vllm.v1.engine.core

HANDSHAKE_TIMEOUT_MINS module-attribute

HANDSHAKE_TIMEOUT_MINS = 5

POLLING_TIMEOUT_S module-attribute

POLLING_TIMEOUT_S = 2.5

_R module-attribute

_R = TypeVar('_R')

logger module-attribute

logger = init_logger(__name__)

DPEngineCoreActor

Bases: DPEngineCoreProc

Ray actor for running EngineCore in a data parallel context

Source code in vllm/v1/engine/core.py

class DPEngineCoreActor(DPEngineCoreProc):
    """
    Ray actor for running EngineCore in a data parallel context
    """

    def __init__(
        self,
        vllm_config: VllmConfig,
        local_client: bool,
        addresses: EngineZmqAddresses,
        executor_class: type[Executor],
        log_stats: bool,
        dp_rank: int = 0,
        local_dp_rank: int = 0,
    ):
        self.addresses = addresses
        vllm_config.parallel_config.data_parallel_rank = dp_rank
        vllm_config.parallel_config.data_parallel_rank_local = \
            local_dp_rank

        # Ray sets CUDA_VISIBLE_DEVICES to empty string,
        # we clean this up to be able to properly initialize
        # data parallel groups.
        del os.environ['CUDA_VISIBLE_DEVICES']

        super().__init__(vllm_config, local_client, "", executor_class,
                         log_stats)

    def _decorate_logs(self):
        pass

    @contextmanager
    def _perform_handshakes(self, handshake_address: str, identity: bytes,
                            local_client: bool, vllm_config: VllmConfig,
                            client_handshake_address: Optional[str]):
        """
        For Ray, we don't need to actually perform handshake.
        All addresses information is known before the actor creation.
        Therefore, we simply yield these addresses.
        """
        yield self.addresses

    def wait_for_init(self):
        """
        Wait until the engine core is initialized.

        This is just an empty method. When ray.get() on this method
        (or any other method of the actor) returns, it is guaranteed
        that actor creation (i.e., __init__) is complete.
        """
        pass

    def run(self):
        """
        Run the engine core busy loop.
        """
        try:
            self.run_busy_loop()
        except SystemExit:
            logger.debug("EngineCore exiting.")
            raise
        except Exception:
            logger.exception("EngineCore encountered a fatal error.")
            raise
        finally:
            self.shutdown()

addresses instance-attribute

addresses = addresses

__init__

__init__(
    vllm_config: VllmConfig,
    local_client: bool,
    addresses: EngineZmqAddresses,
    executor_class: type[Executor],
    log_stats: bool,
    dp_rank: int = 0,
    local_dp_rank: int = 0,
)

Source code in vllm/v1/engine/core.py

def __init__(
    self,
    vllm_config: VllmConfig,
    local_client: bool,
    addresses: EngineZmqAddresses,
    executor_class: type[Executor],
    log_stats: bool,
    dp_rank: int = 0,
    local_dp_rank: int = 0,
):
    self.addresses = addresses
    vllm_config.parallel_config.data_parallel_rank = dp_rank
    vllm_config.parallel_config.data_parallel_rank_local = \
        local_dp_rank

    # Ray sets CUDA_VISIBLE_DEVICES to empty string,
    # we clean this up to be able to properly initialize
    # data parallel groups.
    del os.environ['CUDA_VISIBLE_DEVICES']

    super().__init__(vllm_config, local_client, "", executor_class,
                     log_stats)

_decorate_logs

_decorate_logs()

Source code in vllm/v1/engine/core.py

def _decorate_logs(self):
    pass

_perform_handshakes

_perform_handshakes(
    handshake_address: str,
    identity: bytes,
    local_client: bool,
    vllm_config: VllmConfig,
    client_handshake_address: Optional[str],
)

For Ray, we don't need to actually perform handshake. All addresses information is known before the actor creation. Therefore, we simply yield these addresses.

