model_config.hf_text_config

        # Different MoE architectures expose the per-token expert count under
        # different attribute names (e.g. Gemma4 uses ``top_k_experts``,
        # LongCat-2.0 uses ``moe_topk``).
        moe_topk_attrs = (
            "num_experts_per_tok",
            "num_experts_per_token",
            "top_k_experts",
            "moe_top_k",
            "moe_topk",
        )
        if any(hasattr(config_to_check, attr) for attr in moe_topk_attrs):
            initialize_moe_config()

        # Initialize GEMM-related configuration for FP8 and FP4 backends.
        initialize_fp8_gemm_config()
        initialize_fp4_gemm_config()
        initialize_bf16_gemm_config()

        # This must be called after initialize_moe_config
        self.require_mlp_sync = require_mlp_sync()

    def init_tp_model_worker(self):
        worker_kwargs = dict(
            server_args=self.server_args,
            gpu_id=self.ps.gpu_id,
            ps=self.ps,
            nccl_port=self.nccl_port,
        )

        # FIXME: move tp worker's init logic outside of the scheduler.
        if use_mlx():
            from sglang.srt.hardware_backend.mlx.tp_worker import MlxTpModelWorker

            self.tp_worker = MlxTpModelWorker(**worker_kwargs)
        else:
            from sglang.srt.managers.tp_worker import TpModelWorker

            self.tp_worker = TpModelWorker(**worker_kwargs)

    def maybe_init_draft_worker(self):
        if self.spec_algorithm.is_none():
            self.draft_worker = None
            self.external_corpus_manager = None
            return

        if (
            envs.SGLANG_ENABLE_PP_SPEC.get()
            and get_parallel().pp_size > 1
            and get_parallel().pp_rank != get_parallel().pp_size - 1
        ):
            # PP+spec: the draft model (MTP layer) needs final hidden states and
            # the lm_head, both of which live on the last PP stage only.
            self.draft_worker = None
            self.external_corpus_manager = None
            return

        # Launch a draft worker for speculative decoding. It builds its draft
        # from this process's own config: what differs for the draft — the
        # target's context length, the draft load format, its attention backend
        # — is resolved per runner, not on a config copy.
        draft_worker_kwargs = dict(
            server_args=self.server_args,
            gpu_id=self.ps.gpu_id,
            ps=self.ps,
            nccl_port=self.nccl_port,
            target_worker=self.tp_worker,
        )

        DraftWorkerClass = self.spec_algorithm.create_worker(self.server_args)
        self.draft_worker = DraftWorkerClass(**draft_worker_kwargs)

        if self.spec_algorithm.is_ngram():
            from sglang.srt.speculative.external_corpus_manager import (
                ExternalCorpusManager,
            )

            self.external_corpus_manager = ExternalCorpusManager(
                self.draft_worker,
                self.ipc_channels.send_to_tokenizer.send_output,
            )
        else:
            self.external_corpus_manager = None

    def init_target_memory_pool(self):
        """Allocate target KV cache pools if they have not been allocated yet."""
        if (
            self.tp_worker.model_runner.memory_pool_config is not None
            and self.tp_worker.model_runner.req_to_token_pool is not None
            and self.tp_worker.model_runner.token_to_kv_pool_allocator is not None
        ):
            return
        preloaded_weights_bytes = self.tp_worker.preloaded_weights_bytes
        if self.draft_worker is not None:
            preloaded_weights_bytes += self.draft_worker.preloaded_weights_bytes
        self.tp_worker.model_runner.account_preloaded_weights(preloaded_weights_bytes)
        self.tp_worker.alloc_memory_pool()

    def init_memory_pools(self):
        """Allocate KV cache pools for target and draft workers."""
        self.init_target_memory_pool()
        # Lands the retraction backend on the disagg bag before the draft
        # worker's HiCache plan reads it.
        kv_cache_builder.resolve_decode_retraction_backup(tp_worker=self.tp_worker)
        if self.draft_worker is not None:
            pool, allocator = self.tp_worker.get_memory_pool()
            self.draft_worker.alloc_memory_pool(
                memory_pool_config=self.tp_worker.model_runner.memory_pool_config,
                req_to_token_pool=pool,
                token_to_kv_pool_allocator=allocator,
            )
            self.draft_worker.init_hicache_draft_plan()

    def init_all_attention_backends(self):
        """Initialize attention backends for all workers."""
        self.tp_worker.init_attention_backends()
        if self.draft_worker is not None:
            self.draft_worker.init_attention_backends()

    def init_all_cuda_graphs(self):
        """Capture cuda graphs for all workers."""
        self.tp_worker.init_cuda_graphs()
        if self.draft_worker is not None:
            self.draft_worker.init_cuda_graphs()

    def init_model_worker(self):
        # Load model weights.
        self.init_tp_model_worker()
        if get_model().is_startup_weight_load_overlap:
            self.tp_worker.start_startup_weight_load()
        self.maybe_init_draft_worker()

        # Prepare KV cache pools for all workers
        tic = time.perf_counter()
        self.init_memory_pools()
        self.kv_cache_allocation_time = time.perf_counter() - tic

        self.init_all_attention_backends()
        self.init_all_cuda_graphs()

        model_runner = self.tp_worker.model_runner
        if model_runner.token_to_kv_pool.post_capture_active:
            kv_cache_builder.prepare_hicache_staging(
                tp_worker=self.tp_worker,
                draft_plan=(
                    self.draft_worker.hicache_draft_plan
                    if self.draft_worker is not None
                    else None
                ),
            )
        device_module = torch.get_device_module(model_runner.device)
        self.schedule_stream = None if use_mlx() else device_module.Stream(priority=0)
        # Match run_batch / _pp_launch_batch so warmup allocations stay reusable.
        forward_stream = (
            model_runner.forward_stream
            if self.enable_overlap or get_parallel().pp_size > 1 or use_mlx()
            else self.schedule_stream
        )
        with device_module.stream(forward_stream):
            prewarm_graph_pool_borrow()
            if self.draft_worker is None:
                model_runner.prewarm_sampling()
            else:
                self.draft_worker.prewarm_sampling()
        if model_runner.token_to_kv_pool.post_capture_active:
            tic = time.perf_counter()
            model_runner.post_capture_resize_kv_pool(
                draft_runners=(
                    self.draft_worker._draft_model_runners()
                    if self.draft_worker is not None
                    else ()
                )
            )
            self.kv_cache_allocation_time += time.perf_counter() - tic

        if get_model().is_startup_weight_load_overlap:
            self.tp_worker.finalize_startup_weight_load()

        # Adaptive/speculative graphs and post-capture KV sizing can consume
        # the headroom seen by the initial DeepGEMM layout budget. Refresh it
        # after these allocations, before elastic EP rejoins healthy ranks
        # that do not participate in this startup collective.
        from sglang.srt.model_executor.model_runner_components.cuda_graph_setup import (
            refresh_deep_gemm_layout_memory_budget,
        )

        refresh_deep_gemm_layout_memory_budget(model_runner, only_if_initialized=True)

        if (
            get_exec().moe.elastic_ep_backend is not None
            and get_exec().moe.ep_join_mode == "recover"
        ):
            model_runner.post_capture_elastic_ep_recover()

        # Dispatch the model worker
        if self.spec_algorithm.is_none() or self.draft_worker is None:
            # PP+spec: non-last stages have no draft worker; they run the
            # verify-shaped target forward through the plain tp_worker.
            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.forward_stream,
            _,
            _,
            _,
        ) = self.tp_worker.get_worker_info()
        # DFlash auto-enables the legacy formula; other workloads opt in via
        # --min-free-slots-delay. Built independently of the prefill delayer.
        self.min_free_slots_delayer: Optional[MinFreeSlotsDelayer] = None
        min_free_slots = resolve_min_free_slots(
            get_schedule().min_free_slots_delay,
            self.max_running_requests,
            is_dflash_family=self.spec_algorithm.is_dflash_family(),
        )
        if min_free_slots is not None:
            self.min_free_slots_delayer = MinFreeSlotsDelayer(
                min_free_slots=min_free_slots
            )
        if not get_parallel().pp_max_micro_batch_size:
            get_context().override(
                "scheduler.pp_max_micro_batch_size_default",
                pp_max_micro_batch_size=max(
                    self.max_running_requests // get_parallel().pp_size, 1
                ),
            )

        self.tp_group = get_parallel().tp_group
        self.tp_cpu_group = self.tp_group.cpu_group
        self.attn_tp_group = get_parallel().attn_tp_group
        self.attn_tp_cpu_group = self.attn_tp_group.cpu_group
        self.attn_cp_group = get_parallel().attn_cp_group
        self.attn_cp_cpu_group = self.attn_cp_group.cpu_group
        self.pp_group = get_parallel().pp_group
        self.world_group = get_parallel().world_group

        # NOTE: dp_tp_* are request/data-plane coordination groups (not tensor collectives).
        # When DP attention is enabled, scope to the attention-TP group; otherwise use
        # the base TP group. Entry rank is the local rank 0 in that group.
        # Use the CPU (gloo) group to broadcast VLM Python objects and avoid CUDA
        # stream/device coupling (#11910).
        self.dp_tp_group = (
            self.attn_tp_group if self.enable_dp_attention else self.tp_group
        )
        self.dp_tp_cpu_group = self.dp_tp_group.cpu_group

        # TODO(Jialin): Migrate pad_input_ids implementations to return array.
        self.pad_input_ids_func = self.tp_worker.get_pad_input_ids_func()
        set_random_seed(self.random_seed)

        # Print debug info
        self.startup_available_gpu_memory_gb = get_available_gpu_memory(
            self.device, self.ps.gpu_id, empty_cache=False
        )
        if get_parallel().tp_rank == 0:
            logger.info(
                f"max_total_num_tokens={self.max_total_num_tokens}, "
                f"chunked_prefill_size={get_schedule().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'}="
                f"{self.startup_available_gpu_memory_gb:.2f} GB"
            )

    def emit_metrics_constants(self) -> None:
        if not get_observability().enable_metrics:
            return

        self.metrics_collector.emit_constants(
            max_total_num_tokens=self.max_total_num_tokens,
            max_total_num_tokens_swa=self.swa_tokens_per_layer,
            weight_memory_usage_gb=self.tp_worker.model_runner.weight_load_mem_usage,
            kv_cache_memory_usage_gb=(
                self.token_to_kv_pool_allocator.get_kvcache().mem_usage
            ),
            graph_memory_usage_gb=combine_graph_memory_usage(
                self.tp_worker.graph_memory_usage,
                (
                    None
                    if self.draft_worker is None
                    else self.draft_worker.graph_memory_usage
                ),
            ),
            # TODO: max_running_requests_under_SLO has no setter — dead chain.
            max_running_requests_under_SLO=None,
            page_size=self.page_size,
            num_pages=self.max_total_num_tokens // self.page_size,
            context_len=self.model_config.context_len,
            startup_available_gpu_memory_gb=self.startup_available_gpu_memory_gb,
        )

    def init_hisparse_coordinator(self) -> None:
        self.hisparse_coordinator: Optional[HiSparseCoordinator] = None
        if not self.enable_hisparse:
            return

        # Coordinator was created inside ModelRunner.initialize() before CUDA graph capture.
        self.hisparse_coordinator = self.tp_worker.model_runner.hisparse_coordinator
        self.hisparse_coordinator.set_decode_producer_stream(self.forward_stream)

    def init_running_status(self):
        # Set by a runtime PD role switch to break out of the current event loop.
        self._event_loop_should_restart = False
        # Guards against concurrent/re-entrant PD role switches.
        self._pd_role_switch_in_progress = False
        # Set if a role switch tore down the old role but failed to rebuild
        # either the new or the old role; the instance can no longer serve.
        self._pd_role_switch_unhealthy = False
        # Set by the ShutdownReq handler to break the event loop for graceful shutdown.
        self.gracefully_exit = False
        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_for_debug: Optional[ScheduleBatch] = None
        # The last forward batch
        self.last_batch: Optional[ScheduleBatch] = None
        self.forward_ct = 0
        self.return_health_check_ipcs: Deque[Optional[str]] = deque()
        self.flush_wrapper = SchedulerFlushWrapper(
            flush_cache=self.flush_cache,
            is_fully_idle=self.is_fully_idle,
            ipc_channels=self.ipc_channels,
        )
        self._last_logged_elastic_radix_namespace: Optional[str] = None
        self.session_controller = SessionController(self.tree_cache)
        self.forward_sleep_time = None
        self._engine_paused = False

    def init_chunked_prefill(self):
        self.chunked_prefill_size = get_schedule().chunked_prefill_size
        self.prefill_decode_interval = get_schedule().prefill_decode_interval or 0
        self._prefill_decode_interval_remaining = 0
        uses_transformers_backend = (
            get_resolved_model_impl(self.model_config) == ModelImpl.TRANSFORMERS
        )
        if (
            self.chunked_prefill_size is not None
            and self.chunked_prefill_size > 0
            and self.model_config.is_multimodal
            and uses_transformers_backend
        ):
            logger.warning(
                "Chunked prefill is disabled for multimodal models with the "
                "Transformers backend to avoid partial multimodal chunk mismatches."
            )
            self.chunked_prefill_size = None
        elif self.chunked_prefill_size is not None and self.chunked_prefill_size <= 0:
            self.chunked_prefill_size = None
        self.chunked_req = None
        self._pending_chunked_abort_req = None
        self.is_mixed_chunk = (
            self.chunked_prefill_size is not None and get_schedule().enable_mixed_chunk
        )

    def maybe_init_dynamic_chunk_sizer(self) -> None:
        """Profile a PP prefill latency model that sizes chunks per stage."""
        self.dynamic_chunk_sizer: Optional[DynamicChunkSizer] = None
        if not (get_schedule().enable_dynamic_chunking and get_parallel().pp_size > 1):
            return
        sizer = DynamicChunkSizer(
            model_runner=self.tp_worker.model_runner,
            model_config=self.model_config,
            tree_cache=self.tree_cache,
            req_to_token_pool=self.req_to_token_pool,
            token_to_kv_pool_allocator=self.token_to_kv_pool_allocator,
            spec_algorithm=self.spec_algorithm,
            chunked_prefill_size=self.chunked_prefill_size,
            max_prefill_tokens=self.max_prefill_tokens,
            page_size=self.page_size,
            device=self.device,
            pp_group=self.pp_group,
            world_group=self.world_group,
            pp_rank=get_parallel().pp_rank,
        )
        if sizer.profile_and_fit():
            self.dynamic_chunk_sizer = sizer

    def _should_defer_prefill(self) -> bool:
        if self._prefill_decode_interval_remaining == 0:
            return False

        self._prefill_decode_interval_remaining -= 1
        return True

    def _arm_prefill_decode_interval(self, batch: Optional[ScheduleBatch]) -> None:
        if self.prefill_decode_interval == 0 or batch is None:
            return

        # DP attention synchronizes this flag across ranks. This keeps every
        # rank on the same prefill/decode cadence even when only one rank has
        # local prefill work. Non-DP scheduling can use the local mode directly.
        is_extend = (
            batch.is_extend_in_batch
            if self.require_mlp_sync
            else batch.forward_mode.is_extend()
        )
        if is_extend:
            self._prefill_decode_interval_remaining = self.prefill_decode_interval

    def init_metrics_reporter(
        self, tp_rank: int, pp_rank: int, dp_rank: Optional[int]
    ) -> None:
        # Override point for deployments that need a specialized reporter.
        self.metrics_reporter = SchedulerMetricsReporter(
            scheduler=self,
            tp_rank=tp_rank,
            pp_rank=pp_rank,
            dp_rank=dp_rank,
            metrics_collector_context=self.metrics_collector_context,
            metrics_collector=self.metrics_collector,
        )

    def init_schedule_policy(self):
        # 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,
        )
        self.prefill_delayer: Optional[PrefillDelayer] = None
        self.prefill_bs_tracker = RecentPrefillBatchSizeTracker(
            window_size=envs.SGLANG_PREFILL_DELAYER_MAX_PREFILL_BS_WINDOW_SIZE.get()
        )
        self.max_prefill_bs: int = 0
        if get_schedule().enable_prefill_delayer:
            if get_disagg().disaggregation_mode == "decode":
                logger.info(
                    "Ignoring --enable-prefill-delayer on decode engine "
                    "(no prefill scheduling path; delayer would be a no-op)."
                )
            else:
                self.prefill_delayer = PrefillDelayer(
                    dp_size=get_parallel().dp_size,
                    attn_tp_size=get_parallel().attn_tp_size,
                    cpu_group=self.tp_cpu_group,
                    device_group=self.tp_group.device_group,
                    metrics_collector=(
                        self.metrics_collector
                        if self.metrics_reporter.enable_metrics
                        else None
                    ),
                    max_delay_passes=get_schedule().prefill_delayer_max_delay_passes,
                    token_usage_low_watermark=get_schedule().prefill_delayer_token_usage_low_watermark,
                    device=self.tp_group.device,
                    debug_log_enabled=get_parallel().attn_tp_rank == 0,
                )

        # NOTE: preemption is enabled by default for priority scheduling.
        self.enable_priority_preemption = (
            self.enable_priority_scheduling
            and not get_schedule().disable_priority_preemption
        )

        self.new_token_ratio_tracker = NewTokenRatioTracker.from_config()

    def init_soft_watchdog(self):
        if (x := get_device().soft_watchdog_timeout) is not None:
            self.soft_watchdog = create_scheduler_watchdog(
                self, watchdog_timeout=x, soft=True
            )

    def init_watch_dog_memory_saver_input_blocker(self):
        # Start watchdog thread
        self.watchdog = create_scheduler_watchdog(
            self, watchdog_timeout=get_device().watchdog_timeout
        )

        # Init memory saver, profiler and metric stats
        self.memory_saver_adapter = TorchMemorySaverAdapter.create(
            enable=get_exec().features.enable_memory_saver
        )

        # Init recv skipper and input blocker
        self.recv_skipper = SchedulerRecvSkipper.maybe_create()
        self.input_blocker = (
            SchedulerInputBlocker(noop=get_parallel().attn_tp_rank != 0)
            if get_bool_env_var("SGLANG_ENABLE_COLOCATED_BATCH_GEN")
            else None
        )

        # Configure GC logger
        if envs.SGLANG_LOG_GC.get():
            configure_gc_logger()

    def init_disaggregation(self):
        self.mm_receiver = None
        self.disagg_prefill_bootstrap_queue = None
        self.disagg_prefill_inflight_queue = None
        self.disagg_decode_prealloc_queue = None
        self.disagg_decode_transfer_queue = None

        self.disaggregation_mode = DisaggregationMode(get_disagg().disaggregation_mode)
        self.transfer_backend = TransferBackend(
            get_disagg().disaggregation_transfer_backend
        )

        # In rust-server mode the KV bootstrap registry is already serving on
        # the rust api listener (maybe_init_rust_server runs before this
        # method — the PrefillBootstrapQueue's KVManager below registers to it
        # synchronously, and a failed registration only retries ~60s then logs,
        # leaving every PD request unroutable). Only the ascend config store,
        # which start_disagg_service would otherwise create, is left to do.
        if (
            self.disaggregation_mode == DisaggregationMode.PREFILL
            and self._hosts_rust_server()
        ):
            maybe_create_ascend_config_store(
                transfer_backend=self.transfer_backend,
            )

        draft_token_to_kv_pool = (
            self.draft_worker.primary_draft_kv_pool
            if self.draft_worker is not None
            else None
        )

        if self.spec_algorithm.carries_draft_hidden_states():
            # Derive the rank-uniform PD wire schema from config because only the
            # last prefill PP stage owns a draft runner.
            draft_model_config = ModelConfig.from_server_args(
                self.server_args,
                model_path=get_spec().speculative_draft_model_path,
                model_revision=get_spec().speculative_draft_model_revision,
                is_draft_model=True,
            )
            disagg_hidden_size, disagg_hidden_states_dtype = (
                get_draft_recurrent_hidden_state_spec_from_config(
                    draft_model_config, self.spec_algorithm
                )
            )
        else:
            disagg_hidden_size = 16  # minimal padding size for RDMA
            disagg_hidden_states_dtype = torch.float32

        # The PD metadata wire schema must match on P and D even when only D
        # enables spec decoding; a seedless prefill writes the invalid sentinel.
        output_dsa_topk_indices_dim = get_dsa_seed_metadata_dim(
            self.model_config.hf_config
        )

        if (
            self.disaggregation_mode == DisaggregationMode.DECODE
        ):  # *8 headroom for MiniMax-M3; *2 for other models.
            buffer_multiplier = (
                8 if is_minimax_sparse(self.model_config.hf_config) else 2
            )
            buffer_size = (self.req_to_token_pool.size) * buffer_multiplier
            self.req_to_metadata_buffer_idx_allocator = ReqToMetadataIdxAllocator(
                buffer_size
            )
            self.disagg_metadata_buffers = MetadataBuffers(
                buffer_size,
                hidden_size=disagg_hidden_size,
                hidden_states_dtype=disagg_hidden_states_dtype,
                max_sampling_mask_tokens=self.server_args.sampling_mask_max_tokens,
                custom_mem_pool=self.token_to_kv_pool_allocator.get_kvcache().maybe_get_custom_mem_pool(),
                output_dsa_topk_indices_dim=output_dsa_topk_indices_dim,
                kv_checksum_enabled=get_disagg().disaggregation_enable_kv_checksum,
            )

            # 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=get_parallel().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=draft_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=get_parallel().tp_rank,
                tp_size=get_parallel().tp_size,
                dp_size=get_parallel().dp_size,
                gpu_id=self.ps.gpu_id,
                bootstrap_port=get_disagg().disaggregation_bootstrap_port,
                max_total_num_tokens=self.max_total_num_tokens,
                pp_rank=get_parallel().pp_rank,
                num_reserved_decode_tokens=get_disagg().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=disagg_hidden_size,
                hidden_states_dtype=disagg_hidden_states_dtype,
                max_sampling_mask_tokens=self.server_args.sampling_mask_max_tokens,
                custom_mem_pool=self.token_to_kv_pool_allocator.get_kvcache().maybe_get_custom_mem_pool(),
                output_dsa_topk_indices_dim=output_dsa_topk_indices_dim,
                kv_checksum_enabled=get_disagg().disaggregation_enable_kv_checksum,
            )

            self.disagg_prefill_bootstrap_queue = PrefillBootstrapQueue(
                token_to_kv_pool=self.token_to_kv_pool_allocator.get_kvcache(),
                draft_token_to_kv_pool=draft_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=get_parallel().tp_rank,
                tp_size=get_parallel().tp_size,
                gpu_id=self.ps.gpu_id,
                bootstrap_port=get_disagg().disaggregation_bootstrap_port,
                gloo_group=self.attn_tp_cpu_group,
                max_total_num_tokens=self.max_total_num_tokens,
                scheduler=self,
                scheduler_stage_metrics=self.scheduler_stage_metrics,
                pp_rank=get_parallel().pp_rank,
                pp_size=get_parallel().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] = []
            # Requests with a sent chunk that are not yet on the inflight queue.
            self.disagg_prefill_pending_chunk_rids: Set[str] = set()

            self.enable_staging = envs.SGLANG_DISAGG_STAGING_BUFFER.get()

        if (
            self.enable_unified_memory
            and self.disaggregation_mode != DisaggregationMode.NULL
        ):
            self.token_to_kv_pool_allocator.set_disagg_move_gate(
                unified_memory_disagg_move_gate(self)
            )