Source code in vllm/v1/engine/core.py

@contextmanager
def _perform_handshakes(self, handshake_address: str, identity: bytes,
                        local_client: bool, vllm_config: VllmConfig,
                        client_handshake_address: Optional[str]):
    """
    For Ray, we don't need to actually perform handshake.
    All addresses information is known before the actor creation.
    Therefore, we simply yield these addresses.
    """
    yield self.addresses

run

run()

Run the engine core busy loop.

Source code in vllm/v1/engine/core.py

def run(self):
    """
    Run the engine core busy loop.
    """
    try:
        self.run_busy_loop()
    except SystemExit:
        logger.debug("EngineCore exiting.")
        raise
    except Exception:
        logger.exception("EngineCore encountered a fatal error.")
        raise
    finally:
        self.shutdown()

wait_for_init

wait_for_init()

Wait until the engine core is initialized.

This is just an empty method. When ray.get() on this method (or any other method of the actor) returns, it is guaranteed that actor creation (i.e., init) is complete.

Source code in vllm/v1/engine/core.py

def wait_for_init(self):
    """
    Wait until the engine core is initialized.

    This is just an empty method. When ray.get() on this method
    (or any other method of the actor) returns, it is guaranteed
    that actor creation (i.e., __init__) is complete.
    """
    pass

DPEngineCoreProc

Bases: EngineCoreProc

ZMQ-wrapper for running EngineCore in background process in a data parallel context.

Source code in vllm/v1/engine/core.py

class DPEngineCoreProc(EngineCoreProc):
    """ZMQ-wrapper for running EngineCore in background process
    in a data parallel context."""

    def __init__(
        self,
        vllm_config: VllmConfig,
        local_client: bool,
        handshake_address: str,
        executor_class: type[Executor],
        log_stats: bool,
        client_handshake_address: Optional[str] = None,
    ):
        self._decorate_logs()

        # Counts forward-passes of the model so that we can synchronize
        # finished with DP peers every N steps.
        self.counter = 0
        self.current_wave = 0
        self.last_counts = (0, 0)

        # Initialize the engine.
        dp_rank = vllm_config.parallel_config.data_parallel_rank
        super().__init__(vllm_config, local_client, handshake_address,
                         executor_class, log_stats, client_handshake_address,
                         dp_rank)

    def _decorate_logs(self):
        # Add process-specific prefix to stdout and stderr before
        # we initialize the engine.
        from multiprocessing import current_process
        process_name = current_process().name
        pid = os.getpid()
        _add_prefix(sys.stdout, process_name, pid)
        _add_prefix(sys.stderr, process_name, pid)

    def _init_data_parallel(self, vllm_config: VllmConfig):

        # Configure GPUs and stateless process group for data parallel.
        dp_rank = vllm_config.parallel_config.data_parallel_rank
        dp_size = vllm_config.parallel_config.data_parallel_size
        local_dp_rank = vllm_config.parallel_config.data_parallel_rank_local

        assert dp_size > 1
        assert 0 <= local_dp_rank <= dp_rank < dp_size

        if vllm_config.kv_transfer_config is not None:
            # modify the engine_id and append the local_dp_rank to it to ensure
            # that the kv_transfer_config is unique for each DP rank.
            vllm_config.kv_transfer_config.engine_id = (
                f"{vllm_config.kv_transfer_config.engine_id}_dp{local_dp_rank}"
            )
            logger.debug("Setting kv_transfer_config.engine_id to %s",
                         vllm_config.kv_transfer_config.engine_id)

        from vllm.platforms import current_platform
        device_control_env_var = current_platform.device_control_env_var
        world_size = vllm_config.parallel_config.world_size
        # Set CUDA_VISIBLE_DEVICES or equivalent.
        try:
            os.environ[device_control_env_var] = ",".join(
                str(current_platform.device_id_to_physical_device_id(i))
                for i in range(local_dp_rank *
                               world_size, (local_dp_rank + 1) * world_size))
        except IndexError as e:
            raise Exception(
                f"Error setting {device_control_env_var}: "
                f"local range: [{local_dp_rank * world_size}, "
                f"{(local_dp_rank + 1) * world_size}) "
                f"base value: \"{os.getenv(device_control_env_var)}\"") from e

        self.dp_rank = dp_rank
        self.dp_group = vllm_config.parallel_config.stateless_init_dp_group()

    def shutdown(self):
        super().shutdown()
        if dp_group := getattr(self, "dp_group", None):
            stateless_destroy_torch_distributed_process_group(dp_group)

    def add_request(self, request: EngineCoreRequest):
        if self.has_coordinator and request.current_wave != self.current_wave:
            if request.current_wave > self.current_wave:
                self.current_wave = request.current_wave
            elif not self.engines_running:
                # Request received for an already-completed wave, notify
                # front-end that we need to start the next one.
                self.output_queue.put_nowait(
                    (-1, EngineCoreOutputs(start_wave=self.current_wave)))

        super().add_request(request)

    def _handle_client_request(self, request_type: EngineCoreRequestType,
                               request: Any) -> None:
        if request_type == EngineCoreRequestType.START_DP_WAVE:
            new_wave, exclude_eng_index = request
            if exclude_eng_index != self.engine_index and (
                    new_wave >= self.current_wave):
                self.current_wave = new_wave
                if not self.engines_running:
                    logger.debug("EngineCore starting idle loop for wave %d.",
                                 new_wave)
                    self.engines_running = True
        else:
            super()._handle_client_request(request_type, request)

    def _maybe_publish_request_counts(self):
        if not self.publish_dp_lb_stats:
            return

        # Publish our request counts (if they've changed).
        counts = self.scheduler.get_request_counts()
        if counts != self.last_counts:
            self.last_counts = counts
            stats = SchedulerStats(*counts)
            self.output_queue.put_nowait(
                (-1, EngineCoreOutputs(scheduler_stats=stats)))

    def run_busy_loop(self):
        """Core busy loop of the EngineCore for data parallel case."""

        # Loop until process is sent a SIGINT or SIGTERM
        while True:
            # 1) Poll the input queue until there is work to do.
            self._process_input_queue()

            # 2) Step the engine core.
            executed = self._process_engine_step()
            self._maybe_publish_request_counts()

            local_unfinished_reqs = self.scheduler.has_unfinished_requests()
            if not executed:
                if not local_unfinished_reqs and not self.engines_running:
                    # All engines are idle.
                    continue

                # We are in a running state and so must execute a dummy pass
                # if the model didn't execute any ready requests.
                self.execute_dummy_batch()

            # 3) All-reduce operation to determine global unfinished reqs.
            self.engines_running = self._has_global_unfinished_reqs(
                local_unfinished_reqs)

            if not self.engines_running:
                if self.dp_rank == 0 or not self.has_coordinator:
                    # Notify client that we are pausing the loop.
                    logger.debug("Wave %d finished, pausing engine loop.",
                                 self.current_wave)
                    # In the coordinator case, dp rank 0 sends updates to the
                    # coordinator. Otherwise (offline spmd case), each rank
                    # sends the update to its colocated front-end process.
                    client_index = -1 if self.has_coordinator else 0
                    self.output_queue.put_nowait(
                        (client_index,
                         EngineCoreOutputs(wave_complete=self.current_wave)))
                self.current_wave += 1

    def _has_global_unfinished_reqs(self, local_unfinished: bool) -> bool:

        # Optimization - only perform finish-sync all-reduce every 32 steps.
        self.counter += 1
        if self.counter != 32:
            return True
        self.counter = 0

        return ParallelConfig.has_unfinished_dp(self.dp_group,
                                                local_unfinished)

counter instance-attribute

counter = 0

current_wave instance-attribute

current_wave = 0

last_counts instance-attribute

last_counts = (0, 0)

__init__

__init__(
    vllm_config: VllmConfig,
    local_client: bool,
    handshake_address: str,
    executor_class: type[Executor],
    log_stats: bool,
    client_handshake_address: Optional[str] = None,
)

Source code in vllm/v1/engine/core.py

def __init__(
    self,
    vllm_config: VllmConfig,
    local_client: bool,
    handshake_address: str,
    executor_class: type[Executor],
    log_stats: bool,
    client_handshake_address: Optional[str] = None,
):
    self._decorate_logs()