        # Init mm receiver for EPD disaggregation mode
        if get_disagg().language_only and get_disagg().encoder_transfer_backend in [
            "zmq_to_scheduler",
            "mooncake",
        ]:
            self.mm_receiver = create_mm_receiver(
                self.server_args,
                dtype=self.model_config.dtype,
                hf_config=self.model_config.hf_config,
                pp_rank=get_parallel().pp_rank,
                tp_rank=get_parallel().tp_rank,
                tp_group=self.tp_group,
                scheduler=self,
            )

    def init_overlap(self):
        self.device_module = torch.get_device_module(self.device)

        # FutureMap is always-on: input_ids relay used in both modes.
        # Workers without the spec_v2_attn_backends override fall back to
        # target-only so the helper still produces a safe decision (no
        # accidental opt-out for unaudited shapes).
        if self.draft_worker is not None:
            attn_backends = getattr(
                self.draft_worker,
                "spec_v2_attn_backends",
                (self.tp_worker.model_runner.attn_backend,),
            )
        else:
            attn_backends = (self.tp_worker.model_runner.attn_backend,)
        needs_cpu_seq_lens = decide_needs_cpu_seq_lens(attn_backends)
        needs_confidence_relay = decide_needs_confidence_relay()
        self.future_map = self.spec_algorithm.create_future_map(
            self.device,
            self.req_to_token_pool,
            needs_cpu_seq_lens=needs_cpu_seq_lens,
            needs_confidence_relay=needs_confidence_relay,
        )

        self._confidence_budget_prepare = None
        if (
            needs_confidence_relay
            and self.enable_overlap
            and self.draft_worker is not None
        ):
            self._confidence_budget_prepare = (
                self.draft_worker.get_confidence_budget_prepare()
            )

        if use_mlx():
            # MLX uses its own overlap loop and does not create CUDA streams,
            # but the normal non-overlap scheduler path still relays decode
            # input IDs through FutureMap.
            self.result_queue: Deque = deque()
            return

        # forward_stream_ctx / copy_stream are also used by PP (non-overlap)
        # via scheduler_pp_mixin; init unconditionally to match main.
        self.forward_stream_ctx: CudaStreamContext = self.device_module.stream(
            self.forward_stream
        )
        self.copy_stream: CudaStream = self.device_module.Stream()
        self.copy_stream_ctx: CudaStreamContext = self.device_module.stream(
            self.copy_stream
        )

        if not self.enable_overlap:
            return

        self.batch_record_buf = [None] * 2
        self.batch_record_ct = 0

    def maybe_init_ngram_embedding(self):
        self.ngram_embedding_manager = (
            self.tp_worker.model_runner.ngram_embedding_manager
        )
        self.use_ngram_embedding = self.tp_worker.model_config.use_ngram_embedding

    def init_deterministic_inference_config(self):
        """Initialize deterministic inference configuration for different attention backends."""
        if not get_exec().deterministic.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),
        }
        # Both entries are prefill knobs (SPLIT_TILE / PREFILL_TRUNCATION):
        # the prefill half decides.
        prefill_backend, _ = attention_backends()
        env_var, default_size = backend_sizes.get(prefill_backend, (None, None))
        self.truncation_align_size = (
            get_int_env_var(env_var, default_size) if env_var else None
        )

    def init_dsa_kpool_truncation_align(self):
        """Kpool compress-write asserts chunked extends start on pool boundaries.
        Use the LCM to preserve any existing deterministic-inference alignment."""
        from sglang.srt.configs.model_config import (
            get_dsa_index_kpool,
            is_deepseek_dsa,
        )

        if not is_deepseek_dsa(self.model_config.hf_config):
            return

        dsa_index_kpool = get_dsa_index_kpool(self.model_config.hf_config)
        if dsa_index_kpool <= 1:
            return

        if self.truncation_align_size is None:
            self.truncation_align_size = dsa_index_kpool
        else:
            self.truncation_align_size = math.lcm(
                self.truncation_align_size, dsa_index_kpool
            )

    def init_request_dispatcher(self):
        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_wrapper.handle),
                (ClearHiCacheReqInput, self.clear_hicache_storage_wrapped),
                (AttachHiCacheStorageReqInput, self.attach_hicache_storage_wrapped),
                (DetachHiCacheStorageReqInput, self.detach_hicache_storage_wrapped),
                (AbortReq, self.abort_request),
                (OpenSessionReqInput, self.open_session),
                (CloseSessionReqInput, self.close_session),
                (
                    UpdateWeightFromDiskReqInput,
                    self.weight_updater.update_weights_from_disk,
                ),
                (
                    InitWeightsUpdateGroupReqInput,
                    self.weight_updater.init_weights_update_group,
                ),
                (
                    DestroyWeightsUpdateGroupReqInput,
                    self.weight_updater.destroy_weights_update_group,
                ),
                (
                    InitWeightsSendGroupForRemoteInstanceReqInput,
                    self.init_weights_send_group_for_remote_instance,
                ),
                (
                    SendWeightsToRemoteInstanceReqInput,
                    self.send_weights_to_remote_instance,
                ),
                (
                    UpdateWeightsFromDistributedReqInput,
                    self.weight_updater.update_weights_from_distributed,
                ),
                (
                    UpdateWeightsFromTensorReqInput,
                    self.weight_updater.update_weights_from_tensor,
                ),
                (
                    UpdateWeightsFromIPCReqInput,
                    self.weight_updater.update_weights_from_ipc,
                ),
                (
                    UpdateWeightVersionReqInput,
                    self.handle_update_weight_version,
                ),
                (
                    GetWeightsByNameReqInput,
                    self.weight_updater.get_weights_by_name,
                ),
                (
                    ReleaseMemoryOccupationReqInput,
                    self.weight_updater.release_memory_occupation,
                ),
                (
                    ResumeMemoryOccupationReqInput,
                    self.weight_updater.resume_memory_occupation,
                ),
                (
                    CheckWeightsReqInput,
                    self.weight_updater.check_weights,
                ),
                (SlowDownReqInput, self.slow_down),
                (PdRoleSwitchReqInput, self.handle_pd_role_switch),
                (
                    ProfileReq,
                    lambda req: self.profiler_manager._profile(req),
                ),
                (FreezeGCReq, self.handle_freeze_gc),
                (ShutdownReq, self.handle_shutdown),
                (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),
                (
                    LoadLoRAAdapterFromTensorsReqInput,
                    self.load_lora_adapter_from_tensors,
                ),
                (UnloadLoRAAdapterReqInput, self.unload_lora_adapter),
                (PauseGenerationReqInput, self.pause_generation),
                (ContinueGenerationReqInput, self.continue_generation),
                (ConfigureLoggingReq, self.configure_logging),
                (ScaleElasticEPReqInput, self.handle_scale_elastic_ep),
                (DumperControlReqInput, self.handle_dumper_control),
                (AddExternalCorpusReqInput, self.add_external_corpus),
                (
                    RemoveExternalCorpusReqInput,
                    self.remove_external_corpus,
                ),
                (
                    ListExternalCorporaReqInput,
                    self.list_external_corpora,
                ),
            ]
        )

    def get_init_info(self) -> Dict[str, Any]:
        """Return scheduler initialization info for handshake.

        This method provides the initialization info needed by the tokenizer manager
        and other components to verify the scheduler is ready.
        """
        result_dict = {
            "status": "ready",
            "max_total_num_tokens": self.max_total_num_tokens,
            "max_req_input_len": self.max_req_input_len,
            "startup_time": self.startup_time,
        }

        return result_dict

    def release_host_resources(self) -> None:
        # Release pinned host buffers in userspace on graceful shutdown; see
        # HostKVCache.destroy. Called from run_scheduler_process's finally.
        if self.hisparse_coordinator is not None:
            self.hisparse_coordinator.destroy()
        self.tree_cache.release_host_resources()
        if self.decode_offload_manager is not None:
            self.decode_offload_manager.release_host_resources()
        destroy_global_experts_capturer()
        destroy_global_indexer_capturer()

        rank_consensus_checker.shutdown()

    def run_event_loop(self) -> None:
        """Run the scheduler's event loop.

        The event loop blocks until shutdown.
        """
        # Engine init (graph capture, warmups) is done; from here on any
        # Triton kernel device-load is a lazy first-use at serving time.
        triton_load_watch.install()
        triton_load_watch.mark_serving_started()

        if use_mlx():
            # MLX overlap uses mx.async_eval for CPU/GPU overlap,
            # not PyTorch MPS streams.
            self.metrics_reporter.start_scheduler_time_accounting()
            dispatch_event_loop(self)
            return

        if self.device == "cpu":
            self.schedule_stream.synchronize = lambda: None  # No-op for CPU
        elif (is_cuda() or _is_hip) and (
            self.enable_overlap or get_parallel().pp_size > 1
        ):
            # CUDA/HIP streams come from a fixed round-robin pool. Redraw if this
            # stream aliases forward_stream, which would eliminate scheduler
            # overlap. Only CUDA/HIP streams expose a ``cuda_stream`` handle;
            # other accelerators (e.g. NPU/XPU) skip the alias check.
            if self.schedule_stream.cuda_stream == self.forward_stream.cuda_stream:
                self.schedule_stream = allocate_distinct_stream(
                    self.device_module, (self.forward_stream,)
                )
        # The global WAR barrier fences the scheduler's next shared-buffer write
        # on the previous forward's read of the unified memory pool.
        self._war_barrier_enabled = is_cuda() or envs.SGLANG_ENABLE_WAR_BARRIER.get()
        with self.device_module.StreamContext(self.schedule_stream):
            self.metrics_reporter.start_scheduler_time_accounting()
            dispatch_event_loop(self)

    def _apply_war_barrier(self):
        # WAR: keep later schedule_stream writes behind this forward's shared reads.
        # Clearing matters: a phase that skips the publish then falls back to coarse.
        if not self._war_barrier_enabled:
            return
        runner = self.model_worker.last_shared_read_runner
        ev = runner.shared_read_done_event
        runner.shared_read_done_event = None
        if ev is not None and not envs.SGLANG_FORCE_COARSE_WAR_BARRIER.get():
            self.schedule_stream.wait_event(ev)
        else:
            self.schedule_stream.wait_stream(self.forward_stream)

    @DynamicGradMode()
    def event_loop_normal(self):
        """A normal scheduler loop."""
        while True:
            if self.gracefully_exit:
                break

            # Receive requests
            self.ingest_requests()
            if self._engine_paused:
                self._record_scheduler_state_for_paused_engine()
                continue

            # Get the next batch to run
            plan = self.get_next_batch_to_run(
                running_batch=self.running_batch, last_batch=self.last_batch
            )
            self.running_batch = plan.running_batch
            batch = plan.batch_to_run
            self.cur_batch_for_debug = batch

            # Launch the current 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._sched_idled = True
                self.on_idle()

            # Update last_batch
            self.last_batch = batch
            if envs.SGLANG_ENABLE_STRICT_MEM_CHECK_DURING_BUSY.get():
                self.invariant_checker.self_check_during_busy()

    @DynamicGradMode()
    def event_loop_overlap(self):
        """A scheduler loop that overlaps the CPU processing and GPU computation."""
        self.result_queue: Deque[
            Tuple[ScheduleBatch, Union[GenerationBatchResult, EmbeddingBatchResult]]
        ] = deque()

        def pop_and_process():
            # Process the results of the last batch
            tmp_batch, tmp_result = self.result_queue.popleft()
            self.process_batch_result(tmp_batch, tmp_result)

        while True:
            if self.gracefully_exit:
                break

            # Receive requests
            self.ingest_requests()
            if self._engine_paused:
                self._record_scheduler_state_for_paused_engine()
                continue

            # Get the next batch to run
            plan = self.get_next_batch_to_run(
                running_batch=self.running_batch, last_batch=self.last_batch
            )
            self.running_batch = plan.running_batch
            batch = plan.batch_to_run
            self.cur_batch_for_debug = batch
            disable_overlap_for_batch = self.is_disable_overlap_for_batch(
                batch, last_batch=self.last_batch
            )

            # If we do not need to overlap the current batch with the last batch,
            # we can process the last batch immediately.
            if disable_overlap_for_batch:
                pop_and_process()
                # Opportunistic flush at the disable_overlap sync boundary:
                # forward_stream is idle (prev forward drained, next not launched),
                # so `_flush`'s non-urgent guard compacts freely. Sync-free, best-effort.
                if self.enable_unified_memory:
                    try:
                        self.token_to_kv_pool_allocator.flush_opportunistic()
                    except Exception:
                        pass

            # Launch the current batch
            if batch:
                batch_result = self.run_batch(batch)
                # Fence result processing behind this forward's shared reads.
                self._apply_war_barrier()
                self.result_queue.append((batch.copy(), batch_result))
            else:
                batch_result = None
                self._sched_idled = True

            # Process the last batch
            if self.last_batch:
                if not disable_overlap_for_batch:
                    pop_and_process()
            elif batch is None:
                # When the server is idle, do self-check and re-init some states
                self.on_idle()

            # Run sample of the current batch
            # It depends on the result of the last batch (e.g., grammar), so we run it after the last batch is processed.
            if self.is_generation:
                self.launch_batch_sample_if_needed(batch_result, batch)

            # Update last_batch
            self.last_batch = batch

            if envs.SGLANG_ENABLE_STRICT_MEM_CHECK_DURING_BUSY.get():
                self.invariant_checker.self_check_during_busy()

    def is_disable_overlap_for_batch(
        self, batch: ScheduleBatch, last_batch: Optional[ScheduleBatch]
    ) -> bool:
        # For two consecutive prefill batches, we disable overlap to improve the TTFT of the first batch.
        # This might slightly hurt the throughput, so we use an environment variable to control it.
        # In DP attention mode, use the globally synchronized is_extend_in_batch
        # so all DP ranks make the same overlap decision (avoiding deadlock).
        # In non-DP mode, use the local forward_mode directly.
        if self.require_mlp_sync:
            is_extend = lambda b: b and b.is_extend_in_batch
        else:
            is_extend = lambda b: b and b.forward_mode.is_extend()

        batch_is_extend = is_extend(batch)
        last_batch_is_extend = is_extend(last_batch)

        disable_overlap_for_batch = (
            envs.SGLANG_DISABLE_CONSECUTIVE_PREFILL_OVERLAP.get()
            and batch_is_extend
            and last_batch_is_extend
        )

        # Sync so the FSM advance lands before the next batch's bitmask. Permanent
        # path for host-draft algorithms, not a pending migration.
        need_grammar_sync = (
            batch
            and not batch.spec_algorithm.is_none()
            and batch.grammar_needs_sync()
            and batch.forward_mode.is_decode()
            and len(self.result_queue) > 0
        )

        # Algorithms that support grammar overlap advance the FSM inside verify()
        # via the grammar barrier (overlapping the target forward), which resolves
        # whatever result is still pending in the queue — including the
        # extend->decode boundary — so no grammar-specific overlap disable is needed.
        return disable_overlap_for_batch or need_grammar_sync

    def _advance_pending_grammar(self):
        """Grammar barrier (spec-v2 overlap): advance the FSM over any not-yet
        -processed decode result still in the queue, so a following verify()'s
        bitmask sees the previous batch's committed tokens. Invoked mid-worker
        (before generate_token_bitmask) so the CPU advance overlaps the target
        verify forward. Idempotent; no-op when the queue is empty or has no grammar.
        """
        for prev_batch, prev_result in self.result_queue:
            self.batch_result_processor.advance_grammar_fsm(prev_result, prev_batch)

    def ingest_requests(self) -> List:
        """Receive, broadcast and dispatch this iteration's external input.

        The one place a new per-iteration input source belongs; the return
        value exists for the pipeline stages that relay requests onward.
        """
        local_reqs = []
        if (
            get_parallel().pp_rank == 0
            and get_parallel().attn_tp_rank == 0
            and get_parallel().attn_cp_rank == 0
        ):
            local_reqs = self._poll_timeout_aborts()
        recv_reqs = self.request_receiver.recv_requests(local_reqs=local_reqs)
        if recv_reqs:
            self.metrics_reporter.record_scheduler_active()
        self.process_input_requests(recv_reqs)
        return recv_reqs

    @scheduler_stage_method(SCHEDULER_STAGE_PROCESS_REQUESTS)
    def process_input_requests(self, recv_reqs: List):
        now = time.monotonic()
        self.session_controller.maybe_reap(now)

        for recv_req in recv_reqs:
            vmm_errors = None
            if get_mm().mm_feature_transport == "cuda_vmm":
                vmm_errors = self._materialize_cuda_vmm_inputs(recv_req)

            # Skip health check when server is busy — ongoing requests already carry health info.
            if is_health_check_generate_req(recv_req) and not self.is_fully_idle(
                for_health_check=True
            ):
                self.return_health_check_ipcs.append(
                    getattr(recv_req, "http_worker_ipc", None)
                )
                continue

            if vmm_errors is not None and any(vmm_errors):
                self._dispatch_tokenized_mm_requests(recv_req, vmm_errors)
                continue

            output = self._request_dispatcher(recv_req)
            if output is not None:
                if self.rust_server is not None:
                    # Embedded Rust server: every control-request response goes
                    # back through the egress ring (the zmq tokenizer socket is
                    # not consumed); the Rust api_server shapes it per-endpoint.
                    self.rust_server.push_control_output(recv_req, output)
                elif isinstance(output, RpcReqOutput):
                    if self.ipc_channels.recv_from_rpc is not None:
                        sock_send(self.ipc_channels.recv_from_rpc, output)
                else:
                    self.ipc_channels.send_to_tokenizer.send_output(output, recv_req)

        self.flush_wrapper.check_pending()
        if self.external_corpus_manager is not None:
            self.external_corpus_manager.check_pending_load()

        # A runtime PD role switch rebuilt the disaggregation structures for a new
        # role. The response has already been sent above; now break out of the
        # current (old-role) event loop so the supervisor can re-dispatch.
        if get_disagg().enable_pd_role_switch and self._event_loop_should_restart:
            self._event_loop_should_restart = False
            raise role_switch.PdRoleSwitchRestart()

    @staticmethod
    def _tokenized_requests(recv_req):
        if isinstance(
            recv_req, (TokenizedGenerateReqInput, TokenizedEmbeddingReqInput)
        ):
            return (recv_req,)
        if isinstance(
            recv_req,
            (BatchTokenizedGenerateReqInput, BatchTokenizedEmbeddingReqInput),
        ):
            return tuple(recv_req)
        return ()

    def _gather_vmm_materialization_errors(
        self, local_error: Optional[str]
    ) -> List[Optional[str]]:
        if not (
            torch.distributed.is_available()
            and torch.distributed.is_initialized()
            and self.dp_tp_cpu_group is not None
        ):
            return [local_error]

        world_size = torch.distributed.get_world_size(group=self.dp_tp_cpu_group)
        errors = [None] * world_size
        torch.distributed.all_gather_object(
            errors,
            local_error,
            group=self.dp_tp_cpu_group,
        )
        return errors

    def _materialize_cuda_vmm_inputs(self, recv_req) -> Optional[List[Optional[str]]]:
        """Materialize each request and agree on failures across TP ranks."""
        tokenized_reqs = self._tokenized_requests(recv_req)
        if not tokenized_reqs:
            return None

        request_errors = []
        for tokenized_req in tokenized_reqs:
            # The request broadcast makes this skip consistent across ranks.
            if tokenized_req.mm_inputs is None:
                request_errors.append(None)
                continue

            local_error = None
            try:
                if not isinstance(tokenized_req.mm_inputs, MultimodalInputs):
                    tokenized_req.mm_inputs = MultimodalInputs.from_processor_output(
                        tokenized_req.mm_inputs,
                        requires_mm_token_modalities=self.model_config.requires_mm_token_modalities,
                    )
            except Exception as error:
                local_error = f"{type(error).__name__}: {error}"

            rank_errors = self._gather_vmm_materialization_errors(local_error)
            failed_ranks = [
                rank for rank, error in enumerate(rank_errors) if error is not None
            ]
            if failed_ranks:
                details = "; ".join(
                    f"rank {rank}: {rank_errors[rank]}" for rank in failed_ranks
                )
                error_msg = f"Multimodal feature reconstruction failed ({details})"
                logger.error(error_msg)
                tokenized_req.mm_inputs = None
                request_errors.append(error_msg)
            else:
                request_errors.append(None)
        return request_errors

    def _dispatch_tokenized_mm_requests(
        self, recv_req, errors: List[Optional[str]]
    ) -> None:
        tokenized_reqs = self._tokenized_requests(recv_req)
        if len(tokenized_reqs) != len(errors):
            raise RuntimeError("VMM materialization results do not match requests")
        for tokenized_req, error in zip(tokenized_reqs, errors, strict=True):
            if isinstance(tokenized_req, TokenizedGenerateReqInput):
                self.handle_generate_request(tokenized_req, mm_input_error=error)
            elif isinstance(tokenized_req, TokenizedEmbeddingReqInput):
                self.handle_embedding_request(tokenized_req, mm_input_error=error)
            else:
                raise TypeError(
                    f"Unsupported tokenized request type: {type(tokenized_req).__name__}"
                )

    def init_profiler(self) -> None:
        self.profiler_manager = SchedulerProfilerManager(
            ps=self.ps,
            dp_tp_cpu_group=self.dp_tp_cpu_group,
            get_forward_ct=lambda: self.forward_ct,
        )

    def init_weight_updater(self) -> None:
        self.weight_updater = SchedulerWeightUpdaterManager(
            tp_worker=self.tp_worker,
            draft_worker=self.draft_worker,
            tp_cpu_group=self.tp_cpu_group,
            memory_saver_adapter=self.memory_saver_adapter,
            flush_cache=self.flush_cache,
            is_fully_idle=self.is_fully_idle,
            scheduler=self,
            metrics_collector=self.metrics_collector,
        )

    def init_lora_drainer(self) -> None:
        if get_lora().lora_drain_wait_threshold > 0.0:
            self.lora_drainer = LoRADrainer(
                get_lora().max_loras_per_batch,
                get_lora().lora_drain_wait_threshold,
            )
        else:
            self.lora_drainer = None

    def init_lora_overlap_loader(self) -> None:
        if self.enable_lora_overlap_loading:
            self.lora_overlap_loader = LoRAOverlapLoader(
                self.tp_worker.model_runner.lora_manager
            )

    def init_grammar_manager(self) -> None:
        self.grammar_manager = GrammarManager(self)

    def maybe_init_scripted_scheduler_hook(self) -> None:
        if envs.SGLANG_TEST_SCRIPTED_RUNTIME.get():
            from sglang.test.scripted_runtime.scheduler_hook import (
                ScriptedSchedulerHook,
            )

            self.scripted_scheduler_hook = ScriptedSchedulerHook(
                scheduler=self,
                tokenizer_recv_proxy=self.ipc_channels.recv_from_tokenizer,
            )
        else:
            self.scripted_scheduler_hook = None

    def _hosts_rust_server(self) -> bool:
        """Whether this scheduler rank embeds the Rust server (rank 0 only) —
        and with it the server-process duties a Python ``TokenizerManager``
        would otherwise own (e.g. serving the PD KV bootstrap registry)."""
        return envs.SGLANG_RUST_SERVER.get() and (
            get_parallel().pp_rank == 0
            and get_parallel().attn_tp_rank == 0
            and get_parallel().attn_cp_rank == 0
        )

    def maybe_init_rust_server(self) -> None:
        """Start the embedded Rust server (rank 0) if ``SGLANG_RUST_SERVER`` is
        set, and point the ingress receiver at it. All the plumbing lives in
        ``RustServer`` (scheduler_components/rust_scheduler.py)."""