    # Counts forward-passes of the model so that we can synchronize
    # finished with DP peers every N steps.
    self.counter = 0
    self.current_wave = 0
    self.last_counts = (0, 0)

    # Initialize the engine.
    dp_rank = vllm_config.parallel_config.data_parallel_rank
    super().__init__(vllm_config, local_client, handshake_address,
                     executor_class, log_stats, client_handshake_address,
                     dp_rank)

_decorate_logs

_decorate_logs()

Source code in vllm/v1/engine/core.py

def _decorate_logs(self):
    # Add process-specific prefix to stdout and stderr before
    # we initialize the engine.
    from multiprocessing import current_process
    process_name = current_process().name
    pid = os.getpid()
    _add_prefix(sys.stdout, process_name, pid)
    _add_prefix(sys.stderr, process_name, pid)

_handle_client_request

_handle_client_request(
    request_type: EngineCoreRequestType, request: Any
) -> None

Source code in vllm/v1/engine/core.py

def _handle_client_request(self, request_type: EngineCoreRequestType,
                           request: Any) -> None:
    if request_type == EngineCoreRequestType.START_DP_WAVE:
        new_wave, exclude_eng_index = request
        if exclude_eng_index != self.engine_index and (
                new_wave >= self.current_wave):
            self.current_wave = new_wave
            if not self.engines_running:
                logger.debug("EngineCore starting idle loop for wave %d.",
                             new_wave)
                self.engines_running = True
    else:
        super()._handle_client_request(request_type, request)

_has_global_unfinished_reqs

_has_global_unfinished_reqs(local_unfinished: bool) -> bool

Source code in vllm/v1/engine/core.py

def _has_global_unfinished_reqs(self, local_unfinished: bool) -> bool:

    # Optimization - only perform finish-sync all-reduce every 32 steps.
    self.counter += 1
    if self.counter != 32:
        return True
    self.counter = 0

    return ParallelConfig.has_unfinished_dp(self.dp_group,
                                            local_unfinished)

_init_data_parallel

_init_data_parallel(vllm_config: VllmConfig)

Source code in vllm/v1/engine/core.py

def _init_data_parallel(self, vllm_config: VllmConfig):

    # Configure GPUs and stateless process group for data parallel.
    dp_rank = vllm_config.parallel_config.data_parallel_rank
    dp_size = vllm_config.parallel_config.data_parallel_size
    local_dp_rank = vllm_config.parallel_config.data_parallel_rank_local

    assert dp_size > 1
    assert 0 <= local_dp_rank <= dp_rank < dp_size

    if vllm_config.kv_transfer_config is not None:
        # modify the engine_id and append the local_dp_rank to it to ensure
        # that the kv_transfer_config is unique for each DP rank.
        vllm_config.kv_transfer_config.engine_id = (
            f"{vllm_config.kv_transfer_config.engine_id}_dp{local_dp_rank}"
        )
        logger.debug("Setting kv_transfer_config.engine_id to %s",
                     vllm_config.kv_transfer_config.engine_id)

    from vllm.platforms import current_platform
    device_control_env_var = current_platform.device_control_env_var
    world_size = vllm_config.parallel_config.world_size
    # Set CUDA_VISIBLE_DEVICES or equivalent.
    try:
        os.environ[device_control_env_var] = ",".join(
            str(current_platform.device_id_to_physical_device_id(i))
            for i in range(local_dp_rank *
                           world_size, (local_dp_rank + 1) * world_size))
    except IndexError as e:
        raise Exception(
            f"Error setting {device_control_env_var}: "
            f"local range: [{local_dp_rank * world_size}, "
            f"{(local_dp_rank + 1) * world_size}) "
            f"base value: \"{os.getenv(device_control_env_var)}\"") from e

    self.dp_rank = dp_rank
    self.dp_group = vllm_config.parallel_config.stateless_init_dp_group()

_maybe_publish_request_counts

_maybe_publish_request_counts()

Source code in vllm/v1/engine/core.py

def _maybe_publish_request_counts(self):
    if not self.publish_dp_lb_stats:
        return

    # Publish our request counts (if they've changed).
    counts = self.scheduler.get_request_counts()
    if counts != self.last_counts:
        self.last_counts = counts
        stats = SchedulerStats(*counts)
        self.output_queue.put_nowait(
            (-1, EngineCoreOutputs(scheduler_stats=stats)))

add_request

add_request(request: EngineCoreRequest)

Source code in vllm/v1/engine/core.py

def add_request(self, request: EngineCoreRequest):
    if self.has_coordinator and request.current_wave != self.current_wave:
        if request.current_wave > self.current_wave:
            self.current_wave = request.current_wave
        elif not self.engines_running:
            # Request received for an already-completed wave, notify
            # front-end that we need to start the next one.
            self.output_queue.put_nowait(
                (-1, EngineCoreOutputs(start_wave=self.current_wave)))

    super().add_request(request)

run_busy_loop

run_busy_loop()

Core busy loop of the EngineCore for data parallel case.

Source code in vllm/v1/engine/core.py

def run_busy_loop(self):
    """Core busy loop of the EngineCore for data parallel case."""

    # Loop until process is sent a SIGINT or SIGTERM
    while True:
        # 1) Poll the input queue until there is work to do.
        self._process_input_queue()

        # 2) Step the engine core.
        executed = self._process_engine_step()
        self._maybe_publish_request_counts()

        local_unfinished_reqs = self.scheduler.has_unfinished_requests()
        if not executed:
            if not local_unfinished_reqs and not self.engines_running:
                # All engines are idle.
                continue

            # We are in a running state and so must execute a dummy pass
            # if the model didn't execute any ready requests.
            self.execute_dummy_batch()

        # 3) All-reduce operation to determine global unfinished reqs.
        self.engines_running = self._has_global_unfinished_reqs(
            local_unfinished_reqs)

        if not self.engines_running:
            if self.dp_rank == 0 or not self.has_coordinator:
                # Notify client that we are pausing the loop.
                logger.debug("Wave %d finished, pausing engine loop.",
                             self.current_wave)
                # In the coordinator case, dp rank 0 sends updates to the
                # coordinator. Otherwise (offline spmd case), each rank
                # sends the update to its colocated front-end process.
                client_index = -1 if self.has_coordinator else 0
                self.output_queue.put_nowait(
                    (client_index,
                     EngineCoreOutputs(wave_complete=self.current_wave)))
            self.current_wave += 1

shutdown

shutdown()

Source code in vllm/v1/engine/core.py

def shutdown(self):
    super().shutdown()
    if dp_group := getattr(self, "dp_group", None):
        stateless_destroy_torch_distributed_process_group(dp_group)

EngineCore

Inner loop of vLLM's Engine.

Source code in vllm/v1/engine/core.py

class EngineCore:
    """Inner loop of vLLM's Engine."""

    def __init__(self,
                 vllm_config: VllmConfig,
                 executor_class: type[Executor],
                 log_stats: bool,
                 executor_fail_callback: Optional[Callable] = None):

        # plugins need to be loaded at the engine/scheduler level too
        from vllm.plugins import load_general_plugins
        load_general_plugins()

        self.vllm_config = vllm_config
        logger.info("Initializing a V1 LLM engine (v%s) with config: %s",
                    VLLM_VERSION, vllm_config)

        self.log_stats = log_stats

        # Setup Model.
        self.model_executor = executor_class(vllm_config)
        if executor_fail_callback is not None:
            self.model_executor.register_failure_callback(
                executor_fail_callback)

        # Setup KV Caches and update CacheConfig after profiling.
        num_gpu_blocks, num_cpu_blocks, kv_cache_config = \
            self._initialize_kv_caches(vllm_config)

        vllm_config.cache_config.num_gpu_blocks = num_gpu_blocks
        vllm_config.cache_config.num_cpu_blocks = num_cpu_blocks
        self.collective_rpc("initialize_cache",
                            args=(num_gpu_blocks, num_cpu_blocks))

        self.structured_output_manager = StructuredOutputManager(vllm_config)