        if not self._hosts_rust_server():
            # Always define the attribute: init_output_streamer and the
            # process_input_requests hook read self.rust_server unconditionally.
            self.rust_server = None
            return

        rust_server = self.get_rust_server_class().launch(self)
        self.rust_server = rust_server
        # The rust server *is* the ingress source: SchedulerRequestReceiver
        # drains its request ring (rust_server_mode) instead of a zmq socket.
        self.recv_from_tokenizer = rust_server
        # Park the idle loop on the request ring within the rank-0 rust-server
        self.idle_sleeper = RustServerIdleSleeper(rust_server)

    def get_rust_server_class(self) -> type[RustServer]:
        return RustServer

    def rust_server_tokenizer_path(self) -> str:
        return get_serving().tokenizer_path

    def init_request_receiver(self) -> None:
        self.request_receiver = SchedulerRequestReceiver(
            recv_from_tokenizer=self.recv_from_tokenizer,
            recv_from_rpc=self.ipc_channels.recv_from_rpc,
            recv_skipper=self.recv_skipper,
            input_blocker=self.input_blocker,
            mm_receiver=self.mm_receiver,
            tp_group=self.tp_group,
            tp_cpu_group=self.tp_cpu_group,
            attn_tp_group=self.attn_tp_group,
            attn_tp_cpu_group=self.attn_tp_cpu_group,
            attn_cp_group=self.attn_cp_group,
            attn_cp_cpu_group=self.attn_cp_cpu_group,
            world_group=self.world_group,
            server_args=self.server_args,
            model_config=self.model_config,
            max_recv_per_poll=self.max_recv_per_poll,
            stream_output=lambda *a, **kw: self.output_streamer.stream_output(*a, **kw),
            get_last_batch=lambda: self.last_batch,
            scripted_scheduler_hook=self.scripted_scheduler_hook,
            scheduler_stage_metrics=self.scheduler_stage_metrics,
        )

    def init_dp_attn_adapter(self) -> None:
        # Spec workers have no .model_runner of their own; the prefill graph
        # runner that votes belongs to the target model.
        target_worker = (
            self.tp_worker.target_worker
            if isinstance(self.tp_worker, BaseSpecWorker)
            else self.tp_worker
        )
        self.dp_attn_adapter = SchedulerDPAttnAdapter(
            model_runner=target_worker.model_runner,
            tp_group=self.tp_group,
            req_to_token_pool=self.req_to_token_pool,
            token_to_kv_pool_allocator=self.token_to_kv_pool_allocator,
            tree_cache=self.tree_cache,
            offload_tags=self.weight_updater.offload_tags,
            ps=self.ps,
            model_config=self.model_config,
            enable_overlap=self.enable_overlap,
            spec_algorithm=self.spec_algorithm,
            get_require_mlp_sync=lambda: self.require_mlp_sync,
        )

    def init_pool_stats_observer(self) -> None:
        self.pool_stats_observer = SchedulerPoolStatsObserver(
            tree_cache=self.tree_cache,
            token_to_kv_pool_allocator=self.token_to_kv_pool_allocator,
            req_to_token_pool=self.req_to_token_pool,
            session_controller=self.session_controller,
            hisparse_coordinator=self.hisparse_coordinator,
            is_hybrid_swa=self.is_hybrid_swa,
            is_hybrid_ssm=self.is_hybrid_ssm,
            enable_hisparse=self.enable_hisparse,
            full_tokens_per_layer=self.full_tokens_per_layer,
            swa_tokens_per_layer=self.swa_tokens_per_layer,
            max_total_num_tokens=self.max_total_num_tokens
            * get_parallel().attn_dcp_size,
            get_last_batch=lambda: self.last_batch,
            get_running_batch=lambda: self.running_batch,
        )

    def init_invariant_checker(self) -> None:
        self.invariant_checker = SchedulerInvariantChecker(
            is_hybrid_swa=self.is_hybrid_swa,
            is_hybrid_ssm=self.is_hybrid_ssm,
            disaggregation_mode=self.disaggregation_mode,
            page_size=self.page_size,
            full_tokens_per_layer=self.full_tokens_per_layer,
            swa_tokens_per_layer=self.swa_tokens_per_layer,
            max_total_num_tokens=self.max_total_num_tokens,
            tree_cache=self.tree_cache,
            token_to_kv_pool_allocator=self.token_to_kv_pool_allocator,
            req_to_token_pool=self.req_to_token_pool,
            pool_stats_observer=self.pool_stats_observer,
            get_last_batch=lambda: self.last_batch,
            get_running_batch=lambda: self.running_batch,
            get_chunked_req=lambda: self.chunked_req,
            scheduler_stage_metrics=self.scheduler_stage_metrics,
        )

    def init_rank_consensus_checker(self) -> None:
        groups = []
        if self.attn_cp_group is not None and self.attn_tp_group is not None:
            groups += [self.attn_cp_group, self.attn_tp_group]
        else:
            groups += [self.tp_group]
        if self.pp_group is not None:
            groups += [self.pp_group]
        rank_consensus_checker.configure(groups)

    def init_kv_events_publisher(self) -> None:
        self.kv_events_publisher = SchedulerKvEventsPublisher(
            kv_events_config=get_observability().kv_events_config,
            attn_tp_rank=get_parallel().attn_tp_rank,
            attn_cp_rank=get_parallel().attn_cp_rank,
            attn_dp_rank=get_parallel().attn_dp_rank,
            dp_rank=get_parallel().dp_rank,
            tree_cache=self.tree_cache,
            send_metrics_from_scheduler=self.ipc_channels.send_metrics_from_scheduler,
            max_running_requests=self.max_running_requests,
            max_total_num_tokens=self.max_total_num_tokens,
            get_stats=lambda: self.metrics_reporter.stats,
        )

    def init_load_publisher(self) -> None:
        # Router-facing load reporting; rank gating and no-op fallback live
        # inside the component. Same interval as the DP-balancing writer so
        # the two fire in phase and the load sink always reuses that snapshot
        # instead of walking the queues itself.
        self.load_publisher = SchedulerLoadPublisher(
            kv_events_config=get_observability().kv_events_config,
            load_publish_endpoint=get_observability().load_publish_endpoint,
            publish_interval=get_observability().load_snapshot_publish_interval,
        )

    def init_load_inquirer(self) -> None:
        self.total_prefill_uncached_tokens = 0
        self.total_prefill_busy_us = 0
        self.decode_moment_totals: list[float] = [0.0] * 6
        self._prev_step: Optional[Tuple[int, float, bool]] = None
        self._prev_prefill_end_ts: Optional[float] = None
        self._sched_idled = False
        self.load_inquirer = SchedulerLoadInquirer(
            disaggregation_mode=self.disaggregation_mode,
            ps=self.ps,
            server_args=self.server_args,
            max_total_num_tokens=self.max_total_num_tokens,
            max_running_requests=self.max_running_requests,
            pool_stats_observer=self.pool_stats_observer,
            tp_worker=self.tp_worker,
            token_to_kv_pool_allocator=self.token_to_kv_pool_allocator,
            spec_algorithm=self.spec_algorithm,
            get_running_batch=lambda: self.running_batch,
            get_waiting_queue=lambda: self.waiting_queue,
            waiting_queue_prefix_matched=lambda: (
                self.policy.waiting_queue_prefix_matched(self.waiting_queue)
            ),
            get_recent_cache_hit_rate=lambda: (
                self.metrics_reporter.recent_cache_hit_rate
            ),
            get_stats=lambda: self.metrics_reporter.stats,
            get_chunked_req=lambda: self.chunked_req,
            get_disagg_prefill_bootstrap_queue=lambda: (
                self.disagg_prefill_bootstrap_queue
            ),
            get_disagg_prefill_inflight_queue=lambda: (
                self.disagg_prefill_inflight_queue
            ),
            get_disagg_decode_prealloc_queue=lambda: self.disagg_decode_prealloc_queue,
            get_disagg_decode_transfer_queue=lambda: self.disagg_decode_transfer_queue,
            get_spec_total_num_accept_tokens=lambda: (
                self.metrics_reporter.spec_total_num_accept_tokens
            ),
            get_spec_total_num_forward_ct=lambda: (
                self.metrics_reporter.spec_total_num_forward_ct
            ),
            get_total_prefill_uncached_tokens=lambda: (
                self.total_prefill_uncached_tokens
            ),
            get_total_prefill_busy_us=lambda: self.total_prefill_busy_us,
            get_decode_moment_totals=lambda: self.decode_moment_totals,
        )

    def init_output_streamer(self) -> None:
        self.output_streamer = self.get_output_streamer_class()(
            send_to_detokenizer=self.ipc_channels.send_to_detokenizer,
            tree_cache=self.tree_cache,
            ps=self.ps,
            server_args=self.server_args,
            is_generation=self.is_generation,
            spec_algorithm=self.spec_algorithm,
            disaggregation_mode=self.disaggregation_mode,
            enable_hicache_storage=lambda: self.enable_hicache_storage,
            rust_server=self.rust_server,
        )

    def get_output_streamer_class(self) -> type[SchedulerOutputStreamer]:
        return SchedulerOutputStreamer

    def init_beam_coordinator(self) -> None:
        self.beam_coordinator = BeamCoordinator(
            model_config=self.model_config,
            spec_algorithm=self.spec_algorithm,
            dllm_enabled=self.dllm_config is not None,
            max_req_len=self.max_req_len,
            req_to_token_pool=self.req_to_token_pool,
            token_to_kv_pool_allocator=self.token_to_kv_pool_allocator,
            tree_cache=self.tree_cache,
            future_map=self.future_map,
        )

    def init_batch_result_processor(self) -> None:
        self.init_beam_coordinator()
        self.batch_result_processor = SchedulerBatchResultProcessor(
            is_generation=self.is_generation,
            disaggregation_mode=self.disaggregation_mode,
            enable_overlap=self.enable_overlap,
            enable_overlap_mlx=self.enable_overlap_mlx,
            model_config=self.model_config,
            token_to_kv_pool_allocator=self.token_to_kv_pool_allocator,
            tree_cache=self.tree_cache,
            hisparse_coordinator=self.hisparse_coordinator,
            req_to_token_pool=self.req_to_token_pool,
            decode_offload_manager=self.decode_offload_manager,
            metrics_collector=self.metrics_collector,
            metrics_reporter=self.metrics_reporter,
            draft_worker=self.draft_worker,
            model_worker=self.model_worker,
            logprob_result_processor=SchedulerLogprobResultProcessor(
                model_config=self.model_config
            ),
            output_streamer=self.output_streamer,
            beam_coordinator=self.beam_coordinator,
            abort_request=self.abort_request,
        )

    def init_req_max_new_tokens(self, req):
        input_len = len(req.origin_input_ids)
        max_new_tokens = (
            req.sampling_params.max_new_tokens
            if req.sampling_params.max_new_tokens is not None
            else 1 << 30
        )
        if self.max_new_tokens_limit is not None and self.max_new_tokens_limit > 0:
            if max_new_tokens > self.max_new_tokens_limit:
                logger.warning(
                    f"Capping max_new_tokens of request {req.rid} to "
                    f"SGLANG_MAX_NEW_TOKENS_LIMIT={self.max_new_tokens_limit} "
                    f"(requested: {req.sampling_params.max_new_tokens})."
                )
            max_new_tokens = min(max_new_tokens, self.max_new_tokens_limit)

        # Keep this bound consistent with PrefillAdder's admission budget.
        max_new_tokens = max(
            0,
            min(
                max_new_tokens,
                self.max_req_len - input_len - 1,
            ),
        )
        max_new_tokens = self.token_to_kv_pool_allocator.max_new_tokens_for_memory(
            input_len,
            max_new_tokens,
            token_capacity=self.max_total_num_tokens * get_parallel().attn_dcp_size,
            sliding_window_size=self.sliding_window_size,
            chunk_size=self.chunked_prefill_size,
        )
        if max_new_tokens is None:
            req.set_finish_with_abort(
                f"Request prompt exceeds the KV memory budget: input_len={input_len}."
            )
            max_new_tokens = 0

        req.sampling_params.max_new_tokens = max(0, max_new_tokens)
        # Clipping above can push max_new_tokens below min_new_tokens, which
        # would suppress EOS for the whole generation. Restore the invariant.
        if req.sampling_params.min_new_tokens > req.sampling_params.max_new_tokens:
            req.sampling_params.min_new_tokens = req.sampling_params.max_new_tokens

    def _process_and_broadcast_mm_inputs(
        self,
        raw_mm_inputs,
    ):
        """Materialize MultimodalInputs once on the entry rank and broadcast to others.

        Entry rank:
        - constructs MultimodalInputs.from_processor_output() 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_processor_output

        Returns:
            MultimodalInputs | None

        Raises:
            _MultimodalInputProcessingError: The entry rank could not build the
                request's multimodal inputs. The same error is broadcast to all
                ranks before it is raised.
        """
        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.dp_tp_cpu_group is not None
            ):
                group_world_size = torch.distributed.get_world_size(
                    group=self.dp_tp_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.
        result = None
        if self.dp_tp_group.rank_in_group == 0:
            try:
                result = _MultimodalInputBroadcast(
                    inputs=MultimodalInputs.from_processor_output(
                        raw_mm_inputs,
                        requires_mm_token_modalities=self.model_config.requires_mm_token_modalities,
                    )
                )
            except Exception as error:
                result = _MultimodalInputBroadcast(
                    error=(
                        "Multimodal input processing failed on the TP entry rank: "
                        f"{type(error).__name__}: {error}"
                    )
                )

            # Broadcast either the prepared inputs or the request-local error.
            if group_world_size > 1:
                obj_list = [result]
                torch.distributed.broadcast_object_list(
                    obj_list,
                    src=self.dp_tp_group.first_rank,
                    group=self.dp_tp_cpu_group,
                )
                result = 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=self.dp_tp_group.first_rank,
                    group=self.dp_tp_cpu_group,
                )
                result = obj_list[0]
            else:
                result = _MultimodalInputBroadcast(
                    inputs=MultimodalInputs.from_processor_output(
                        raw_mm_inputs,
                        requires_mm_token_modalities=self.model_config.requires_mm_token_modalities,
                    )
                )

        if result.error is not None:
            raise _MultimodalInputProcessingError(result.error)
        return result.inputs

    def _get_multimodal_inputs(self, mm_inputs):
        if isinstance(mm_inputs, MMInputsProcessError):
            raise _MultimodalInputProcessingError(mm_inputs.message)
        if isinstance(mm_inputs, MultimodalInputs):
            return mm_inputs

        if get_mm().enable_broadcast_mm_inputs_process:
            return self._process_and_broadcast_mm_inputs(mm_inputs)
        return MultimodalInputs.from_processor_output(
            mm_inputs,
            requires_mm_token_modalities=self.model_config.requires_mm_token_modalities,
        )

    @staticmethod
    def _try_apply_padded_mm_input_ids(recv_req, req, image_inputs) -> bool:
        """setup origin_input_ids with trying to reuse existing MultimodalInputs.padded_input_ids first,
        if absent, call pad_input_ids_func"""
        padded_input_ids = image_inputs.padded_input_ids
        if padded_input_ids is None or recv_req.input_ids is None:
            return False

        recv_input_len = len(recv_req.input_ids)
        if len(padded_input_ids) != recv_input_len:
            return False

        prefix_len = len(req.origin_input_ids) - recv_input_len
        if prefix_len < 0:
            return False

        padded_input_ids = array("q", padded_input_ids)
        if prefix_len == 0:
            req.origin_input_ids = padded_input_ids
        else:
            req.origin_input_ids = req.origin_input_ids[:prefix_len] + padded_input_ids
        return True

    def _maybe_compute_mrope_positions(self, req) -> None:
        """Compute M-RoPE positions when they are missing (e.g. gRPC preprocessed path)."""
        if self._mm_processor is None:
            return
        mm = req.multimodal_inputs
        if mm is None or mm.mrope_positions is not None:
            return

        mrope_positions, mrope_position_delta = (
            self._mm_processor.compute_mrope_positions(
                req.origin_input_ids, mm.mm_items
            )
        )
        if mrope_positions is not None:
            mm.mrope_positions = mrope_positions
            mm.mrope_position_delta = mrope_position_delta

    def _maybe_namespace_elastic_radix_cache(self, req: Req) -> None:
        if (
            get_exec().moe.elastic_ep_backend is None
            or self.disable_radix_cache
            or not self.tree_cache.is_tree_cache()
        ):
            return

        from sglang.srt.elastic_ep.elastic_ep import ElasticEPStateManager

        inst = ElasticEPStateManager.instance()
        if inst is None:
            return

        namespace = f"elastic_ep_size={ElasticEPStateManager.get_effective_ep_size()}"
        if req.extra_key:
            req.extra_key = f"{req.extra_key}|{namespace}"
        else:
            req.extra_key = namespace

        if self._last_logged_elastic_radix_namespace != namespace:
            self._last_logged_elastic_radix_namespace = namespace
            logger.debug(
                "[Elastic EP][scale] radix cache namespace is now %s",
                namespace,
            )

    def _maybe_clear_mm_inputs(self, batch: ScheduleBatch) -> None:
        for req in batch.reqs:
            if not req.finished() or not (mm_inputs := req.multimodal_inputs):
                continue
            # For session requests, keep mm_inputs for the next request
            if req.session:
                continue
            # For non-session requests, clear features and mm_inputs
            mm_inputs.release_features()
            req.multimodal_inputs = None

    def handle_generate_request(
        self,
        recv_req: TokenizedGenerateReqInput,
        *,
        mm_input_error: Optional[str] = None,
    ):
        # Route: normal request / session request / session-not-found
        session_id = (
            recv_req.session_params.id if recv_req.session_params is not None else None
        )
        if recv_req.bootstrap_port is None:
            recv_req.bootstrap_port = get_disagg().disaggregation_bootstrap_port
        # Radix-native sessions use only the top-level session_id.
        radix_native_session = (
            recv_req.session_id is not None and self.enable_session_radix_cache
        )

        if session_id is None or radix_native_session:
            # Normal non-session request, or a radix-native session request
            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)
                recv_req.input_ids = array("q", [1]) * seq_length

            is_beam = BeamCoordinator.request_beam_width(recv_req) > 1
            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,
                return_sampling_mask=recv_req.return_sampling_mask,
                return_flat_raw_top_logprobs=recv_req.return_flat_raw_top_logprobs,
                stream=recv_req.stream,
                lora_id=recv_req.lora_id,
                session_id=recv_req.session_id,
                input_embeds=recv_req.input_embeds,
                positional_embed_overrides=recv_req.positional_embed_overrides,
                token_type_ids=recv_req.token_type_ids,
                custom_logit_processor=recv_req.custom_logit_processor,
                require_reasoning=recv_req.require_reasoning,
                return_hidden_states=recv_req.return_hidden_states,
                return_routed_experts=recv_req.return_routed_experts,
                routed_experts_start_len=recv_req.routed_experts_start_len,
                return_indexer_topk=recv_req.return_indexer_topk,
                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,
                routed_dp_rank=recv_req.routed_dp_rank,
                disagg_prefill_dp_rank=recv_req.disagg_prefill_dp_rank,
                vocab_size=self.model_config.vocab_size,
                priority=recv_req.priority,
                metrics_collector=(
                    self.metrics_collector
                    if self.metrics_reporter.enable_metrics
                    else None
                ),
                routing_key=recv_req.routing_key,
                extra_key=recv_req.extra_key,
                cache_salt=recv_req.cache_salt,
                http_worker_ipc=recv_req.http_worker_ipc,
                dllm_config=self.dllm_config,
                time_stats=recv_req.time_stats,
                multi_item_delimiter_indices=recv_req.multi_item_delimiter_indices,
            )
            req.tokenizer = self.tokenizer

            if radix_native_session:
                req.session_generation = self.tree_cache.ensure_session_generation(
                    recv_req.session_id
                )

            if is_beam:
                error_msg = self.beam_coordinator.validate_and_init(req, recv_req)
                if error_msg:
                    logger.error(error_msg)
                    prepare_abort(req, error_msg, status_code=HTTPStatus.BAD_REQUEST)
                    self.output_streamer.stream_output([req], req.return_logprob)
                    return

            if self.disaggregation_mode != DisaggregationMode.NULL:
                # Invalid request for disaggregated mode
                if (
                    recv_req.bootstrap_room is None
                    and self.transfer_backend != TransferBackend.FAKE
                ):
                    error_msg = (
                        f"Invalid request: Disaggregated request received without "
                        f"bootstrap room id. {req.rid=}"
                    )
                    logger.error(error_msg)
                    if not envs.SGLANG_RUST_SERVER.get():
                        recv_req.time_stats.trace_ctx.abort(
                            abort_info={"reason": error_msg}
                        )
                    prepare_abort(req, error_msg, status_code=HTTPStatus.BAD_REQUEST)
                    self.output_streamer.stream_output([req], req.return_logprob)
                    return