        # Setup scheduler.
        if isinstance(vllm_config.scheduler_config.scheduler_cls, str):
            Scheduler = resolve_obj_by_qualname(
                vllm_config.scheduler_config.scheduler_cls)
        else:
            Scheduler = vllm_config.scheduler_config.scheduler_cls

        # This warning can be removed once the V1 Scheduler interface is
        # finalized and we can maintain support for scheduler classes that
        # implement it
        if Scheduler is not V1Scheduler:
            logger.warning(
                "Using configured V1 scheduler class %s. "
                "This scheduler interface is not public and "
                "compatibility may not be maintained.",
                vllm_config.scheduler_config.scheduler_cls)

        self.scheduler: SchedulerInterface = Scheduler(
            vllm_config=vllm_config,
            kv_cache_config=kv_cache_config,
            structured_output_manager=self.structured_output_manager,
            include_finished_set=vllm_config.parallel_config.data_parallel_size
            > 1,
            log_stats=self.log_stats,
        )

        # Setup MM Input Mapper.
        self.mm_input_cache_server = MirroredProcessingCache(
            vllm_config.model_config)

        # Setup batch queue for pipeline parallelism.
        # Batch queue for scheduled batches. This enables us to asynchronously
        # schedule and execute batches, and is required by pipeline parallelism
        # to eliminate pipeline bubbles.
        self.batch_queue_size = self.model_executor.max_concurrent_batches
        self.batch_queue: Optional[queue.Queue[tuple[Future[ModelRunnerOutput],
                                                     SchedulerOutput]]] = None
        if self.batch_queue_size > 1:
            logger.info("Batch queue is enabled with size %d",
                        self.batch_queue_size)
            self.batch_queue = queue.Queue(self.batch_queue_size)

    def _initialize_kv_caches(
            self, vllm_config: VllmConfig) -> tuple[int, int, KVCacheConfig]:
        start = time.time()

        # Get all kv cache needed by the model
        kv_cache_specs = self.model_executor.get_kv_cache_specs()

        # Profiles the peak memory usage of the model to determine how much
        # memory can be allocated for kv cache.
        available_gpu_memory = self.model_executor.determine_available_memory()

        assert len(kv_cache_specs) == len(available_gpu_memory)
        # Get the kv cache tensor size
        kv_cache_configs = [
            get_kv_cache_config(vllm_config, kv_cache_spec_one_worker,
                                available_gpu_memory_one_worker)
            for kv_cache_spec_one_worker, available_gpu_memory_one_worker in
            zip(kv_cache_specs, available_gpu_memory)
        ]

        # Since we use a shared centralized controller, we need the
        # `kv_cache_config` to be consistent across all workers to make sure
        # all the memory operators can be applied to all workers.
        unify_kv_cache_configs(kv_cache_configs)

        # All workers have the same kv_cache_config except layer names, so use
        # an arbitrary one to initialize the scheduler.
        assert all([
            cfg.num_blocks == kv_cache_configs[0].num_blocks
            for cfg in kv_cache_configs
        ])
        num_gpu_blocks = kv_cache_configs[0].num_blocks
        num_cpu_blocks = 0
        scheduler_kv_cache_config = kv_cache_configs[0]

        # Initialize kv cache and warmup the execution
        self.model_executor.initialize_from_config(kv_cache_configs)

        elapsed = time.time() - start
        logger.info(("init engine (profile, create kv cache, "
                     "warmup model) took %.2f seconds"), elapsed)
        return num_gpu_blocks, num_cpu_blocks, scheduler_kv_cache_config

    def add_request(self, request: EngineCoreRequest):
        """Add request to the scheduler."""