        elif (
            session_id in self.session_controller
            and not self.session_controller.get(session_id).close_on_finish
        ):
            # Session exists and is not closing: create request from session
            session = self.session_controller.get(session_id)
            req = session.create_req(
                recv_req,
                self.tokenizer,
                self.model_config.vocab_size,
                eos_token_ids=self.model_config.hf_eos_token_id,
                disagg_mode=self.disaggregation_mode,
            )
            if self.enable_session_radix_cache:
                req.session_generation = self.tree_cache.ensure_session_generation(
                    session_id
                )
            # TODO: set trace context
            if self.metrics_reporter.enable_metrics:
                req.time_stats.set_metrics_collector(self.metrics_collector)
            if isinstance(req.finished_reason, FINISH_ABORT):
                self.init_req_max_new_tokens(req)
                self._add_request_to_queue(req)
                return

        else:
            # Session not found, or session is closing
            if session_id in self.session_controller:
                error_msg = (
                    f"Invalid request: close was requested for session {session_id}"
                )
            else:
                error_msg = f"Invalid request: session id {session_id} does not exist"
            req = Req(
                recv_req.rid,
                recv_req.input_text,
                recv_req.input_ids,
                recv_req.sampling_params,
                vocab_size=self.model_config.vocab_size,
                http_worker_ipc=recv_req.http_worker_ipc,
            )
            req.tokenizer = self.tokenizer
            req.set_finish_with_abort(error_msg)
            self.init_req_max_new_tokens(req)
            self._add_request_to_queue(req)
            return

        if recv_req.pp_prefetch_ticketed is True:
            self.tree_cache.bind_prefetch_ticket(req.rid)
        self._maybe_namespace_elastic_radix_cache(req)

        if mm_input_error is not None:
            req.set_finish_with_abort(
                mm_input_error,
                status_code=HTTPStatus.INTERNAL_SERVER_ERROR,
                err_type="InternalServerError",
            )
            self.init_req_max_new_tokens(req)
            self._add_request_to_queue(req)
            return

        if self.spec_algorithm.is_dflash_family():
            error_msg = validate_dflash_request(req, self.enable_overlap)
            if error_msg is not None:
                req.set_finish_with_abort(error_msg)
                self.init_req_max_new_tokens(req)
                self._add_request_to_queue(req)
                return

        if self.spec_algorithm.is_uno():
            error_msg = validate_uno_request(req)
            if error_msg is not None:
                req.set_finish_with_abort(error_msg)
                self.init_req_max_new_tokens(req)
                self._add_request_to_queue(req)
                return

        if req.return_sampling_mask:
            if (
                self.disaggregation_mode != DisaggregationMode.NULL
                and not self.disagg_metadata_buffers.enable_sampling_mask
            ):
                self._reject_sampling_mask_request(
                    req,
                    "return_sampling_mask requires "
                    "SGLANG_ENABLE_DISAGG_SAMPLING_MASK=1 on both prefill and "
                    "decode servers when using disaggregated serving.",
                )
                return
            top_k = req.sampling_params.top_k
            sampling_mask_cap = self.server_args.sampling_mask_max_tokens
            if top_k != 1 and not (1 < top_k <= sampling_mask_cap):
                error_msg = (
                    "return_sampling_mask requires top_k=1 for greedy sampling "
                    f"or finite 1 < top_k <= {sampling_mask_cap}; got top_k="
                    f"{top_k}. Lower top_k or increase "
                    "--sampling-mask-max-tokens."
                )
                self._reject_sampling_mask_request(req, error_msg)
                return

        if req.return_sampling_mask and not self.spec_algorithm.is_none():
            # Spec workers do not emit one sampling support per accepted token, so
            # the returned mask would not align 1:1 with generated tokens. Reject
            # the combination instead of silently returning a misaligned mask.
            error_msg = (
                "return_sampling_mask is not supported with speculative decoding."
            )
            self._reject_sampling_mask_request(req, error_msg)
            return

        if req.return_sampling_mask and get_exec().kernel.sampling_backend == "ascend":
            # The ascend backend samples from logits directly and never builds the
            # top-k/top-p support, so it cannot produce a sampling mask.
            error_msg = (
                "return_sampling_mask is not supported with the ascend "
                "sampling backend."
            )
            self._reject_sampling_mask_request(req, error_msg)
            return

        # Handle multimodal inputs
        if recv_req.mm_inputs is not None:
            try:
                image_inputs = self._get_multimodal_inputs(recv_req.mm_inputs)
            except _MultimodalInputProcessingError as error:
                req.set_finish_with_abort(
                    str(error),
                    status_code=HTTPStatus.INTERNAL_SERVER_ERROR,
                    err_type="InternalServerError",
                )
                self.init_req_max_new_tokens(req)
                self._add_request_to_queue(req)
                return

            SessionController.adjust_mm_offsets(recv_req, req, image_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.
            # The pad function is model-specific and can be None for some backends.
            if (
                not self._try_apply_padded_mm_input_ids(recv_req, req, image_inputs)
                and self.pad_input_ids_func
            ):
                req.origin_input_ids = array(
                    "q", self.pad_input_ids_func(req.origin_input_ids, image_inputs)
                )
            req.extend_image_inputs(image_inputs)
            self._maybe_compute_mrope_positions(req)

            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,
            get_serving().allow_auto_truncate,
        )
        if error_msg:
            req.set_finish_with_abort(error_msg)
            self._add_request_to_queue(req)
            return

        if not recv_req.return_logprob and recv_req.logprob_start_len != -1:
            # When return_logprob is False, logprob_start_len should be ignored
            recv_req.logprob_start_len = -1

        if recv_req.logprob_start_len == -1:
            if recv_req.return_logprob and recv_req.token_ids_logprob is None:
                # If logprob is required but neither token_ids_logprob nor logprob_start_len is
                # set, return the logprobs for output tokens by default
                req.logprob_start_len = len(req.origin_input_ids)
            elif req.is_prefill_only:
                # For prefill-only requests with logprob_start_len == -1, set logprob_start_len
                # beyond input sequence to skip input logprob computation entirely
                req.logprob_start_len = len(req.origin_input_ids)
            else:
                # If return_logprob is False, only the last token requires logprob computation
                req.logprob_start_len = -1
        else:
            req.logprob_start_len = recv_req.logprob_start_len

        if 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 = -1
            req.set_finish_with_abort(error_msg)
            self._add_request_to_queue(req)
            return

        if (
            get_device().mlx_enable_sampling
            and req.return_logprob
            and 0 <= req.logprob_start_len < len(req.origin_input_ids)
        ):
            # The MLX sampling path computes output logprobs only; the
            # prefill result carries no input_token_logprobs, so letting
            # this through would crash output processing.
            error_msg = (
                "Prompt input logprobs (logprob_start_len) are not supported "
                "on the MLX sampling path; omit logprob_start_len to get "
                "output logprobs."
            )
            req.logprob_start_len = -1
            req.set_finish_with_abort(error_msg)
            self._add_request_to_queue(req)
            return

        if recv_req.return_routed_experts:
            error_msg = None
            if recv_req.routed_experts_start_len < 0:
                error_msg = (
                    f"{recv_req.routed_experts_start_len=} is lower than 0. "
                    "Please use a non-negative routed_experts_start_len."
                )

            if recv_req.routed_experts_start_len > len(req.origin_input_ids):
                error_msg = (
                    f"{recv_req.routed_experts_start_len=} is higher than the "
                    f"number of input tokens {len(req.origin_input_ids)=}. Please "
                    f"use a smaller routed_experts_start_len."
                )

            if error_msg is not None:
                req.routed_experts_start_len = 0
                req.set_finish_with_abort(error_msg)
                self._add_request_to_queue(req)
                return

        if self.ps.pp_rank == 0 and getattr(
            self.tree_cache.cache_controller, "pp_prefetch_command_group", None
        ):
            recv_req.pp_prefetch_ticketed = bool(self._prefetch_kvcache(req))
            self.tree_cache.bind_prefetch_ticket(req.rid, recv_req.pp_prefetch_ticketed)

        added_to_grammar_queue = self.grammar_manager.process_req_with_grammar(req)
        if not added_to_grammar_queue:
            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, storage_hit_end: Optional[int] = None):
        if self.enable_hicache_storage:
            req.init_next_round_input(self.tree_cache, cow_mamba=False)
            tree_cache = self.tree_cache
            buffer_mode = get_memory().hicache_host_memory_mode == "buffer_only"
            last_host_node = req.last_host_node
            # Buffer mode host-backups nothing, so match_prefix anchors at
            # root; re-anchor on the deepest device node. The anchor is only
            # read for hash/extra-key context here (never locked), so a device
            # node serves. Cache mode keeps the is_backuped gate below: its
            # write-through prefix is contiguous from root, so an unbacked
            # anchor means a guaranteed storage miss.
            if (
                buffer_mode
                and tree_cache.is_root(last_host_node)
                and not tree_cache.is_root(req.last_node)
            ):
                last_host_node = req.last_node

            matched_len = len(req.prefix_indices) + req.host_hit_length
            req.storage_prefetch_last_match_len = matched_len

            if (
                tree_cache.is_backuped(last_host_node)
                or tree_cache.is_root(last_host_node)
                or (
                    buffer_mode
                    and tree_cache.get_last_hash_value(last_host_node) is not None
                )
            ):
                match_end = req._compute_max_prefix_len(
                    len(req.full_untruncated_fill_ids)
                )
                new_input_tokens = req.full_untruncated_fill_ids[matched_len:match_end]
                prefix_keys = (
                    tree_cache.get_prefix_hash_values(last_host_node)
                    if tree_cache.hicache_storage_pass_prefix_keys
                    else None
                )
                return tree_cache.prefetch_from_storage(
                    req.cache_request_handle,
                    last_host_node,
                    new_input_tokens,
                    tree_cache.get_last_hash_value(last_host_node),
                    prefix_keys,
                    matched_prefix_tokens=req.full_untruncated_fill_ids[:matched_len],
                    extra_key=req.extra_key,
                    cache_salt=req.cache_salt,
                    storage_hit_end=storage_hit_end,
                )

    def _process_storage_prefetch_retries(self):
        """Issue due L3 attempts in the current waiting-queue order."""
        retries = self.tree_cache.storage_prefetch_retries
        if retries is None:
            return
        memory = get_memory()
        for req, storage_hit_end in retries.pop_ready(
            self.waiting_queue,
            memory.hicache_storage_prefetch_retry_poll_interval,
            memory.hicache_storage_prefetch_retry_max_attempts,
        ):
            self._retry_storage_prefetch(req, storage_hit_end)

    def _retry_storage_prefetch(
        self, req: Req, storage_hit_end: Optional[int] = None
    ) -> None:
        req.storage_prefetch_retry_attempts += 1
        max_attempts = get_memory().hicache_storage_prefetch_retry_max_attempts
        if req.storage_prefetch_retry_attempts >= max_attempts:
            logger.warning(
                "HiCache storage prefetch reissue cap reached req=%s attempts=%d; "
                "the request is admitted without further L3 lookups",
                req.rid,
                req.storage_prefetch_retry_attempts,
            )
        else:
            logger.debug(
                "HiCache storage prefetch re-issue req=%s attempt=%d",
                req.rid,
                req.storage_prefetch_retry_attempts,
            )
        self._prefetch_kvcache(req, storage_hit_end)

    def _prefetch_after_device_hit_loss(self, req: Req) -> bool:
        """Re-query an L3 range newly exposed by queue-time device eviction."""
        previous_match_len = req.storage_prefetch_last_match_len
        buffer_pipeline = self.tree_cache.buffer_pipeline
        if not previous_match_len or (
            buffer_pipeline is not None
            and buffer_pipeline.has_staged(req.cache_request_handle)
        ):
            return False
        current_match_len = len(req.prefix_indices) + req.host_hit_length
        if current_match_len >= previous_match_len:
            return False
        if (
            req.storage_prefetch_retry_attempts
            >= get_memory().hicache_storage_prefetch_retry_max_attempts
        ):
            # Past the re-issue cap the shorter live match is admitted as is.
            req.storage_prefetch_last_match_len = current_match_len
            return False
        logger.warning(
            "HiCache device prefix shrank before admission req=%s "
            "lookup_match=%d current_match=%d; reissuing storage lookup",
            req.rid,
            previous_match_len,
            current_match_len,
        )
        self._retry_storage_prefetch(req)
        return True

    def retire_unadmitted_request(self, req: Req) -> None:
        """Finish a request the disaggregation queues rejected at their door."""
        # `create_req` marks a streaming session in-flight, and the pre-abort
        # detach lives in `StreamingSession.find_active_slot`, which only runs
        # while scheduling; a session left in-flight rejects every later request.
        if req.session is not None and req.session.streaming:
            req.session.abort_req()
            req.session = None
        # `beam_coordinator.validate_and_init` counts the group in ahead of the
        # checks that reject; no-op when the request has no group.
        self.beam_coordinator.retire_group(req)
        # PREFILL runs `_prefetch_kvcache` before its door, so even the
        # one-token stub is registered with the cache by now:
        # `prefetch_from_storage` arms the paced-retry set for this attempt's
        # cache handle. Only a `finish`/ABORT and a `waiting_queue` sweep clear
        # that, and a retired request reaches neither.
        self._release_aborted_request(req)
        # `update_finish_state` returns early once `finished()`, so an already
        # set `finished_reason` is what the client receives; report the same.
        reason = req.finished_reason or req.to_finish
        req.time_stats.trace_ctx.abort(abort_info={"reason": reason.message})
        req.update_finish_state()
        self.output_streamer.stream_output([req], req.return_logprob)

    def _add_request_to_queue(self, req: Req, is_retracted: bool = False):
        if not self._set_or_validate_priority(req):
            self._release_aborted_request(req)
            return
        if is_retracted:
            req.storage_prefetch_retry_attempts = 0
            req.storage_prefetch_last_match_len = None
            req.staged_prefetch_plan = None
        if self.disaggregation_mode == DisaggregationMode.NULL:
            if self._abort_on_queued_limit(req):
                self._release_aborted_request(req)
                return
            self._prefetch_kvcache(req)
            self.waiting_queue.append(req)
            req.time_stats.set_wait_queue_entry_time()
            req.arrival_processed_tokens = self.processed_tokens_counter
        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.set_prefill_bootstrap_queue_entry_time()
        elif self.disaggregation_mode == DisaggregationMode.DECODE:
            self.disagg_decode_prealloc_queue.add(req, is_retracted=is_retracted)
            if not is_retracted:
                req.time_stats.set_decode_prealloc_queue_entry_time()
            else:
                req.time_stats.set_retract_time()
        else:
            raise ValueError(f"Invalid {self.disaggregation_mode=}")

    def _reject_sampling_mask_request(self, req: Req, error_msg: str) -> None:
        """Return a sampling-mask validation error without running the model."""
        logger.error(f"{error_msg}, {req.rid=}")
        req.time_stats.trace_ctx.abort(abort_info={"reason": error_msg})
        prepare_abort(req, error_msg, status_code=HTTPStatus.BAD_REQUEST)
        self.output_streamer.stream_output([req], req.return_logprob)

    def _set_or_validate_priority(self, req: Req) -> bool:
        """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 = _make_abort_req(
                req,
                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.",
                },
            )
            req.time_stats.trace_ctx.abort(abort_info=abort_req.finished_reason)
            self.ipc_channels.send_to_tokenizer.send_output(abort_req, req)
            return False
        return True

    def _release_aborted_request(self, req: Req) -> None:
        """Drop the cache-side state an aborted request left behind."""
        self.tree_cache.finish(req.cache_request_handle, CacheRequestOutcome.ABORT)

    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._release_aborted_request(candidate_req)
                self.waiting_queue.pop(idx)
                self.beam_coordinator.retire_group(candidate_req)
                req_to_abort = candidate_req
                message = "The request is aborted by a higher priority request."

        self.ipc_channels.send_to_tokenizer.send_output(
            _make_abort_req(
                req_to_abort,
                finished_reason={
                    "type": "abort",
                    "status_code": HTTPStatus.SERVICE_UNAVAILABLE,
                    "message": message,
                },
            ),
            req_to_abort,
        )
        req_to_abort.time_stats.trace_ctx.abort(abort_info={"reason": message})
        return req_to_abort.rid == recv_req.rid

    def _poll_timeout_aborts(self) -> List[AbortReq]:
        """Emit aborts only; every rank must drop the same requests in the
        same iteration, or the extend-vs-decode decision splits and the
        collectives hang.
        """
        aborts: List[AbortReq] = []

        if (timeout_s := envs.SGLANG_REQ_WAITING_TIMEOUT.get()) > 0:
            deadline = time.perf_counter() - timeout_s
            for req in self.waiting_queue:
                entry_time = req.time_stats.wait_queue_entry_time
                if 0 < entry_time < deadline:
                    aborts.append(
                        AbortReq(
                            rid=req.rid,
                            abort_message="Request waiting timeout reached.",
                            finished_reason={
                                "type": "abort",
                                "status_code": HTTPStatus.SERVICE_UNAVAILABLE,
                                "message": "Request waiting timeout reached.",
                            },
                        )
                    )

        if (timeout_s := envs.SGLANG_REQ_RUNNING_TIMEOUT.get()) > 0:
            deadline = time.perf_counter() - timeout_s
            if get_parallel().pp_size == 1:
                inflight_batches = [self.running_batch, self.last_batch]
            else:
                inflight_batches = [*self.running_mbs, *self.mbs]
            seen_rids = set()
            for batch in inflight_batches:
                if batch is None:
                    continue
                for req in batch.reqs:
                    if req.rid in seen_rids or req.finished():
                        continue
                    seen_rids.add(req.rid)
                    if 0 < req.time_stats.forward_entry_time < deadline:
                        aborts.append(
                            AbortReq(
                                rid=req.rid,
                                abort_message="Request running timeout reached.",
                                finished_reason={
                                    "type": "abort",
                                    "status_code": HTTPStatus.SERVICE_UNAVAILABLE,
                                    "message": "Request running timeout reached.",
                                },
                            )
                        )

        return aborts

    def handle_embedding_request(
        self,
        recv_req: TokenizedEmbeddingReqInput,
        *,
        mm_input_error: Optional[str] = None,
    ):
        req = Req(
            recv_req.rid,
            recv_req.input_text,
            recv_req.input_ids,
            recv_req.sampling_params,
            positional_embed_overrides=recv_req.positional_embed_overrides,
            token_type_ids=recv_req.token_type_ids,
            routed_dp_rank=recv_req.routed_dp_rank,
            priority=recv_req.priority,
            dimensions=recv_req.dimensions,
            lora_id=recv_req.lora_id,
            http_worker_ipc=recv_req.http_worker_ipc,
            time_stats=recv_req.time_stats,
            return_pooled_hidden_states=recv_req.return_pooled_hidden_states,
            multi_item_delimiter_indices=recv_req.multi_item_delimiter_indices,
        )
        req.tokenizer = self.tokenizer
        self._maybe_namespace_elastic_radix_cache(req)

        if mm_input_error is not None:
            req.set_finish_with_abort(
                mm_input_error,
                status_code=HTTPStatus.INTERNAL_SERVER_ERROR,
                err_type="InternalServerError",
            )
            self._add_request_to_queue(req)
            return

        # Handle multimodal inputs
        if recv_req.mm_inputs is not None:
            try:
                image_inputs = self._get_multimodal_inputs(recv_req.mm_inputs)
            except _MultimodalInputProcessingError as error:
                req.set_finish_with_abort(
                    str(error),
                    status_code=HTTPStatus.INTERNAL_SERVER_ERROR,
                    err_type="InternalServerError",
                )
                self._add_request_to_queue(req)
                return
            # Expand a single image token into multiple dummy tokens for receiving image embeddings
            # The `pad_input_ids_func` is model-specific and may be None for
            # embedding models or models not requiring special padding.
            # If None, `req.origin_input_ids` is expected to be correctly populated already.
            if (
                not self._try_apply_padded_mm_input_ids(recv_req, req, image_inputs)
                and self.pad_input_ids_func
            ):
                # See companion call site above for the array.array wrap rationale.
                req.origin_input_ids = array(
                    "q", self.pad_input_ids_func(req.origin_input_ids, image_inputs)
                )

            req.extend_image_inputs(image_inputs)
            self._maybe_compute_mrope_positions(req)

            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,
            get_serving().allow_auto_truncate,
        )
        if error_msg:
            self._add_request_to_queue(req)
            return

        # Copy more attributes
        req.logprob_start_len = -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 stash_chunked_request(self, req: Req):
        maybe_cache_unfinished_req(req, self.tree_cache, chunked=True)

    def process_pending_chunked_abort(self) -> None:
        """Abort an in-flight chunked-prefill request once it is safe to do so.

        ``abort_request`` only records the target in ``_pending_chunked_abort_req``
        (tearing it down mid-iteration is unsafe). Clearing ``chunked_req`` here at
        the top of the scheduling step stops the next chunk from launching; the
        chunk already launched is drained when its result is resolved. Under overlap
        the result lands a step later, so the batch-result processors keep
        ``inflight_middle_chunks`` accounting intact and skip the aborted chunk:
        ``process_batch_result_disagg_prefill`` via its ``is_aborted`` drop, and
        ``process_batch_result_prefill`` via its chunked branch (the finished req
        is excluded from streaming and its logprob offset is still accounted).
        Mirrors ``handle_bootstrap_failure``.
        """
        req = self._pending_chunked_abort_req
        if req is None:
            return
        if self.chunked_req is not req:
            # Already past chunked prefill; the running-batch abort path handles
            # it. Drop the marker once the request is actually gone.
            if req.finished() or not req.kv.holds_kv:
                self._pending_chunked_abort_req = None
                return
            # The request moved to another scheduler queue after abort_request
            # deferred it, so retry against its current location.
            self._pending_chunked_abort_req = None
            self.abort_request(AbortReq(rid=req.rid))
            return

        prepare_abort(req, "Aborted")
        req.time_stats.trace_ctx.abort(abort_info={"reason": "Aborted"})
        req.to_finish = None
        if self.disaggregation_mode == DisaggregationMode.PREFILL:
            self.clear_pending_chunk_send(req)
            req.disagg_kv_sender.abort()
            maybe_release_metadata_buffer(
                req, self.req_to_metadata_buffer_idx_allocator
            )
            req.pending_bootstrap = False
        self._release_aborted_request(req)
        release_kv_cache(req, self.tree_cache, is_insert=False)

        self.chunked_req = None
        self._pending_chunked_abort_req = None
        self.ipc_channels.send_to_tokenizer.send_output(_make_abort_req(req), req)
        logger.debug(f"Abort chunked prefill request. {req.rid=}")

    def _build_hisparse_decode_batch(self, reqs):
        """Build a ScheduleBatch for hisparse requests transitioning from staging to decode."""
        device = self.device

        batch = ScheduleBatch.init_new(
            reqs=reqs,
            req_to_token_pool=self.req_to_token_pool,
            token_to_kv_pool_allocator=self.token_to_kv_pool_allocator,
            tree_cache=self.tree_cache,
            model_config=self.model_config,
            enable_overlap=self.enable_overlap,
            spec_algorithm=self.spec_algorithm,
        )

        req_pool_indices = [r.kv.req_pool_idx for r in reqs]
        batch.req_pool_indices = torch.tensor(
            req_pool_indices, dtype=torch.int64, device=device
        )
        batch.req_pool_indices_cpu = torch.tensor(req_pool_indices, dtype=torch.int64)
        seq_lens = [len(r.origin_input_ids) + len(r.output_ids) - 1 for r in reqs]
        batch.seq_lens = torch.tensor(seq_lens, dtype=torch.int64, device=device)
        batch.seq_lens_cpu = torch.tensor(seq_lens, dtype=torch.int64)
        batch.orig_seq_lens = torch.tensor(seq_lens, dtype=torch.int32, device=device)
        batch.seq_lens_sum = sum(seq_lens)
        # Stash last token into relay; resolve_forward_inputs will gather.
        last_tokens = torch.tensor(
            [r.output_ids[-1] for r in reqs], dtype=torch.int64, device=device
        )
        self.future_map.stash(
            batch.req_pool_indices, RelayPayload(bonus_tokens=last_tokens)
        )
        batch.input_ids = None

        if batch.return_logprob:
            batch.top_logprobs_nums = [r.logprob.top_logprobs_num for r in reqs]
            batch.token_ids_logprobs = [list(r.origin_input_ids) for r in reqs]

        batch.sampling_info = SamplingBatchInfo.from_schedule_batch(
            batch, self.model_config.vocab_size
        )
        # todo hisparse, maybe other info to contain for the new batch
        return batch

    def _process_hicache_events(self) -> None:
        # The HiCache drain is TP-wide consensus; run it before rank-local
        # decisions (_should_defer_prefill) or ranks enter different collectives.
        if (
            self.enable_hierarchical_cache
            or get_memory().enable_flexkv
            or self.enable_unified_cache_external_linker
        ):
            self.tree_cache.check_hicache_events()
            if self.enable_hicache_storage:
                self._process_storage_prefetch_retries()

    @scheduler_stage_method(SCHEDULER_STAGE_GET_NEXT_BATCH)
    def get_next_batch_to_run(
        self, running_batch: ScheduleBatch, last_batch: Optional[ScheduleBatch]
    ) -> NextBatchPlan:
        self.process_pending_chunked_abort()
        self._process_hicache_events()

        if self.enable_fpm:
            self._fpm_batch_t0 = time.monotonic()
        if self.dllm_config is not None:
            self.dllm_manager.filter_finished_reqs()

        # Merge the prefill batch into the running batch
        chunked_req_to_exclude = set()

        if self.dllm_config is not None and self.dllm_manager.any_staging_reqs():
            chunked_req_to_exclude.update(self.dllm_manager.staging_queue)
            for req in self.dllm_manager.staging_queue:
                if self.dllm_config.first_done_first_out_mode:
                    if not req.dllm_incomplete_ids:
                        self.stash_chunked_request(req)
                        self.req_to_token_pool.free(req)
                    # Otherwise, keep req slot/KV for reuse.
                else:
                    self.stash_chunked_request(req)

        if self.chunked_req is not None:
            # 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)

            # Stash (cache) the previous chunk only when it produced new KV
            # beyond what is already cached. A parked chunk (add_chunked_req
            # hybrid-SWA early-return) leaves extend_range.end ==
            # len(prefix_indices), so there is nothing new to cache and
            # stashing would be a no-op.
            if self.chunked_req.extend_range.end > len(self.chunked_req.prefix_indices):
                self.stash_chunked_request(self.chunked_req)

        # HiSparse has its own prefill-to-decode transition; skip last_batch merge.
        if self.enable_hisparse:
            ready_reqs = self.hisparse_coordinator.collect_ready_reqs()
            if len(ready_reqs) > 0:
                new_batch = self._build_hisparse_decode_batch(ready_reqs)
                if running_batch.is_empty():
                    running_batch = new_batch
                else:
                    running_batch.merge_batch(new_batch)
                running_batch.hisparse_coordinator = self.hisparse_coordinator
            # Reset batch_is_full so the scheduler can schedule more prefills.
            running_batch.batch_is_full = False

        if (
            not self.enable_hisparse
            and last_batch
            and last_batch.forward_mode.is_extend()
        ):
            if 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(last_batch.chunked_req)

            if self.dllm_config is not None and last_batch.reqs:
                chunked_req_to_exclude.update(last_batch.reqs)

            # Filter batch
            last_bs = last_batch.batch_size()
            last_batch.filter_batch(chunked_req_to_exclude=list(chunked_req_to_exclude))
            if last_batch.batch_size() < last_bs:
                running_batch.batch_is_full = False

            # Merge the new batch into the running batch.
            if not last_batch.is_empty():
                if running_batch.is_empty():
                    running_batch = last_batch
                else:
                    # Merge running_batch with prefill batch
                    running_batch.merge_batch(last_batch)

        # For prefill-only batch, filter out finished requests since they
        # won't go through the decode step. This keeps running_batch accurate
        # for load reporting (num_running_reqs via /v1/loads).
        # Runs outside the last_batch block so stale requests are cleaned
        # even when no new batches arrive (e.g. traffic stops).
        if running_batch.is_prefill_only:
            running_batch.filter_batch()
            if running_batch.is_empty():
                running_batch.batch_is_full = False

        if self.dllm_config is not None:
            new_batch = self.get_new_batch_dllm(running_batch)
        elif self._should_defer_prefill():
            new_batch = None
        else:
            prefill_plan = self.get_new_batch_prefill(running_batch)
            new_batch = prefill_plan.batch_to_run
            running_batch = prefill_plan.running_batch

        need_mlp_sync = self.require_mlp_sync
        if (
            need_mlp_sync
            and not self.spec_algorithm.is_none()
            and not get_spec().speculative_skip_dp_mlp_sync
        ):
            # NOTE: This branch makes sure prefill and decode batches will not be mixed when spec and dp-attn is enabled.
            # Before merging the new batch into running batch:
            # 1. All new batches are none -> need_mlp_sync remains true (sync is needed for decode batch).
            # 2. All new batches are some (prefill / idle) -> we do not need prepare mlp sync one more time.
            new_batch = self.dp_attn_adapter.maybe_prepare_mlp_sync_batch(new_batch)
            need_mlp_sync = new_batch is None

        if new_batch is not None:
            # Run prefill first if possible
            ret = new_batch
        else:
            # Run decode (skip for prefill-only batches)
            if not running_batch.is_empty() and not running_batch.is_prefill_only:
                running_batch = self.update_running_batch(running_batch)
                ret = running_batch if not running_batch.is_empty() else None
            else:
                ret = None

        # Handle DP attention and log stats
        ret = self.dp_attn_adapter.maybe_prepare_mlp_sync_batch(
            ret, need_sync=need_mlp_sync
        )
        # Decode->extend conversion keeps a heterogeneous dp step replayable.
        converted = self.dp_attn_adapter.maybe_convert_decode_to_extend(ret)
        if converted is running_batch and converted.forward_mode.is_extend():
            # The converted batch re-enters via the last_batch extend-merge
            # next iteration; empty running_batch or it merges with itself.
            running_batch = ScheduleBatch(
                reqs=[], batch_is_full=running_batch.batch_is_full
            )
        ret = converted
        self._arm_prefill_decode_interval(ret)

        # Handle ngram embedding
        ret = self.ngram_embedding_manager.prepare_for_forward(
            ret, chunked_req=self.chunked_req
        )

        if ret:
            set_schedule_time_batch(ret)
            if self.enable_fpm:
                ret.fpm_start_time = self._fpm_batch_t0

        return NextBatchPlan(batch_to_run=ret, running_batch=running_batch)

    def get_num_allocatable_reqs(
        self,
        running_bs: int,
        beam_width: Optional[int] = None,
        running_batch: Optional[ScheduleBatch] = None,
    ) -> int:
        pp_budget = get_parallel().pp_max_micro_batch_size - running_bs
        available = self.req_to_token_pool.available_size()

        active_batch = running_batch or self.running_batch
        available = max(
            available - self.beam_coordinator.pending_member_rows(active_batch), 0
        )
        res = min(pp_budget, available)
        if beam_width is not None:
            # A beam candidate owns beam_width rows once decoding.
            res = min(res, available // beam_width)

        return res

    def get_new_batch_prefill(self, running_batch: ScheduleBatch) -> NextBatchPlan:
        prefill_delayer_single_pass = None
        if self.prefill_delayer:
            # Get max usage across all pools for prefill delay decision
            max_pool_usage = (
                self.pool_stats_observer.get_pool_stats().get_max_pool_usage()
            )
            prefill_delayer_single_pass = PrefillDelayerSinglePassExecutor(
                self.prefill_delayer, token_usage=max_pool_usage
            )

        ret, running_batch = self._get_new_batch_prefill_raw(
            prefill_delayer_single_pass=prefill_delayer_single_pass,
            running_batch=running_batch,
        )

        if self.prefill_delayer:
            observed_prefill_bs = prefill_delayer_single_pass.finalize(
                actual_prefill_bs=ret.batch_size() if ret is not None else 0
            )
            if observed_prefill_bs > 0:
                self.max_prefill_bs = self.prefill_bs_tracker.observe_attempt(
                    observed_prefill_bs
                )

        return NextBatchPlan(batch_to_run=ret, running_batch=running_batch)

    def _get_new_batch_prefill_raw(
        self,
        prefill_delayer_single_pass: Optional[PrefillDelayerSinglePassExecutor],
        running_batch: ScheduleBatch,
    ) -> Tuple[Optional[ScheduleBatch], ScheduleBatch]:
        # Check if the grammar is ready in the grammar queue
        if self.grammar_manager.has_waiting_grammars():
            ready_grammar_requests = self.grammar_manager.get_ready_grammar_requests()
            for req in ready_grammar_requests:
                self._add_request_to_queue(req)

        if self.enable_priority_preemption or self.is_hybrid_swa:
            # Reset batch_is_full to try preemption with a prefill adder.
            running_batch.batch_is_full = False

        if (
            running_batch.batch_is_full or len(self.waiting_queue) == 0
        ) and self.chunked_req is None:
            return None, running_batch

        running_bs = len(running_batch.reqs)
        # Skipped during a chunked prefill: that pass must proceed regardless.
        if (
            self.min_free_slots_delayer is not None
            and self.chunked_req is None
            and self.min_free_slots_delayer.should_delay(
                running_bs=running_bs,
                num_allocatable_reqs=self.get_num_allocatable_reqs(
                    running_bs, running_batch=running_batch
                ),
            )
        ):
            return None, running_batch

        # 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 requests has just been released.
        # In PP case, chunked requests (or dllm requests) 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 requests to be added, otherwise, there will be a memory leak.
        if (
            self.get_num_allocatable_reqs(running_bs, running_batch=running_batch) <= 0
            and self.chunked_req is None
            and not self.enable_priority_preemption
        ):
            running_batch.batch_is_full = True
            return None, running_batch

        # Get priority queue
        self.policy.calc_priority(
            self.waiting_queue,
            running_batch,
            processed_tokens=self.processed_tokens_counter,
        )

        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, running_batch

        # Determine chunked_prefill_size for this batch
        chunked_prefill_size = self.chunked_prefill_size
        if self.chunked_req is not None and self.dynamic_chunk_sizer is not None:
            history_len = len(self.chunked_req.prefix_indices)
            dynamic_size = self.dynamic_chunk_sizer.predict(history_len)
            if dynamic_size is not None:
                chunked_prefill_size = dynamic_size

        # Prefill policy
        # Get BLOCK_M from the backend for tile-budget admission logic
        attn_backend = self.tp_worker.model_runner.attn_backend
        if hasattr(attn_backend, "extend_attention_block_m"):
            prefill_tile_block_m = attn_backend.extend_attention_block_m
        else:
            prefill_tile_block_m = 64  # Fallback for non-Triton backends

        adder = PrefillAdder(
            self.page_size,
            self.tree_cache,
            self.token_to_kv_pool_allocator,
            running_batch,
            self.new_token_ratio_tracker.current,
            self.max_prefill_tokens,
            chunked_prefill_size,
            running_bs if self.is_mixed_chunk else 0,
            self.priority_scheduling_preemption_threshold,
            max_prefill_bs=int(self.max_prefill_bs),
            max_running_requests=self.max_running_requests,
            prefill_max_requests=get_schedule().prefill_max_requests,
            prefill_delayer_single_pass=prefill_delayer_single_pass,
            dllm_config=self.dllm_config,
            waiting_queue_len=len(self.waiting_queue),
            prefill_tile_block_m=prefill_tile_block_m,
        )

        if self.chunked_req is not None:
            self.chunked_req.init_next_round_input()
            adder.chunked_req_limit = self.policy.shortest_prefill_chunk_limit(
                self.chunked_req,
                self.waiting_queue,
                adder.rem_chunk_tokens or 0,
                self.page_size,
            )
            self.chunked_req = adder.add_chunked_req(self.chunked_req)

        if self.enable_lora:
            running_loras = {
                req.lora_id for req in running_batch.reqs if not req.finished()
            }
            # Account for LoRAs that are already loaded in the adder, such as chunked requests
            running_loras.update(req.lora_id for req in adder.can_run_list)

            if self.lora_drainer:
                self.lora_drainer.update_draining_state(
                    self.waiting_queue,
                    running_batch.reqs,
                )

        mamba_allocator = getattr(self.req_to_token_pool, "mamba_allocator", None)
        if mamba_allocator is not None:
            mamba_allocator.alloc_group_begin(len(self.waiting_queue))
        buffer_pipeline = self.tree_cache.buffer_pipeline
        # Get requests from the waiting queue to a new prefill batch
        for req in self.waiting_queue:
            if self.enable_lora and not self.can_schedule_lora_req(req, running_loras):
                continue

            running_bs = len(running_batch.reqs)
            candidate_beam_width = (
                req.beam_group.beam_width if req.beam_group is not None else None
            )
            if len(adder.can_run_list) >= self.get_num_allocatable_reqs(
                running_bs,
                candidate_beam_width,
                running_batch=running_batch,
            ):
                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():
                    running_batch.batch_is_full = True

            if running_batch.batch_is_full:
                if not self.enable_priority_preemption or not adder.preempt_to_schedule(
                    req
                ):
                    break

            if self.enable_hicache_storage:
                prefetch_done = self.tree_cache.check_prefetch_progress(
                    req.cache_request_handle
                )
                if not prefetch_done:
                    # skip staging requests that are ongoing prefetch
                    continue
                # Pop the L3-loaded span. Unified cache exposes its absolute
                # start so cache-mode L2/L3 attribution survives L3-tail eviction.
                loaded_tokens, loaded_start = self.tree_cache.pop_prefetch_loaded_span(
                    req.cache_request_handle
                )
                if loaded_tokens > 0:
                    req.storage_hit_length = loaded_tokens
                    req.storage_hit_start = loaded_start
                    # Cache-mode host memory is a resident L2 tier. Buffer mode
                    # marks the staged span below once it is surfaced.
                    req.host_hit_is_storage = False

            req.init_next_round_input(self.tree_cache)
            if self.enable_hicache_storage and (
                self._prefetch_after_device_hit_loss(req)
            ):
                continue
            if (
                self.enable_hicache_storage
                and buffer_pipeline is not None
                and not buffer_pipeline.prepare_staged_prefetch(req)
            ):
                continue
            res = adder.add_one_req(
                req,
                has_chunked_req=(self.chunked_req is not None),
                truncation_align_size=self.truncation_align_size,
            )

            if self.enable_lora:
                running_loras.add(req.lora_id)

            if res != AddReqResult.CONTINUE:
                if res == AddReqResult.NO_TOKEN:
                    if (
                        self.enable_hierarchical_cache
                        or self.enable_unified_cache_external_linker
                    ):
                        # Set batch_is_full after making sure there are requests that can be served
                        running_batch.batch_is_full = len(adder.can_run_list) > 0 or (
                            not running_batch.is_empty()
                        )
                    else:
                        running_batch.batch_is_full = True
                # revert matched mamba idx to avoid memory leak, if req is not added.
                # Only free if the slot was freshly allocated in this batch (not
                # pre-existing from a session). Session-held slots have their own
                # lifecycle and freeing them here causes double-free.
                added = len(adder.can_run_list) > 0 and req is adder.can_run_list[-1]
                if not added:
                    # init_next_round_input() may stage deferred Mamba COW/clear
                    # metadata before add_one_req() rejects the request.
                    req.kv.mamba_cow_src_index = None
                    req.kv.mamba_needs_clear = False
                    if req.kv.holds_mamba and not getattr(req, "session", None):
                        self.tree_cache.req_to_token_pool.mamba_allocator.free(
                            req.kv.mamba_pool_idx.unsqueeze(-1)
                        )
                        req.kv.mamba_pool_idx = None
                break

        if mamba_allocator is not None:
            mamba_allocator.alloc_group_end()

        # Update waiting queue
        can_run_list: List[Req] = adder.can_run_list
        if len(can_run_list) == 0:
            return None, running_batch

        can_run_set = set(can_run_list)
        retries = self.tree_cache.storage_prefetch_retries
        if self.enable_hicache_storage and retries is not None:
            for req in can_run_list:
                retries.cancel(req.rid)
        self.waiting_queue = [x for x in self.waiting_queue if x not in can_run_set]
        if adder.preempt_list:
            for req in adder.preempt_list:
                self._add_request_to_queue(req)

        if adder.new_chunked_req is not None:
            # Update chunked prefill
            assert self.chunked_req is None
            self.chunked_req = adder.new_chunked_req

        if self.chunked_req is not None:
            self.chunked_req.inflight_middle_chunks += 1

        set_time_batch(can_run_list, "set_forward_entry_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,
        )

        new_batch.contains_last_prefill_chunk = (
            self.chunked_req is None or len(can_run_list) != 1
        )

        if self.enable_hierarchical_cache or self.enable_unified_cache_external_linker:
            # 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()

        if self.tp_worker.model_runner.prefill_aware_swa:
            for req in can_run_list:
                req.kv.swa_evict_floor = req.extend_range.end

        # Record prefill stats for logging after forward.
        new_batch.prefill_stats = PrefillStats.from_adder(
            adder,
            running_batch.reqs,
            self.enable_priority_scheduling,
            num_pending_tokens=self.load_inquirer._get_num_pending_tokens(
                chunk_deduct=(
                    self.chunked_req.extend_range.length
                    if self.chunked_req is not None
                    else 0
                ),
            ),
        )

        # Mixed-style chunked prefill
        if (
            self.is_mixed_chunk
            and not running_batch.is_empty()
            and not (new_batch.return_logprob or running_batch.return_logprob)
            # mix_with_running cats input_ids but not input_embeds — shapes would mismatch
            and new_batch.input_embeds is None
            # Beam member rows are not supported inside a mixed extend batch.
            and all(r.beam_group is None for r in running_batch.reqs)
        ):
            # TODO (lianmin): support return_logprob + mixed chunked prefill
            running_batch.filter_batch()
            if not running_batch.is_empty():
                running_batch.prepare_for_decode()
                new_batch.mix_with_running(running_batch)
                new_batch.decoding_reqs = running_batch.reqs
                if not self.enable_overlap and not self.spec_algorithm.is_none():
                    # Non-overlap spec never writes the relay; stash the
                    # tails' pending tokens for the mixed input resolve.
                    last_tokens = torch.tensor(
                        [r.output_ids[-1] for r in running_batch.reqs],
                        dtype=torch.int64,
                        device=self.device,
                    )
                    self.future_map.stash_bonus_tokens(
                        running_batch.req_pool_indices, last_tokens
                    )
                running_batch = ScheduleBatch(
                    reqs=[], batch_is_full=running_batch.batch_is_full
                )
        else:
            new_batch.decoding_reqs = None

        return new_batch, running_batch

    def can_schedule_lora_req(
        self, req: Req, running_loras: set[Optional[str]]
    ) -> bool:
        """
        Check if a LoRA request can be scheduled.

        This method checks two conditions:
        1. The drainer allows scheduling (based on draining state)
        2. The LoRA adapter can be loaded (either already running or can be added)
        """
        if self.lora_drainer and not self.lora_drainer.can_schedule(req):
            return False

        if req.lora_id in running_loras:
            return True

        if self.enable_lora_overlap_loading:
            # For overlapping loading of LoRA weights with computation, we will load each
            # adapter one at a time, as opposed to loading them in one batch
            return self.lora_overlap_loader.try_overlap_load_lora(
                req.lora_id, running_loras
            )
        else:
            new_lora_set = {req.lora_id} | running_loras
            return self.tp_worker.model_runner.lora_manager.validate_lora_batch(
                new_lora_set
            )

    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 (kv_full_retract_flag := not batch.check_decode_mem()) or (
            TEST_RETRACT and self.forward_ct % TEST_RETRACT_INTERVAL == 0
        ):
            old_available_tokens = self.token_to_kv_pool_allocator.available_size()
            old_ratio = self.new_token_ratio_tracker.current
            mamba_allocator = getattr(
                self.tree_cache.req_to_token_pool, "mamba_allocator", None
            )
            old_mamba_available = (
                mamba_allocator.available_size()
                if mamba_allocator is not None
                else None
            )
            retracted_reqs, new_token_ratio, reqs_to_abort = batch.retract_decode()
            new_available_tokens = self.token_to_kv_pool_allocator.available_size()
            new_token_gained = new_available_tokens - old_available_tokens
            mamba_num_gained = (
                mamba_allocator.available_size() - old_mamba_available
                if mamba_allocator is not None
                else None
            )

            self.metrics_reporter.num_retracted_reqs = len(retracted_reqs)
            if self.metrics_reporter.enable_metrics and len(retracted_reqs) > 0:
                self.metrics_reporter.metrics_collector.increment_retracted_reqs(
                    num_retracted_reqs=len(retracted_reqs),
                    num_retracted_input_tokens=sum(
                        len(r.origin_input_ids) for r in retracted_reqs
                    ),
                    num_retracted_output_tokens=sum(
                        len(r.output_ids) for r in retracted_reqs
                    ),
                )
            self.new_token_ratio_tracker.current = new_token_ratio
            for req in reqs_to_abort:
                abort_reason: FINISH_ABORT = req.to_finish
                self.ipc_channels.send_to_tokenizer.send_output(
                    _make_abort_req(req, finished_reason=abort_reason.to_json()),
                    req,
                )
            for req in reqs_to_abort:
                # Member rows were freed inside retract_decode; the group only
                # has to leave the live set.
                self.beam_coordinator.retire_group(req)

            msg_prefix = (
                "KV cache pool is full. Retract requests. "
                if kv_full_retract_flag
                else "Testing retraction. "
            )
            msg_details = f"#retracted_reqs: {len(retracted_reqs)}, #new_tokens_gained: {new_token_gained}"
            if mamba_num_gained is not None:
                msg_details += f", #mamba_num_gained: {mamba_num_gained}"
            if kv_full_retract_flag:
                msg_details += (
                    f", #new_token_ratio: {old_ratio:.4f} -> {new_token_ratio:.4f}"
                )
            logger.warning(msg_prefix + msg_details)

            for req in retracted_reqs:
                self._add_request_to_queue(req, is_retracted=True)
        else:
            self.new_token_ratio_tracker.decay_step()

        if batch.batch_size() < initial_bs:
            batch.batch_is_full = False

        if batch.is_empty():
            return batch

        # Update batch tensors
        batch.prepare_for_decode()
        return batch

    def record_batch_in_overlap(self, batch: ScheduleBatch):
        # 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
        # Snapshot all fields: spec V2 rebinds seq_lens / spec_info mid-forward.
        attr_snapshot = [
            getattr(batch, f.name, None) for f in dataclasses.fields(batch)
        ]
        self.batch_record_ct = (self.batch_record_ct + 1) % 2
        # List (not tuple) so that workers can register additional refs via
        # GenerationBatchResult.extra_keep_alive_refs after forward returns.
        self.batch_record_buf[self.batch_record_ct] = [batch, attr_snapshot]

    @contextmanager
    def _forward_isolation(self, batch: ScheduleBatch, *, overlap: bool):
        """Make SB transactional across one forward (overlap and non-overlap).