        if request.mm_hashes is not None:
            # Here, if hash exists for a multimodal input, then it will be
            # fetched from the cache, else it will be added to the cache.
            # Note that the cache here is mirrored with the client cache, so
            # anything that has a hash must have a HIT cache entry here
            # as well.
            assert request.mm_inputs is not None
            request.mm_inputs = self.mm_input_cache_server.get_and_update_p1(
                request.mm_inputs, request.mm_hashes)

        req = Request.from_engine_core_request(request)
        if req.use_structured_output:
            # Start grammar compilation asynchronously
            self.structured_output_manager.grammar_init(req)

        if req.kv_transfer_params is not None and (
                not self.scheduler.get_kv_connector()):
            logger.warning("Got kv_transfer_params, but no KVConnector found. "
                           "Disabling KVTransfer for this request.")

        self.scheduler.add_request(req)

    def abort_requests(self, request_ids: list[str]):
        """Abort requests from the scheduler."""

        # TODO: The scheduler doesn't really need to know the
        # specific finish reason, TBD whether we propagate that
        # (i.e. client-aborted vs stop criteria met).
        self.scheduler.finish_requests(request_ids,
                                       RequestStatus.FINISHED_ABORTED)

    def execute_model(self, scheduler_output: SchedulerOutput):
        try:
            return self.model_executor.execute_model(scheduler_output)
        except Exception as err:
            # We do not want to catch BaseException here since we're only
            # interested in dumping info when the exception is due to an
            # error from execute_model itself.

            # NOTE: This method is exception-free
            dump_engine_exception(self.vllm_config, scheduler_output,
                                  self.scheduler.make_stats())
            raise err

    def step(self) -> tuple[dict[int, EngineCoreOutputs], bool]:
        """Schedule, execute, and make output.

        Returns tuple of outputs and a flag indicating whether the model
        was executed.
        """

        # Check for any requests remaining in the scheduler - unfinished,
        # or finished and not yet removed from the batch.
        if not self.scheduler.has_requests():
            return {}, False
        scheduler_output = self.scheduler.schedule()
        model_output = self.execute_model(scheduler_output)
        engine_core_outputs = self.scheduler.update_from_output(
            scheduler_output, model_output)  # type: ignore

        return (engine_core_outputs,
                scheduler_output.total_num_scheduled_tokens > 0)

    def step_with_batch_queue(
            self) -> tuple[Optional[dict[int, EngineCoreOutputs]], bool]:
        """Schedule and execute batches with the batch queue.
        Note that if nothing to output in this step, None is returned.

        The execution flow is as follows:
        1. Try to schedule a new batch if the batch queue is not full.
        If a new batch is scheduled, directly return an empty engine core
        output. In other words, fulfilling the batch queue has a higher priority
        than getting model outputs.
        2. If there is no new scheduled batch, meaning that the batch queue
        is full or no other requests can be scheduled, we block until the first
        batch in the job queue is finished.
        3. Update the scheduler from the output.
        """
        assert self.batch_queue is not None

        engine_core_outputs = None
        scheduler_output = None
        # Try to schedule a new batch if the batch queue is not full, but
        # the scheduler may return an empty batch if all requests are scheduled.
        # Note that this is not blocking.
        if not self.batch_queue.full():
            scheduler_output = self.scheduler.schedule()
            if scheduler_output.total_num_scheduled_tokens > 0:
                future = self.model_executor.execute_model(scheduler_output)
                self.batch_queue.put_nowait(
                    (future, scheduler_output))  # type: ignore

        scheduled_batch = (scheduler_output is not None
                           and scheduler_output.total_num_scheduled_tokens > 0)

        # If no more requests can be scheduled and the job queue is not empty,
        # block until the first batch in the job queue is finished.
        # TODO(comaniac): Ideally we should peek the first batch in the
        # job queue to check if it's finished before scheduling a new batch,
        # but peeking the first element in a queue is not thread-safe,
        # so we need more work.
        if not scheduled_batch and not self.batch_queue.empty():
            future, scheduler_output = self.batch_queue.get_nowait()
            # Blocking until the first result is available.
            model_output = future.result()
            self.batch_queue.task_done()
            engine_core_outputs = (self.scheduler.update_from_output(
                scheduler_output, model_output))

        return engine_core_outputs, scheduled_batch

    def shutdown(self):
        self.structured_output_manager.clear_backend()
        if self.model_executor:
            self.model_executor.shutdown()
        if self.scheduler:
            self.scheduler.shutdown()

    def profile(self, is_start: bool = True):
        self.model_executor.profile(is_start)

    def reset_mm_cache(self):
        # NOTE: Since this is mainly for debugging, we don't attempt to
        # re-sync the internal caches (P0 processor, P0 mirror, P1 mirror)
        if self.scheduler.has_unfinished_requests():
            logger.warning("Resetting the multi-modal cache when requests are "
                           "in progress may lead to desynced internal caches.")