        1. Snapshot SB fields so V2's mid-forward mutations (forward_mode /
           input_ids / seq_lens / spec_info / ...) can be undone. V1 / non-spec
           only need sampling_info restored - V1 carries spec_info forward as
           next-iter draft input.
        2. Substitute sampling_info with a forward-only copy (orchestrator=None,
           shares the pre-accumulated penalty buffer) so V2's multiple init_new
           calls don't double-accumulate penalties.
        3. (overlap=True only) Pin (batch, snapshot) into batch_record_buf
           for 2 iters so GPU tensors in the snapshot survive the caching
           allocator past the forward stream. Must run AFTER the sampling_info
           swap so the forward-only copy gets pinned. The non-overlap (sync) path
           runs on a single stream and doesn't allocate batch_record_buf, so it
           passes overlap=False.
        """
        # 1. snapshot
        snapshot_v2_full = not batch.spec_algorithm.is_none()
        sched_snapshot = (
            {f.name: getattr(batch, f.name) for f in dataclasses.fields(batch)}
            if snapshot_v2_full
            else None
        )
        sched_sampling_info = batch.sampling_info

        # 2. sampling_info substitute
        if sched_sampling_info is not None:
            batch.sampling_info = sched_sampling_info.copy_for_forward()

        # 3. pin for 2-iter tensor lifetime (overlap path only)
        if overlap:
            self.record_batch_in_overlap(batch)

        try:
            yield
        finally:
            if snapshot_v2_full:
                for name, value in sched_snapshot.items():
                    setattr(batch, name, value)
            else:
                batch.sampling_info = sched_sampling_info

    @scheduler_stage_method(SCHEDULER_STAGE_RUN_BATCH)
    def run_batch(
        self,
        batch: ScheduleBatch,
        pp_proxy_tensors: Optional[PPProxyTensors] = None,
    ) -> Union[GenerationBatchResult, EmbeddingBatchResult]:
        """Run a batch."""
        self.metrics_reporter.record_scheduler_active()
        self.forward_ct += 1
        batch.forward_iter = self.forward_ct
        batch.launch_ts = time.monotonic()
        is_split_prefill = batch.forward_mode.is_split_prefill()
        if not is_split_prefill or batch.split_index == 0:
            batch.after_idle_gap = self._sched_idled
            if is_split_prefill:
                batch.split_prefill_start = (batch.forward_iter, batch.launch_ts)
        self._sched_idled = False

        # Accumulate the prefill-token counter used by the HRRN scheduling policy. Decode / prebuilt batches contribute 0.
        if batch.extend_num_tokens:
            self.processed_tokens_counter += batch.extend_num_tokens

        if self.scripted_scheduler_hook is not None:
            self.scripted_scheduler_hook.on_run_batch(batch)

        # Whether to run the profiler
        self.profiler_manager._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)

        # Place holder handling for pd-disagg decode event loop
        if batch.forward_mode.is_prebuilt():
            return self._run_batch_prebuilt(batch)

        # PD prefill: early-send cached prefix KV, overlapping the suffix forward.
        if self.disaggregation_mode == DisaggregationMode.PREFILL:
            for req in batch.reqs:
                self.maybe_send_cached_prefix_chunk(req)

        # Run forward
        if self.is_generation:
            if self.enable_overlap:
                # Self-gates on batch.spec_info.future_indices; non-spec_v2
                # no-ops (ForwardBatch.init_new lazily computes the sum).
                self.future_map.resolve_seq_lens_cpu(batch)
                if self._confidence_budget_prepare is not None:
                    self._confidence_budget_prepare(batch, self.future_map)

                with self.forward_stream_ctx:
                    self.forward_stream.wait_stream(self.schedule_stream)
                    # resolve consumes SB staging (prefill_input_ids_cpu /
                    # mix_running_indices). Run OUTSIDE isolation so the
                    # snapshot captures the post-consume state — restoring
                    # post-forward must not un-consume staging.
                    resolve_forward_inputs(batch, self.future_map)

                    with self._forward_isolation(batch, overlap=True):
                        future_indices = batch.req_pool_indices

                        # Spec_v2 fires on_publish mid-worker (between verify and
                        # draft_extend) so schedule prep can overlap with draft_extend.
                        # Non-spec has no later work — scheduler publishes after return.
                        fwd_kwargs = {}
                        if not batch.spec_algorithm.is_none():
                            fwd_kwargs["on_publish"] = partial(
                                self.future_map.publish, future_indices
                            )
                            # Grammar-overlap-capable workers advance the grammar FSM
                            # inside verify() before building the bitmask; hand them the
                            # barrier that resolves the previous batch's committed
                            # tokens (overlapping the target forward).
                            if batch.spec_algorithm.supports_grammar_overlap():
                                fwd_kwargs["grammar_barrier"] = (
                                    self._advance_pending_grammar
                                )

                        # FIXME: pp is not compatible with overlap
                        batch_result = self.model_worker.forward_batch_generation(
                            batch, **fwd_kwargs
                        )
                        if batch.spec_algorithm.is_none():
                            self.future_map.publish(future_indices, batch.seq_lens + 1)
                        # Park any refs the worker wants kept alive 2 iters
                        # (cross-stream tensor lifetime; pinned in the same
                        # ring slot as the SB attr snapshot).
                        if batch_result.extra_keep_alive_refs:
                            self.batch_record_buf[self.batch_record_ct].extend(
                                batch_result.extra_keep_alive_refs
                            )
                        if self.enable_unified_memory:
                            # Record a `forward_done` event after the forward (before
                            # copy_to_cpu); lazy-compaction `_flush` gates src reuse on
                            # it. Only the unified pool's allocator exposes these hooks.
                            allocator = self.token_to_kv_pool_allocator
                            forward_done = self.device_module.Event()
                            forward_done.record(stream=self.forward_stream)
                            allocator.set_latest_forward_done_event(forward_done)
                            # Write-set classification: hand the allocator this
                            # forward's virtual out_cache_loc as a tensor ref (no GPU work).
                            allocator.set_inflight_forward(
                                forward_done,
                                batch.out_cache_loc,
                            )
                        # FIXME(lsyin): maybe move this to forward_batch_generation
                        batch_result.copy_done = self.device_module.Event()
                        if batch_result.delay_sample_func is None:
                            self._relay_forward_payload(
                                batch, future_indices, batch_result
                            )
                            if _is_hip:
                                # Cross-stream sync costs more than the tiny D2H it
                                # overlaps.
                                batch_result.copy_to_cpu(
                                    return_logprob=batch.return_logprob,
                                    return_hidden_states=batch.return_hidden_states,
                                )
                            else:
                                # Result D2H on copy_stream overlaps the next forward
                                # instead of serializing on forward_stream; it's a leaf
                                # gated by copy_done, so nothing on forward_stream waits.
                                self.copy_stream.wait_stream(self.forward_stream)
                                with self.copy_stream_ctx:
                                    batch_result.copy_to_cpu(
                                        return_logprob=batch.return_logprob,
                                        return_hidden_states=batch.return_hidden_states,
                                    )
                        else:
                            batch_result.future_indices = future_indices

                # Next-iter input_ids relayed via future_map.
                batch.input_ids = None

                if not batch.spec_algorithm.is_none():
                    batch.spec_info = batch_result.next_draft_input
                    batch.spec_info.future_dsa_topk_indices_available = (
                        batch.spec_info.dsa_topk_indices is not None
                    )
                    batch.spec_info.future_indices = future_indices
            elif self.enable_pdmux and batch.forward_mode.is_split_prefill():
                resolve_forward_inputs(batch, self.future_map)
                batch_result = self.tp_worker.forward_batch_split_prefill(batch)
                self._relay_forward_payload(batch, batch.req_pool_indices, batch_result)
                batch.input_ids = None
                self._copy_auxiliary_output_to_cpu(batch, batch_result)
            elif not batch.spec_algorithm.is_none():
                is_verify_round = get_parallel().pp_size > 1 and not (
                    batch.forward_mode.is_extend() or batch.is_extend_in_batch
                )
                # The relayed tree is what the requests carry between rounds;
                # the rebuild below swaps it for this round's verify input, so
                # hold on to it and put it back once the forward is done.
                relay_input = batch.spec_info if is_verify_round else None
                if is_verify_round:
                    # PP+spec decode: every stage rebuilds the same verify
                    # input from relayed per-req state (draft lives on the
                    # last stage only).
                    self._pp_spec_rebuild_verify_input(batch)
                if not self.pp_group.is_last_rank:
                    # PP+spec: non-last stages run only their model chunk on
                    # the verify-shaped batch; sampling, accept and draft all
                    # live on the last stage. The plain tp_worker path already
                    # returns pp_hidden_states_proxy_tensors for relay.
                    resolve_forward_inputs(batch, self.future_map)
                    if is_verify_round:
                        from sglang.srt.speculative.eagle_utils import (
                            eagle_prepare_for_verify,
                        )

                        # Isolation is load-bearing: eagle_prepare_for_verify
                        # mutates SB fields (forward_mode -> TARGET_VERIFY,
                        # input_ids, out_cache_loc); without the restore the
                        # next get_next_batch_to_run treats this decode batch
                        # as extend and re-merges it (duplicate reqs).
                        with self._forward_isolation(batch, overlap=False):
                            verify_forward_batch, can_run_cuda_graph = (
                                eagle_prepare_for_verify(
                                    batch.spec_info,
                                    self.req_to_token_pool,
                                    batch,
                                    self.tp_worker,
                                )
                            )
                            batch_result = self.tp_worker.forward_batch_generation(
                                batch=None,
                                forward_batch=verify_forward_batch,
                                pp_proxy_tensors=pp_proxy_tensors,
                                is_verify=True,
                            )
                        batch_result.can_run_cuda_graph = can_run_cuda_graph
                        # The isolation above restores batch.out_cache_loc, but
                        # this stage still has to compact its own accepted-path
                        # KV once the last stage relays which nodes it kept, so
                        # the verify slots have to outlive the forward.
                        batch_result.spec_verify_out_cache_loc = (
                            verify_forward_batch.out_cache_loc
                        )
                    else:
                        batch_result = self.model_worker.forward_batch_generation(
                            batch, pp_proxy_tensors=pp_proxy_tensors
                        )
                    batch.input_ids = None
                    # The verify input is per-round; between iterations
                    # spec_info carries the relayed tree, which is
                    # merge/filter-safe.
                    batch.spec_info = relay_input
                else:
                    # Non-overlap: drive the V2 worker synchronously (no
                    # future_map relay / on_publish).
                    resolve_forward_inputs(batch, self.future_map)
                    with self._forward_isolation(batch, overlap=False):
                        batch_result = self.model_worker.forward_batch_generation(
                            batch, pp_proxy_tensors=pp_proxy_tensors
                        )
                    # The isolation restore reverted the worker's in-forward SB edits;
                    # re-apply what must carry to the next iter. Under PP the
                    # tail draft already consumed the draft input in-round, and
                    # the next round's tree comes from the relay, so the last
                    # stage carries the same relayed tree as the others.
                    batch.spec_info = (
                        relay_input
                        if is_verify_round
                        else batch_result.next_draft_input
                    )
                    if batch_result.new_seq_lens is not None:
                        batch.seq_lens = batch_result.new_seq_lens
                        if batch.seq_lens_cpu is not None:
                            batch.seq_lens_cpu = batch_result.new_seq_lens.to("cpu")
                            batch.seq_lens_sum = int(batch.seq_lens_cpu.sum())
                    batch.input_ids = None  # rebuilt next iter from draft_token
                    self.update_cache_from_scheduler(batch, batch_result)
                    # Only the last PP rank owns real results requiring D2H; other ranks
                    # consume device tensors rebuilt from the output ring.
                    batch_result.copy_done = self.device_module.Event()
                    if (
                        batch_result.has_sampled_token_ids
                        and get_parallel().pp_size == 1
                    ):
                        batch_result.copy_to_cpu(
                            return_logprob=batch.return_logprob,
                            return_hidden_states=batch.return_hidden_states,
                        )
            else:
                kwargs = (
                    {"pp_proxy_tensors": pp_proxy_tensors}
                    if self.spec_algorithm.is_none()
                    else {}
                )
                resolve_forward_inputs(batch, self.future_map)
                batch_result = self.model_worker.forward_batch_generation(
                    batch, **kwargs
                )
                if batch_result.has_sampled_token_ids:
                    # Non-spec: relay via future_map, gathered next iter.
                    self._relay_forward_payload(
                        batch, batch.req_pool_indices, batch_result
                    )
                    batch.input_ids = None
                self.update_cache_from_scheduler(batch, batch_result)
                self._copy_auxiliary_output_to_cpu(batch, batch_result)

            # 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 batch.return_hidden_states:
                batch_result.extend_input_len_per_req = [
                    req.extend_range.length if req.extend_range is not None else 0
                    for req in batch.reqs
                ]
            else:
                batch_result.extend_input_len_per_req = None

            if batch.return_logprob:
                batch_result.extend_logprob_start_len_per_req = (
                    batch.extend_logprob_start_lens
                )
            else:
                batch_result.extend_logprob_start_len_per_req = None

            ret = batch_result
        else:  # embedding or reward model
            if self.enable_overlap:
                self.record_batch_in_overlap(batch)
                with self.forward_stream_ctx:
                    self.forward_stream.wait_stream(self.schedule_stream)
                    resolve_forward_inputs(batch, self.future_map)
                    pooler_output, can_run_cuda_graph = (
                        self.tp_worker.forward_batch_embedding(batch)
                    )
                    ret = EmbeddingBatchResult(
                        embeddings=pooler_output.embeddings,
                        pooled_hidden_states=pooler_output.pooled_hidden_states,
                        can_run_cuda_graph=can_run_cuda_graph,
                    )
                    ret.copy_to_cpu()
            else:
                resolve_forward_inputs(batch, self.future_map)
                pooler_output, can_run_cuda_graph = (
                    self.tp_worker.forward_batch_embedding(batch)
                )
                ret = EmbeddingBatchResult(
                    embeddings=pooler_output.embeddings,
                    pooled_hidden_states=pooler_output.pooled_hidden_states,
                    can_run_cuda_graph=can_run_cuda_graph,
                )

        self._maybe_report_active_ranks()

        return ret

    def _maybe_report_active_ranks(self) -> None:
        if not (
            self.enable_dp_attention and get_exec().moe.elastic_ep_backend is not None
        ):
            return
        from sglang.srt.elastic_ep.elastic_ep import ElasticEPStateManager

        inst = ElasticEPStateManager.instance()
        if inst is not None and inst.active_ranks_cpu is not None:
            self.ipc_channels.send_to_tokenizer.send_output(
                ActiveRanksOutput(
                    status=[bool(x) for x in inst.active_ranks_cpu.tolist()]
                )
            )
        else:
            logger.debug("[Elastic EP] active rank state is unavailable")
            return

        model_runner = self.tp_worker.model_runner
        pending = model_runner._pending_elastic_scale_update
        if pending is not None:
            self.ipc_channels.send_to_tokenizer.send_output(pending)
            model_runner._pending_elastic_scale_update = None

    def _relay_forward_payload(
        self,
        batch: ScheduleBatch,
        future_indices: torch.Tensor,
        batch_result: GenerationBatchResult,
    ) -> None:
        """Stash this iter's relay payload for next iter's resolve_forward_inputs."""
        if self.spec_algorithm.is_ngram():
            if batch_result.next_draft_input is not None:
                payload = RelayPayload.from_ngram(batch_result.next_draft_input)
                self.future_map.stash(future_indices, payload)
            return
        if batch_result.next_draft_input is not None:
            payload = RelayPayload.from_draft_input(batch_result.next_draft_input)
        elif batch_result.has_sampled_token_ids:
            payload = RelayPayload(bonus_tokens=batch_result.next_token_ids)
        else:
            return
        if batch.beam_tail is not None:
            # The worker sliced the tail off before sampling, so sampled tokens
            # cover only the reqs-aligned rows; the coordinator relays the rest.
            future_indices = future_indices[: batch.beam_tail.num_base_rows]
        self.future_map.stash(future_indices, payload)
        self.beam_coordinator.maybe_select_and_relay(
            batch, batch_result, chunked_req=self.chunked_req
        )

    def _copy_auxiliary_output_to_cpu(
        self,
        batch: ScheduleBatch,
        result: GenerationBatchResult,
    ) -> None:
        logits_output = result.logits_output
        if (
            logits_output is None
            or logits_output.auxiliary_device_output is None
            or result.auxiliary_host_output is not None
        ):
            return
        # PP transports the device output to the first rank before copying it.
        if get_parallel().pp_size > 1:
            return
        if result.copy_done is not None:
            raise RuntimeError(
                "generation result has an uncopied auxiliary output after its "
                "device-to-host copy was scheduled"
            )
        result.copy_done = self.device_module.Event()
        result.copy_to_cpu(
            return_logprob=batch.return_logprob,
            return_hidden_states=batch.return_hidden_states,
        )

    def launch_batch_sample_if_needed(
        self, batch_result: GenerationBatchResult, cur_batch: ScheduleBatch
    ) -> Union[GenerationBatchResult]:
        # 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.schedule_stream)
            _batch_result = batch_result.delay_sample_func()
            assert _batch_result is batch_result
            # Delay-sample is non-spec only; relays the sampled bonus tokens.
            self._relay_forward_payload(
                cur_batch, batch_result.future_indices, batch_result
            )

        # Run device-to-host copy on a separate stream to avoid blocking the
        # forward stream. The copy waits for the sampled result and can overlap
        # with subsequent forward computation.
        self.copy_stream.wait_stream(self.forward_stream)
        with self.copy_stream_ctx:
            batch_result.copy_to_cpu(
                return_logprob=cur_batch.return_logprob,
                return_hidden_states=cur_batch.return_hidden_states,
            )

        # Release the closure and large GPU tensors that are no longer needed.
        # The delay_sample_func closure captures forward_batch (which holds
        # sampling_info with vocab_mask) and logits_output (which holds
        # next_token_logits). Without clearing these, they stay alive via
        # batch_result in result_queue and batch_record_buf until the next
        # iteration, causing a steady VRAM leak with structured output.
        batch_result.delay_sample_func = None
        if batch_result.logits_output is not None:
            batch_result.logits_output.next_token_logits = None

    @scheduler_stage_method(SCHEDULER_STAGE_PROCESS_BATCH_RESULT)
    def process_batch_result(
        self,
        batch: ScheduleBatch,
        result: Union[GenerationBatchResult, EmbeddingBatchResult],
    ):
        # Flush async trace ops here: in overlap mode this CPU work runs while
        # the next batch's GPU forward is in flight, giving free overlap.
        flush_trace_batch(batch.reqs)
        snapshot = self.publish_load_snapshot(force=batch.forward_mode.is_extend())
        # Router-facing gauge on the dedicated PUB socket, reusing the
        # snapshot above rather than walking the queues again.
        self.load_publisher.publish_load_stat(
            self.load_inquirer.get_loads,
            force=batch.forward_mode.is_extend(),
            snapshot=snapshot,
        )

        if batch.forward_mode.is_decode():
            self.batch_result_processor.process_batch_result_decode(batch, result)
        elif batch.forward_mode.is_extend():
            if batch.is_dllm():
                self.process_batch_result_dllm(batch, result)
            elif self.disaggregation_mode == DisaggregationMode.PREFILL:
                self.process_batch_result_disagg_prefill(batch, result)
            else:
                self.batch_result_processor.process_batch_result_prefill(batch, result)
        elif batch.forward_mode.is_prebuilt():
            self.batch_result_processor.process_batch_result_prebuilt(batch)
        elif batch.forward_mode.is_idle():
            self.batch_result_processor.process_batch_result_idle(batch, result)

        # Submit this batch's queued host backups before the next scheduler step.
        self.tree_cache.flush_pending_backups()

        self._record_step_counters(batch, result)

        self.metrics_reporter.log_batch_result_stats(batch, result)

        # Emit forward pass metrics (every iteration when enabled)
        if self.enable_fpm:
            self.metrics_reporter._emit_forward_pass_metrics(batch, result)

        self._maybe_clear_mm_inputs(batch)
        self.maybe_send_health_check_signal()
        self.metrics_reporter.update_device_timer()

    def _record_step_counters(
        self, batch: ScheduleBatch, result: GenerationBatchResult
    ) -> None:
        mode = batch.forward_mode
        is_prefill = mode.is_extend_without_speculative()
        if not (is_prefill or mode.is_decode() or mode.is_target_verify()):
            return
        if all(is_health_check_generate_req(req) for req in batch.reqs):
            return
        if is_prefill and mode.is_split_prefill():
            start_iter, start_ts = batch.split_prefill_start
        else:
            start_iter, start_ts = batch.forward_iter, batch.launch_ts
        prev = self._prev_step
        self._prev_step = (batch.forward_iter, batch.launch_ts, is_prefill)
        # An idle pass keeps forward_iter contiguous (forward_ct advances in run_batch).
        contiguous = False
        if prev is not None:
            prev_iter, prev_launch_ts, prev_is_prefill = prev
            contiguous = (
                not batch.after_idle_gap
                and prev_iter + 1 == start_iter
                and prev_is_prefill == is_prefill
            )
        if is_prefill:
            # Completion boundaries include the burst tail and scheduling overhead
            # without double-counting overlapping launch-to-result spans.
            end_ts = time.monotonic()
            prev_end_ts = self._prev_prefill_end_ts
            self._prev_prefill_end_ts = end_ts
            if prev_end_ts is not None:
                # A new burst or mode can still overlap an earlier prefill.
                start_ts = prev_end_ts if contiguous else max(start_ts, prev_end_ts)
            step_us = int((end_ts - start_ts) * 1e6)
        else:
            # Decode regression uses launch cadence and skips boundaries.
            if not contiguous:
                return
            step_us = int((batch.launch_ts - prev_launch_ts) * 1e6)
        if not 0 < step_us < STEP_MAX_US:
            return
        if is_prefill:
            self.total_prefill_busy_us += step_us
            self.total_prefill_uncached_tokens += batch.extend_num_tokens
        else:
            batch_size = len(batch.reqs)
            _accumulate_decode_moment(
                self.decode_moment_totals,
                batch_size,
                step_us,
                result.get_num_generated_tokens(batch_size),
            )

    def maybe_send_health_check_signal(self):
        if self.return_health_check_ipcs:
            # 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.ipc_channels.send_to_tokenizer.send_output(
                HealthCheckOutput(
                    http_worker_ipc=self.return_health_check_ipcs.popleft()
                )
            )

    def add_external_corpus(
        self, recv_req: AddExternalCorpusReqInput
    ) -> Optional[AddExternalCorpusReqOutput]:
        if self.external_corpus_manager is None:
            return AddExternalCorpusReqOutput(
                success=False,
                message="Ngram speculative decoding is not enabled.",
            )
        return self.external_corpus_manager.add(recv_req)

    def remove_external_corpus(
        self, recv_req: RemoveExternalCorpusReqInput
    ) -> RemoveExternalCorpusReqOutput:
        if self.external_corpus_manager is None:
            return RemoveExternalCorpusReqOutput(
                success=False,
                message="Ngram speculative decoding is not enabled.",
            )
        return self.external_corpus_manager.remove(recv_req)

    def list_external_corpora(
        self, recv_req: ListExternalCorporaReqInput
    ) -> ListExternalCorporaReqOutput:
        if self.external_corpus_manager is None:
            return ListExternalCorporaReqOutput(
                success=False,
                message="Ngram speculative decoding is not enabled.",
            )
        return self.external_corpus_manager.list(recv_req)

    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)

    @scheduler_stage_method(SCHEDULER_STAGE_IDLE)
    def on_idle(self):
        """Idle housekeeping: guard, check, metrics, reset, sleep."""
        # Flush any health-check signal deferred while the engine was busy.
        self.maybe_send_health_check_signal()

        # Publish before the fully-idle gate: a no-batch-but-not-idle stall
        # (queues parked under KV pressure / disagg transfer) has no
        # process_batch_result to publish the growing gauge, and gating here
        # froze /get_loads, DP balancing, and the LoadStat for the stall. This
        # path is polled repeatedly, so a wall-clock floor bounds the
        # O(queue) get_loads for both sinks; the fully-idle publish runs
        # post-flush below.
        fully_idle = self.is_fully_idle()
        if not fully_idle:
            self.metrics_reporter.record_scheduler_active()
            now = time.monotonic()
            if now - self._last_stall_publish_ts >= LOAD_STALL_REFRESH_S:
                self._last_stall_publish_ts = now
                snapshot = self.publish_load_snapshot(force=True)
                self.load_publisher.publish_load_stat(
                    self.load_inquirer.get_loads, force=True, snapshot=snapshot
                )
            if (
                self.enable_hicache_storage
                or self.disaggregation_mode != DisaggregationMode.NULL
            ):
                # Storage and transfer workers need the GIL between I/O calls.
                # Singleton PD polls no longer yield through a collective.
                time.sleep(0)
            return
        self.metrics_reporter.record_scheduler_idle()

        if self.enable_unified_memory:
            try:
                self.token_to_kv_pool_allocator.flush_opportunistic()
            except Exception:
                pass

        # memory leak check (skipped for hisparse — pool counters intentionally
        # diverge during host-backup, see _get_swa_token_info clamp).
        # Also skipped while deferred KV releases are pending: they hold pages out
        # of the allocator by design, so the pool is transiently below `total` and
        # would trip the idle leak invariant. Resumes once the holds resolve.
        deferred_pending = (
            self.disaggregation_mode == DisaggregationMode.DECODE
            and self.disagg_decode_transfer_queue.has_pending_deferred_releases()
        )
        with self.scheduler_stage_metrics.record(SCHEDULER_STAGE_SANITY_CHECK_CACHE):
            if not self.enable_hisparse and not deferred_pending:
                has_leak, messages = self.invariant_checker._check_all_pools(
                    self.pool_stats_observer.get_pool_stats(),
                )
                if has_leak:
                    self.invariant_checker._report_leak("pool", "\n".join(messages))
                self.invariant_checker._check_req_pool()
                # Byte-conservation diagnostic (allocator-owned; static pools
                # return [] — the token identity above can't see byte leaks).
                byte_violations = (
                    self.token_to_kv_pool_allocator.verify_byte_accounting()
                )
                if byte_violations:
                    self.invariant_checker._report_leak(
                        "pool-bytes", "\n".join(byte_violations)
                    )

            # tree cache sanity check
            self.invariant_checker._check_tree_cache()

        # metrics every 30s
        self.metrics_reporter._maybe_log_idle_metrics()

        # kv event publishing
        self.kv_events_publisher.publish_kv_events()

        # reset token ratio
        self.new_token_ratio_tracker.reset()

        # Fully-idle publish, post-flush so the gauge reflects compacted KV.
        # Forced (immediate) so the busy->idle transition is never delayed.
        snapshot = self.publish_load_snapshot(force=True)
        self.load_publisher.publish_load_stat(
            self.load_inquirer.get_loads, force=True, snapshot=snapshot
        )

        # sleep until next event
        self.maybe_sleep_on_idle()
        self.metrics_reporter.record_scheduler_idle()

    def _record_scheduler_state_for_paused_engine(self) -> None:
        if self.is_fully_idle():
            self.metrics_reporter.record_scheduler_idle()
        else:
            self.metrics_reporter.record_scheduler_active()

    def is_fully_idle(self, for_health_check=False) -> bool:
        # Health check piggybacks on running requests in process_output.
        # Only running_batch + waiting_queue guarantee active GPU processing;
        # disagg queues (bootstrap/prealloc/transfer) may have items without
        # any request actually running on GPU — e.g. stuck handshake, full
        # KV cache, or stalled transfer — so they can't carry health info.
        # Batch running status
        idle = (
            self.running_batch.is_empty()
            and self.chunked_req is None
            and not self.dllm_manager.any_staging_reqs()
            and (self.last_batch is None or self.last_batch.is_empty())
            and (not self.enable_overlap or len(self.result_queue) == 0)
            and self._pp_microbatches_drained()
        )

        # Waiting queues: waiting + bootstrapping + preallocation + kv transfer (decode)
        idle &= len(self.waiting_queue) == 0

        if (
            for_health_check
            and not self._engine_paused
            and self.disaggregation_mode == DisaggregationMode.DECODE
            and self.disagg_decode_prealloc_queue is not None
        ):
            idle &= len(self.disagg_decode_prealloc_queue.retracted_queue) == 0

        if not for_health_check:
            # Grammar queue and prefill inflight queue may not produce batch
            # results instantly, but they still indicate the server is not idle.
            idle &= len(self.grammar_manager.grammar_queue) == 0
            if self.disaggregation_mode == DisaggregationMode.PREFILL:
                idle &= len(self.disagg_prefill_inflight_queue) == 0
                idle &= len(self.disagg_prefill_bootstrap_queue.queue) == 0

            if self.disaggregation_mode == DisaggregationMode.DECODE:
                idle &= len(self.disagg_decode_prealloc_queue.queue) == 0
                idle &= len(self.disagg_decode_prealloc_queue.retracted_queue) == 0
                idle &= len(self.disagg_decode_transfer_queue.queue) == 0
                if self.decode_offload_manager is not None:
                    idle &= len(self.decode_offload_manager.ongoing_offload) == 0

            # HiSparse: staging requests transitioning prefill -> decode
            if self.enable_hisparse:
                idle &= not self.hisparse_coordinator.has_ongoing_staging()

            # HiCache: in-flight async ops (GPU↔Host↔L3) must drain before
            # destructive operations like attach/detach/flush_cache.
            if self.enable_hierarchical_cache:
                tc = self.tree_cache
                idle &= len(tc.ongoing_write_through) == 0
                idle &= len(tc.ongoing_load_back) == 0
                if tc.enable_storage:
                    idle &= len(tc.ongoing_prefetch) == 0
                    idle &= len(tc.ongoing_backup) == 0
                    if get_memory().hicache_host_memory_mode == "buffer_only":
                        # Queued writes, staged prefetches, and in-flight
                        # storage writes still hold host staging
                        # (buffer-mode unified tree only).
                        idle &= tc.buffer_pipeline.is_idle()

        return idle

    def _pp_microbatches_drained(self) -> bool:
        if get_parallel().pp_size == 1:
            return True
        return all(x.is_empty() for x in self.running_mbs) and all(
            mb is None or mb.is_empty() for mb in self.mbs
        )

    def attach_hicache_storage_wrapped(
        self, recv_req: AttachHiCacheStorageReqInput
    ) -> AttachHiCacheStorageReqOutput:
        if not self.enable_hierarchical_cache:
            return AttachHiCacheStorageReqOutput(
                success=False, message="Hierarchical cache is not enabled."
            )

        if not self.is_fully_idle():
            return AttachHiCacheStorageReqOutput(
                success=False,
                message=(
                    "Reject attach: scheduler is not idle. "
                    f"#queue-req={len(self.waiting_queue)} "
                    f"#running-req={len(self.running_batch.reqs)}"
                ),
            )

        if not hasattr(self.tree_cache, "attach_storage_backend"):
            return AttachHiCacheStorageReqOutput(
                success=False,
                message="Current tree_cache implementation does not support dynamic attach.",
            )

        try:
            ok, msg = self.tree_cache.attach_storage_backend(
                storage_backend=recv_req.hicache_storage_backend,
                storage_backend_extra_config_json=recv_req.hicache_storage_backend_extra_config_json,
                served_model_name=get_serving().served_model_name,
                hicache_storage_prefetch_policy=recv_req.hicache_storage_prefetch_policy,
                hicache_write_policy=recv_req.hicache_write_policy,
            )
        except Exception as e:
            logger.exception("Attach HiCache storage backend failed with exception.")
            return AttachHiCacheStorageReqOutput(success=False, message=str(e))
        if ok:
            self.enable_hicache_storage = True
            hicache_fields = {
                "hicache_storage_backend": recv_req.hicache_storage_backend
            }
            if recv_req.hicache_storage_backend_extra_config_json is not None:
                hicache_fields["hicache_storage_backend_extra_config"] = (
                    recv_req.hicache_storage_backend_extra_config_json
                )
            if recv_req.hicache_storage_prefetch_policy is not None:
                hicache_fields["hicache_storage_prefetch_policy"] = (
                    recv_req.hicache_storage_prefetch_policy
                )
            if recv_req.hicache_write_policy is not None:
                hicache_fields["hicache_write_policy"] = recv_req.hicache_write_policy
            get_context().override("scheduler.attach_hicache", **hicache_fields)
            logger.info(
                f"Attached HiCache storage backend: {recv_req.hicache_storage_backend}"
            )
        return AttachHiCacheStorageReqOutput(success=ok, message=msg)

    def detach_hicache_storage_wrapped(
        self, recv_req: DetachHiCacheStorageReqInput
    ) -> DetachHiCacheStorageReqOutput:
        if not self.enable_hierarchical_cache:
            return DetachHiCacheStorageReqOutput(
                success=False, message="Hierarchical cache is not enabled."
            )

        if not self.is_fully_idle():
            return DetachHiCacheStorageReqOutput(
                success=False,
                message=(
                    "Reject detach: scheduler is not idle. "
                    f"#queue-req={len(self.waiting_queue)} "
                    f"#running-req={len(self.running_batch.reqs)}"
                ),
            )

        if not hasattr(self.tree_cache, "detach_storage_backend"):
            return DetachHiCacheStorageReqOutput(
                success=False,
                message="Current tree_cache implementation does not support dynamic detach.",
            )

        # Idempotent detach: even if scheduler thinks storage is disabled, we still
        # attempt best-effort cleanup in tree_cache (it may have leftover state).
        try:
            ok, msg = self.tree_cache.detach_storage_backend()
        except Exception as e:
            logger.exception("Detach HiCache storage backend failed with exception.")
            return DetachHiCacheStorageReqOutput(success=False, message=str(e))

        if ok or (not self.enable_hicache_storage):
            # Treat "already disabled / nothing to do" as success for idempotence.
            self.enable_hicache_storage = False
            get_context().override(
                "scheduler.detach_hicache",
                hicache_storage_backend=None,
                hicache_storage_backend_extra_config=None,
            )
            logger.info("Detached HiCache storage backend.")
            return DetachHiCacheStorageReqOutput(
                success=True, message=msg or "HiCache storage backend is detached."
            )

        return DetachHiCacheStorageReqOutput(success=False, message=msg)

    def flush_cache(self, empty_cache: bool = True):
        """Flush memory pools (e.g., KV cache, Mamba cache) and optionally empty device allocator cache."""
        if self.is_fully_idle():
            self.cur_batch_for_debug = None
            self.last_batch = None
            self.tree_cache.reset()
            self.req_to_token_pool.clear()
            self.token_to_kv_pool_allocator.clear()
            self.req_to_token_pool.reset_aux_cache_allocator()
            self.grammar_manager.clear()
            self.metrics_reporter.reset_metrics()

            if self.draft_worker:
                self.draft_worker.clear_cache_pool()

            if empty_cache:
                current_platform.empty_cache()
            # Per-DP-group leader logs once: ranks within a DP group are
            # state-synchronous, but DP groups may diverge.
            if self.metrics_reporter.is_stats_logging_rank:
                logger.info("Cache flushed successfully!")
            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)}"
            )
            success = False
        return success

    def get_internal_state(self, recv_req: GetInternalStateReq):
        # Resolved config (pristine server_args + post-publish overrides) so a
        # readback reflects values changed via /set_internal_state, not startup.
        ret = get_context().resolved_server_args_dict()
        ret["world_size"] = compute_world_size(
            enable_dp_attention=get_parallel().enable_dp_attention,
            dp_size=get_parallel().dp_size,
            tp_size=get_parallel().tp_size,
            pp_size=get_parallel().pp_size,
        )
        ret["last_gen_throughput"] = self.metrics_reporter.last_gen_throughput
        draft_graph_memory_usage = (
            None if self.draft_worker is None else self.draft_worker.graph_memory_usage
        )
        memory_usage = build_memory_usage(
            weight_gb=self.tp_worker.model_runner.weight_load_mem_usage,
            kv_cache_gb=self.token_to_kv_pool_allocator.get_kvcache().mem_usage,
            startup_available_gb=self.startup_available_gpu_memory_gb,
            token_capacity=self.max_total_num_tokens,
            token_capacity_swa=self.swa_tokens_per_layer,
            target_graph_memory_usage=self.tp_worker.graph_memory_usage,
            draft_graph_memory_usage=draft_graph_memory_usage,
        )
        ret["memory_usage"] = memory_usage
        ret["startup_time"] = self.startup_time
        ret["effective_max_running_requests_per_dp"] = self.max_running_requests
        # PD role switch: report this instance's role and the decode CUDA graph
        # batch sizes it captured, which a router feeds back as
        # PdRoleSwitchReqInput.decode_cuda_graph_bs. Unset until
        # init_disaggregation runs, which also re-derives it on every flip.
        disaggregation_mode = getattr(self, "disaggregation_mode", None)
        if disaggregation_mode is not None:
            ret["disaggregation_mode"] = disaggregation_mode.value
            ret["decode_cuda_graph_bs"] = self.tp_worker.get_decode_cuda_graph_bs()
            ret["decode_cuda_graph_memory_gb"] = round(
                sum(
                    memory_usage["graph"][phase]
                    for phase in (
                        "decode",
                        "target_verify",
                        "draft_decode",
                        "draft_extend",
                    )
                ),
                3,
            )

        if get_exec().moe.elastic_ep_backend is not None:
            from sglang.srt.elastic_ep.elastic_ep import ElasticEPStateManager

            ret["is_scaling_elastic_ep"] = ElasticEPStateManager.is_scaling()
            ret["effective_ep_size"] = ElasticEPStateManager.get_effective_ep_size()
            ret["pending_ep_size"] = ElasticEPStateManager.get_pending_ep_size()
            ret["scale_phase"] = ElasticEPStateManager.get_scale_phase()
            ret["elastic_ep_last_error"] = ElasticEPStateManager.get_last_error()

        if (
            not self.spec_algorithm.is_none()
            and self.metrics_reporter.spec_total_num_forward_ct > 0
        ):
            ret["avg_spec_accept_length"] = (
                self.metrics_reporter.spec_total_num_accept_tokens
                / self.metrics_reporter.spec_total_num_forward_ct
            )

        if RECORD_STEP_TIME:
            ret["step_time_dict"] = self.metrics_reporter.step_time_dict

        if self.spec_algorithm.is_dspark() and self.draft_worker is not None:
            info_record = self.draft_worker.dump_info_records()
            if info_record is not None:
                ret["dspark_info_record"] = info_record

        if envs.SGLANG_EXPOSE_OWN_ENV_VARS.get():
            ret["env_vars"] = exportable_env_vars()

        # A bound signal handler is not msgpack-serializable, and no reader
        # consumes it.
        ret.pop("custom_sigquit_handler", None)

        return GetInternalStateReqOutput(internal_state=msgspec_to_builtins(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",
                "dspark_force_budget_frac",
                "dspark_clear_info_records",
            ]
        )

        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 // get_parallel().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 // get_parallel().pp_size}]."
                )
                if_success = False
                break
            elif k == "dspark_force_budget_frac":
                if not self.spec_algorithm.is_dspark() or not hasattr(
                    self.draft_worker, "set_dspark_forced_budget_frac"
                ):
                    logging.warning(
                        "dspark_force_budget_frac requires a DSpark draft worker."
                    )
                    if_success = False
                    break
                if v is not None and not (0.0 < float(v) <= 1.0):
                    logging.warning(
                        f"dspark_force_budget_frac must be in (0, 1] or null, got {v}."
                    )
                    if_success = False
                    break
            elif k == "dspark_clear_info_records":
                if not self.spec_algorithm.is_dspark() or not hasattr(
                    self.draft_worker, "clear_info_records"
                ):
                    logging.warning(
                        "dspark_clear_info_records requires a DSpark draft worker."
                    )
                    if_success = False
                    break

        if if_success:
            if (
                not self.spec_algorithm.is_none()
                and self.metrics_reporter.spec_total_num_forward_ct > 0
            ):
                avg_spec_accept_length = (
                    self.metrics_reporter.spec_total_num_accept_tokens
                    / self.metrics_reporter.spec_total_num_forward_ct
                )
                logger.info(f"{avg_spec_accept_length=}")
            self.metrics_reporter.spec_total_num_accept_tokens = (
                self.metrics_reporter.spec_total_num_forward_ct
            ) = 0
            # DSpark control keys are worker commands, not server args; route
            # them to the draft worker and keep them out of the override.
            remaining = dict(server_args_dict)
            frac = remaining.pop("dspark_force_budget_frac", None)
            if "dspark_force_budget_frac" in server_args_dict:
                self.draft_worker.set_dspark_forced_budget_frac(
                    None if frac is None else float(frac)
                )
            if remaining.pop("dspark_clear_info_records", None):
                self.draft_worker.clear_info_records()
            if remaining:
                get_context().override(source="update_server_args", **remaining)
            logger.info(f"Config updated via context override: {remaining}")

        return SetInternalStateReqOutput(updated=if_success)

    def save_remote_model(self, **kwargs):
        self.weight_updater.save_remote_model(kwargs)

    def save_sharded_model(self, **kwargs):
        self.weight_updater.save_sharded_model(kwargs)

    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)
            if recv_req.parameters is not None:
                func(**recv_req.parameters)
            else:
                func()
        except Exception as e:
            success = False
            exec = e
            logger.error(f"Failed to call rpc {recv_req.method}: {str(e)}")

        barrier(group=self.tp_group.cpu_group)
        return RpcReqOutput(success=success, message="" if not exec else str(exec))

    def handle_update_weight_version(
        self, recv_req: UpdateWeightVersionReqInput
    ) -> UpdateWeightVersionReqOutput:
        self.record_weight_version_change(new_version=recv_req.new_version)
        return UpdateWeightVersionReqOutput()

    def record_weight_version_change(self, new_version: Optional[str]) -> None:
        if new_version is None or new_version == get_serving().weight_version:
            return

        old_version = get_serving().weight_version
        get_context().override("scheduler.weight_version", weight_version=new_version)

        live_reqs = {
            *self.collect_inflight_reqs(),
            *self.waiting_queue,
            *([self.chunked_req] if self.chunked_req is not None else []),
        }
        if self.hisparse_coordinator is not None:
            live_reqs.update(
                act.req for act in self.hisparse_coordinator.ack_staging_queue
            )
        num_recorded = record_weight_version_events(live_reqs, old_version=old_version)
        logger.info(
            f"Weight version changed. {old_version=} {new_version=} {num_recorded=}"
        )

    def collect_inflight_reqs(self) -> Set[Req]:
        if get_parallel().pp_size == 1:
            inflight_batches = [self.running_batch, self.last_batch]
        else:
            inflight_batches = [*self.running_mbs, *self.mbs]
        return {
            req for batch in inflight_batches if batch is not None for req in batch.reqs
        }

    def abort_request(self, recv_req: AbortReq):
        if (chunked_req := self.chunked_req) is not None:
            if recv_req.abort_all or chunked_req.rid.startswith(recv_req.rid):
                self._pending_chunked_abort_req = chunked_req

        # todo hisparse, release resources for abort requests in hisparse coordinator
        # Abort requests still waiting for encoder embeddings (EPD language-only)
        if self.mm_receiver is not None:
            self.mm_receiver.abort_waiting_requests(recv_req)

        # 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)
            self._release_aborted_request(req)
            self.beam_coordinator.retire_group(req)
            # Without the initiator's reason the tokenizer falls back to a
            # generic abort message.
            self.ipc_channels.send_to_tokenizer.send_output(
                _make_abort_req(req, finished_reason=recv_req.finished_reason), req
            )
            # For disaggregation decode mode, the request in the waiting queue has KV cache allocated.
            if self.disaggregation_mode == DisaggregationMode.DECODE:
                release_kv_cache(req, self.tree_cache)
            # For disaggregation prefill mode, free the metadata buffer index
            if self.disaggregation_mode == DisaggregationMode.PREFILL:
                bootstrap_pending = req.pending_bootstrap
                maybe_release_metadata_buffer(
                    req, self.req_to_metadata_buffer_idx_allocator
                )
                if (
                    bootstrap_pending
                    and hasattr(req, "disagg_kv_sender")
                    and req.disagg_kv_sender is not None
                ):
                    if hasattr(req.disagg_kv_sender, "abort"):
                        req.disagg_kv_sender.abort()

            # For mamba radix cache
            if (
                req.kv.holds_mamba
                and self.disaggregation_mode != DisaggregationMode.DECODE
            ):
                release_kv_cache(req, self.tree_cache, is_insert=False)
            logger.debug(f"Abort queued request. {req.rid=}")

        if self.dllm_config is not None:
            for req in self.dllm_manager.pop_aborted_reqs(
                recv_req.abort_all, recv_req.rid
            ):
                self._release_aborted_request(req)
                self.ipc_channels.send_to_tokenizer.send_output(
                    _make_abort_req(req), req
                )
                if req.kv.holds_kv or req.kv.holds_mamba:
                    release_kv_cache(req, self.tree_cache, is_insert=False)
                logger.debug(f"Abort dLLM queued request. {req.rid=}")

        # Delete the requests in the 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.
        self.grammar_manager.abort_requests(recv_req)

        # 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=}")
                    self._release_aborted_request(req)

                    if hasattr(req.disagg_kv_sender, "abort"):
                        req.disagg_kv_sender.abort()
                    if get_parallel().pp_size > 1:
                        prepare_abort(req, "Aborted by AbortReq.")

            # 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=}")
                    decode_req.kv_receiver.abort()
                    if get_parallel().pp_size > 1:
                        prepare_abort(decode_req.req, "Aborted by AbortReq.")

            # 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=}")
                    receiver = decode_req.kv_receiver
                    receiver.abort()
                    # Arm drain-ack accounting once the ABORT is sent, so acks
                    # arriving before this req is deferred (e.g. during the next
                    # forward step) are captured. A fresh set also drops stale acks
                    # from a prior request that reused this bootstrap_room. A
                    # redundant abort only re-wipes -- holds longer, never releases
                    # early -- so no transition guard is needed.
                    if (
                        receiver.kv_mgr.enable_deferred_decode_kv_release
                        and receiver.abort_notified
                    ):
                        receiver.kv_mgr.register_deferred_abort_room(
                            decode_req.req.bootstrap_room
                        )

            # Abort requests whose KV is already backed up for retraction.
            if self.disagg_decode_prealloc_queue.retracted_queue:
                remaining_retracted = []
                for decode_req in self.disagg_decode_prealloc_queue.retracted_queue:
                    if recv_req.abort_all or decode_req.rid.startswith(recv_req.rid):
                        retraction_discard(
                            decode_req,
                            self.tree_cache,
                            get_disagg().disaggregation_decode_retraction_backup,
                        )
                        self.ipc_channels.send_to_tokenizer.send_output(
                            _make_abort_req(decode_req), decode_req
                        )
                    else:
                        remaining_retracted.append(decode_req)
                self.disagg_decode_prealloc_queue.retracted_queue = remaining_retracted

        # Delete requests in the running batch
        for req in self.collect_inflight_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=}")
                if recv_req.abort_message:
                    # Timeout aborts carry an SLA message + 503 for the client.
                    req.to_finish = FINISH_ABORT(
                        recv_req.abort_message, HTTPStatus.SERVICE_UNAVAILABLE
                    )
                else:
                    req.to_finish = FINISH_ABORT()

    def _pause_engine(self) -> Tuple[List[Req], int]:
        raise NotImplementedError()

    def pause_generation(self, recv_req: PauseGenerationReqInput):
        assert recv_req.mode in ("in_place", "retract")
        self._engine_paused = True

        if recv_req.mode == "in_place":
            # In-place pause: just set the flag and return immediately.
            # All scheduler state (running_batch, last_batch, chunked_req,
            # result_queue) is left untouched. On resume, the normal event
            # loop (get_next_batch_to_run) handles last_batch merge,
            # chunked_req cleanup, and overlap result processing through
            # the standard code paths. This avoids duplicating batch
            # manipulation logic and the accounting bugs that come with it.
            return

        if self.enable_overlap and 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)

        retract_reqs = [r for r in self.running_batch.reqs if not r.finished()]
        if (
            self.last_batch is not None
            and self.last_batch.forward_mode.is_extend()
            # Skip merge for disagg prefill: completed prefill requests are
            # already in disagg_prefill_inflight_queue. Merging them into
            # running_batch leaks them, since the prefill event loop never
            # calls update_running_batch to clean them up.
            and self.disaggregation_mode != DisaggregationMode.PREFILL
        ):
            retract_reqs += [r for r in self.last_batch.reqs if not r.finished()]

        if (
            self.chunked_req is not None
            and not self.chunked_req.finished()
            and self.chunked_req not in retract_reqs
            and self.disaggregation_mode != DisaggregationMode.PREFILL
        ):
            retract_reqs.append(self.chunked_req)

        self.last_batch = None
        self.cur_batch_for_debug = None

        if retract_reqs:
            # Decode-side retract always rebootstraps (recomputes the KV from
            # the prefill), so skip the device->host KV offload that release_req
            # would otherwise do; the offloaded copy would be immediately
            # discarded. Non-decode modes ignore offload_kv (they never offload).
            retract_all(
                reqs=retract_reqs,
                req_to_token_pool=self.req_to_token_pool,
                token_to_kv_pool_allocator=self.token_to_kv_pool_allocator,
                tree_cache=self.tree_cache,
                hisparse_coordinator=self.hisparse_coordinator,
                offload_kv=False,
            )
        self.running_batch.reqs = []
        for req in retract_reqs:
            if self.disaggregation_mode == DisaggregationMode.DECODE:
                if req.output_ids:
                    req.pd_rebootstrap_forced_output_id = req.output_ids.pop()
                req.pd_rebootstrap_in_progress = True
                req.time_stats.set_retract_time()
                self.disagg_decode_prealloc_queue.hold_rebootstrap(req)
            else:
                self._add_request_to_queue(req)
        self.running_batch.batch_is_full = False
        # In disagg-PREFILL, keep a live mid-chunk chunked_req rather than retract it:
        # freeing its KV under a live disagg KV-sender crashes pop_bootstrapped or
        # sends freed/reused KV to decode. Kept, it resumes prefill after the pause.
        # TODO(disagg-prefill-retract): tear the sender down (abort + release metadata
        # buffer + reset pending_bootstrap) before freeing KV, then retract for real.
        # Until then a weight-update pause leaves stale-weight prefix KV (off-policy).
        if self.disaggregation_mode != DisaggregationMode.PREFILL:
            self.chunked_req = None

        # Surface the paused state to dashboards immediately. The scheduler
        # event loop short-circuits before reaching ``on_idle`` while paused,
        # so without this hop ``gen_throughput`` retains its last non-zero
        # value and KV events are not flushed for the entire pause window
        # (e.g. across a weight update). Zero the gauge, force a one-shot
        # idle log by resetting the rate-limit timestamp, and flush pending
        # KV events.
        self.metrics_reporter.last_gen_throughput = 0.0
        if self.metrics_reporter.current_scheduler_metrics_enabled:
            self.metrics_reporter.metrics_collector.last_log_time = 0.0
            self.metrics_reporter._maybe_log_idle_metrics()
        self.kv_events_publisher.publish_kv_events()

    def continue_generation(self, recv_req: ContinueGenerationReqInput):
        if recv_req.torch_empty_cache:
            before_mb = torch.cuda.memory_reserved() / (1024 * 1024)
            torch.cuda.empty_cache()
            after_mb = torch.cuda.memory_reserved() / (1024 * 1024)
            logger.info(
                f"[continue_generation] torch.cuda.empty_cache() called: "
                f"reserved {before_mb:.1f} MB -> {after_mb:.1f} MB "
                f"(freed {before_mb - after_mb:.1f} MB)"
            )
        # Enqueue any rebootstrap requests that were staged during a
        # retract-mode pause. Deferring until resume keeps the preallocation
        # queue empty during the pause window (so an intervening weight update
        # can flush the cache) and recomputes the prefix KV under the updated
        # weights.
        if (
            self.disaggregation_mode == DisaggregationMode.DECODE
            and self.disagg_decode_prealloc_queue is not None
        ):
            self.disagg_decode_prealloc_queue.enqueue_held_rebootstrap()
        self._engine_paused = False

    def handle_scale_elastic_ep(
        self, recv_req: ScaleElasticEPReqInput
    ) -> ScaleElasticEPReqOutput:
        """Begin a pending elastic EP scale-up request."""
        from sglang.srt.elastic_ep.elastic_ep import ElasticEPStateManager

        old_ep_size = ElasticEPStateManager.get_effective_ep_size()
        new_ep_size = recv_req.new_ep_size
        max_ep_size = get_parallel().max_world_size

        logger.debug(
            "[Elastic EP][scale] request received: new_ep_size=%d "
            "old_ep_size=%d max_ep_size=%d",
            new_ep_size,
            old_ep_size,
            max_ep_size,
        )

        if new_ep_size <= old_ep_size:
            return ScaleElasticEPReqOutput(
                success=False,
                message=(
                    f"new_ep_size ({new_ep_size}) must be greater than current "
                    f"effective_ep_size ({old_ep_size})."
                ),
                old_ep_size=old_ep_size,
                new_ep_size=new_ep_size,
            )
        if new_ep_size > max_ep_size:
            return ScaleElasticEPReqOutput(
                success=False,
                message=(
                    f"new_ep_size ({new_ep_size}) exceeds --max-ep-size "
                    f"({max_ep_size}). Restart with a larger --max-ep-size."
                ),
                old_ep_size=old_ep_size,
                new_ep_size=new_ep_size,
            )
        if ElasticEPStateManager.is_scaling():
            return ScaleElasticEPReqOutput(
                success=False,
                message=(
                    "A previous scale operation has not completed yet. Wait until "
                    "all pending ranks have joined before issuing another scale."
                ),
                old_ep_size=old_ep_size,
                new_ep_size=new_ep_size,
                pending_ep_size=ElasticEPStateManager.get_pending_ep_size(),
                scale_phase=ElasticEPStateManager.get_scale_phase(),
            )

        if not ElasticEPStateManager.request_scale(new_ep_size):
            return ScaleElasticEPReqOutput(
                success=False,
                message=(
                    "Failed to queue elastic EP scale: no elastic state or "
                    "scale already pending."
                ),
                old_ep_size=old_ep_size,
                new_ep_size=new_ep_size,
                pending_ep_size=ElasticEPStateManager.get_pending_ep_size(),
                scale_phase=ElasticEPStateManager.get_scale_phase(),
            )
        if (eplb_manager := self.tp_worker.model_runner.eplb_manager) is not None:
            eplb_manager.disable_rebalance("elastic EP scale-up is pending")
        logger.debug(
            "[Elastic EP][scale] scale requested: target_ep_size=%d; "
            "waiting for a joining cohort",
            new_ep_size,
        )

        return ScaleElasticEPReqOutput(
            success=True,
            message=f"Scaling initiated from {old_ep_size} to {new_ep_size}",
            old_ep_size=old_ep_size,
            new_ep_size=new_ep_size,
            pending_ep_size=ElasticEPStateManager.get_pending_ep_size(),
            scale_phase=ElasticEPStateManager.get_scale_phase(),
        )

    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 load_lora_adapter_from_tensors(
        self, recv_req: LoadLoRAAdapterFromTensorsReqInput
    ) -> LoadLoRAAdapterFromTensorsReqOutput:
        """In-place loading a new lora adapter from serialized tensors."""

        result = self.tp_worker.load_lora_adapter_from_tensors(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=success, message=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=success, message=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 handle_pd_role_switch(self, recv_req: PdRoleSwitchReqInput):
        return role_switch.handle_pd_role_switch(self, recv_req)

    def _sync_disaggregation_mode_to_subcomponents(self):
        # Push the (possibly flipped) mode into sub-components that cache it.
        # object.__setattr__ because some are frozen dataclasses.
        for name in (
            "invariant_checker",
            "load_inquirer",
            "output_streamer",
            "batch_result_processor",
        ):
            comp = getattr(self, name, None)
            if comp is not None and hasattr(comp, "disaggregation_mode"):
                object.__setattr__(
                    comp, "disaggregation_mode", self.disaggregation_mode
                )

    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):
        output = self.session_controller.open(recv_req)
        if output.success and self.enable_session_radix_cache:
            self.tree_cache.open_radix_session(recv_req.session_id)
        if (
            get_parallel().pp_rank == 0
            and get_parallel().tp_rank == 0
            and get_parallel().attn_cp_rank == 0
        ):
            return output
        return None

    def close_session(self, recv_req: CloseSessionReqInput):
        if self.enable_session_radix_cache:
            self.tree_cache.release_radix_session(recv_req.session_id)
        if (
            recv_req.session_id in self.session_controller
            or not self.enable_session_radix_cache
        ):
            self.session_controller.close(recv_req)

    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.ipc_channels.send_to_detokenizer.send_output(recv_req, recv_req)
        return None

    def handle_shutdown(self, recv_req: ShutdownReq):
        # Break the event loop; the finally in run_scheduler_process releases resources.
        self.gracefully_exit = True
        return None

    def configure_logging(self, recv_req: ConfigureLoggingReq):
        if recv_req.log_level is not None:
            logging.getLogger().setLevel(recv_req.log_level.upper())
        self.ipc_channels.send_to_detokenizer.send_output(recv_req, recv_req)

    def handle_dumper_control(self, recv_req: DumperControlReqInput):
        from sglang.srt.debug_utils.dumper import dumper

        try:
            response: list = []
            if (
                not torch.distributed.is_initialized()
                or torch.distributed.get_rank() == 0
            ):
                response = dumper._http_manager.handle_request(
                    method=recv_req.method, body=recv_req.body
                )
            self.ipc_channels.send_to_tokenizer.send_output(
                DumperControlReqOutput(success=True, response=response), recv_req
            )
        except Exception as e:
            print(f"[Scheduler] handle_dumper_control error: {e}", flush=True)
            self.ipc_channels.send_to_tokenizer.send_output(
                DumperControlReqOutput(success=False, response=[], error=str(e)),
                recv_req,
            )

    # placeholder for override
    def update_cache_from_scheduler(
        self, schedule_batch: ScheduleBatch, batch_result: GenerationBatchResult
    ):
        pass


def dispatch_event_loop(scheduler: Scheduler):
    if scheduler.server_args.enable_pd_role_switch:
        return role_switch.run_event_loop_supervisor(
            scheduler,
            _dispatch_event_loop_once,
        )
    return _dispatch_event_loop_once(scheduler)


def _dispatch_event_loop_once(scheduler: Scheduler):
    # The live PP property asserts before torch.distributed init (MLX stub).
    disaggregation_mode: DisaggregationMode = scheduler.disaggregation_mode
    if disaggregation_mode == DisaggregationMode.NULL:
        if scheduler.enable_pdmux:
            scheduler.event_loop_pdmux()
        elif get_parallel().pp_size > 1:
            scheduler.event_loop_pp()
        elif scheduler.enable_overlap_mlx:
            scheduler.event_loop_overlap_mlx()
        elif scheduler.enable_overlap:
            scheduler.event_loop_overlap()
        else:
            scheduler.event_loop_normal()
    elif disaggregation_mode == DisaggregationMode.PREFILL:
        if get_parallel().pp_size > 1:
            scheduler.event_loop_pp_disagg_prefill()
        elif scheduler.enable_overlap:
            scheduler.event_loop_overlap_disagg_prefill()
        else:
            scheduler.event_loop_normal_disagg_prefill()
    elif disaggregation_mode == DisaggregationMode.DECODE:
        if get_parallel().pp_size > 1:
            scheduler.event_loop_pp_disagg_decode()
        elif scheduler.enable_overlap:
            scheduler.event_loop_overlap_disagg_decode()
        else:
            scheduler.event_loop_normal_disagg_decode()


def resolve_spawn_dp_rank(dp_rank: Optional[int]) -> Optional[int]:
    """The `dp_rank` this process was spawned with, in either of its two forms.

    A router does not pass it as an argument, it sets `SGLANG_DP_RANK`. Both
    forms are the launcher naming this process's place, so both have to be in
    hand before `publish` records the placement -- resolving one of them after
    would leave the context answering `None` for a process that has a rank.
    """
    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
        return int(os.environ["SGLANG_DP_RANK"])
    return dp_rank


def configure_scheduler_process(
    server_args: ServerArgs,
    gpu_id: int,
    tp_rank: int,
    attn_cp_rank: int,
    moe_dp_rank: int,
    moe_ep_rank: int,
    pp_rank: int,
    dp_rank: Optional[int],
    display_tp_rank: Optional[int] = None,
    display_dp_rank: Optional[int] = None,
    display_moe_ep_rank: Optional[int] = None,
) -> None:
    """Configure scheduler worker logging and process title.

    display_* ranks are cosmetic; runtime ranks stay local. `dp_rank` arrives
    already resolved -- see `resolve_spawn_dp_rank`.
    """
    kill_itself_when_parent_died()

    # Generate the logger prefix
    shown_dp = display_dp_rank if display_dp_rank is not None else dp_rank
    shown_tp = display_tp_rank if display_tp_rank is not None else tp_rank
    shown_moe_ep = (
        display_moe_ep_rank if display_moe_ep_rank is not None else moe_ep_rank
    )

    prefix = ""
    if shown_dp is not None:
        prefix += f" DP{shown_dp}"
    if get_parallel().pp_size > 1:
        prefix += f" PP{pp_rank}"
    if get_parallel().attn_cp_size > 1:
        prefix += f" ATTN_CP{attn_cp_rank}"
    if get_parallel().moe_dp_size > 1:
        prefix += f" MOE_DP{moe_dp_rank}"
    if get_parallel().tp_size > 1:
        prefix += f" TP{shown_tp}"
    if get_parallel().ep_size > 1:
        prefix += f" EP{shown_moe_ep}"

    # Config the process
    setproctitle.setproctitle(f"sglang::scheduler{prefix.replace(' ', '_')}")
    faulthandler.enable()

    # Configure the logger
    configure_logger(server_args, prefix=prefix)
    suppress_other_loggers()

    # Set cpu affinity to this gpu process
    if envs.SGLANG_SET_CPU_AFFINITY.get():
        set_gpu_proc_affinity(
            get_parallel().pp_size,
            get_parallel().tp_size,
            get_parallel().nnodes,
            gpu_id,
        )
    if not envs.SGLANG_NUMA_BIND_V2.get():
        numa_node = get_numa_node_if_available(server_args, gpu_id)
        if numa_node is not None:
            numa_bind_to_node(numa_node)


def run_scheduler_process(
    server_args: ServerArgs,
    port_args: PortArgs,
    gpu_id: int,
    tp_rank: int,
    attn_cp_rank: int,
    moe_dp_rank: int,
    moe_ep_rank: int,
    pp_rank: int,
    dp_rank: Optional[int],
    pipe_writer,
    display_tp_rank: Optional[int] = None,
    display_dp_rank: Optional[int] = None,
    display_moe_ep_rank: Optional[int] = None,
):
    # Load plugins so hooks can override Scheduler and its dependencies.
    load_plugins()
    dp_rank = resolve_spawn_dp_rank(dp_rank)
    # Publish before anything in this process reads configuration, with the
    # placement the launcher decided: from here on a rank read is answered
    # without a process group, which is what every reader needs before
    # `init_torch_distributed` has run.
    publish(
        server_args,
        role="scheduler",
        ranks=SpawnRanks(
            world_rank=spawn_world_rank(server_args, tp_rank=tp_rank, pp_rank=pp_rank),
            dp_rank=dp_rank,
        ),
    )
    configure_scheduler_process(
        server_args,
        gpu_id,
        tp_rank,
        attn_cp_rank,
        moe_dp_rank,
        moe_ep_rank,
        pp_rank,
        dp_rank,
        display_tp_rank=display_tp_rank,
        display_dp_rank=display_dp_rank,
        display_moe_ep_rank=display_moe_ep_rank,
    )
    parent_process = psutil.Process().parent()

    # Set up tracing
    if get_observability().enable_trace:
        process_tracing_init(
            get_observability().otlp_traces_endpoint,
            get_observability().otlp_service_name,
            trace_modules=get_observability().trace_modules,
        )
        thread_label = "Scheduler"
        if get_disagg().disaggregation_mode == "prefill":
            thread_label = "Prefill Scheduler"
        elif get_disagg().disaggregation_mode == "decode":
            thread_label = "Decode Scheduler"
        trace_set_thread_info(thread_label, tp_rank, dp_rank, pp_rank)

    # Create a scheduler and run the event loop
    scheduler = None
    try:
        scheduler = Scheduler(
            server_args,
            port_args,
            gpu_id,
            tp_rank,
            moe_ep_rank,
            pp_rank,
            attn_cp_rank,
            moe_dp_rank,
            dp_rank,
        )

        # Send initialization info back to the parent process
        pipe_writer.send(scheduler.get_init_info())

        # Run the event loop (blocks until a ShutdownReq sets gracefully_exit)
        scheduler.run_event_loop()

    except Exception:
        traceback = get_exception_traceback()
        logger.error(f"Scheduler hit an exception: {traceback}")
        parent_process.send_signal(signal.SIGQUIT)
        # Opt-in: SIGKILL the pgroup so sibling ranks don't spew thousands
        # of NCCL/TCPStore tracebacks before they finally die.
        if envs.SGLANG_KILLPG_ON_SCHEDULER_EXCEPTION.get():
            try:
                os.killpg(os.getpgrp(), signal.SIGKILL)
            except Exception:
                pass
    finally:
        if scheduler is not None:
            # FPM has a background ZMQ publisher thread that needs explicit
            # teardown to flush queued metrics and close the socket cleanly.
            scheduler.metrics_reporter._shutdown_fpm()
            # Graceful path only: on the exception path the GPU may be wedged
            # and the synchronize() in destroy() could itself hang.
            if scheduler.gracefully_exit:
                scheduler.release_host_resources()
                # Last: anything above may still need a working communicator.
                abort_distributed_environment()


def _make_abort_req(
    req: Req, finished_reason: Optional[FinishReasonDict] = None
) -> AbortReq:
    return AbortReq(
        rid=req.rid,
        finished_reason=finished_reason,
        weight_versions=compute_weight_version_spans(
            req.weight_version_events,
            current_version=get_serving().weight_version,
            num_output_tokens=len(req.output_ids),
        ),
    )