        self.mm_input_cache_server.reset()

    def reset_prefix_cache(self):
        self.scheduler.reset_prefix_cache()

    def sleep(self, level: int = 1):
        self.model_executor.sleep(level)

    def wake_up(self, tags: Optional[list[str]] = None):
        self.model_executor.wake_up(tags)

    def is_sleeping(self) -> bool:
        return self.model_executor.is_sleeping

    def execute_dummy_batch(self):
        self.model_executor.collective_rpc("execute_dummy_batch")

    def add_lora(self, lora_request: LoRARequest) -> bool:
        return self.model_executor.add_lora(lora_request)

    def remove_lora(self, lora_id: int) -> bool:
        return self.model_executor.remove_lora(lora_id)

    def list_loras(self) -> set[int]:
        return self.model_executor.list_loras()

    def pin_lora(self, lora_id: int) -> bool:
        return self.model_executor.pin_lora(lora_id)

    def save_sharded_state(
        self,
        path: str,
        pattern: Optional[str] = None,
        max_size: Optional[int] = None,
    ) -> None:
        self.model_executor.save_sharded_state(path=path,
                                               pattern=pattern,
                                               max_size=max_size)

    def collective_rpc(self,
                       method: Union[str, Callable[..., _R]],
                       timeout: Optional[float] = None,
                       args: tuple = (),
                       kwargs: Optional[dict[str, Any]] = None) -> list[_R]:
        return self.model_executor.collective_rpc(method, timeout, args,
                                                  kwargs)

    def save_tensorized_model(
        self,
        tensorizer_config,
    ) -> None:
        self.model_executor.save_tensorized_model(
            tensorizer_config=tensorizer_config, )

batch_queue instance-attribute

batch_queue: Optional[
    Queue[tuple[Future[ModelRunnerOutput], SchedulerOutput]]
] = None

batch_queue_size instance-attribute

batch_queue_size = max_concurrent_batches

log_stats instance-attribute

log_stats = log_stats

mm_input_cache_server instance-attribute

mm_input_cache_server = MirroredProcessingCache(
    model_config
)

model_executor instance-attribute

model_executor = executor_class(vllm_config)

scheduler instance-attribute

scheduler: SchedulerInterface = Scheduler(
    vllm_config=vllm_config,
    kv_cache_config=kv_cache_config,
    structured_output_manager=structured_output_manager,
    include_finished_set=data_parallel_size > 1,
    log_stats=log_stats,
)

structured_output_manager instance-attribute

structured_output_manager = StructuredOutputManager(
    vllm_config
)

vllm_config instance-attribute

vllm_config = vllm_config

__init__

__init__(
    vllm_config: VllmConfig,
    executor_class: type[Executor],
    log_stats: bool,
    executor_fail_callback: Optional[Callable] = None,
)

Source code in vllm/v1/engine/core.py

def __init__(self,
             vllm_config: VllmConfig,
             executor_class: type[Executor],
             log_stats: bool,
             executor_fail_callback: Optional[Callable] = None):

    # plugins need to be loaded at the engine/scheduler level too
    from vllm.plugins import load_general_plugins
    load_general_plugins()

    self.vllm_config = vllm_config
    logger.info("Initializing a V1 LLM engine (v%s) with config: %s",
                VLLM_VERSION, vllm_config)

    self.log_stats = log_stats

    # Setup Model.
    self.model_executor = executor_class(vllm_config)
    if executor_fail_callback is not None:
        self.model_executor.register_failure_callback(
            executor_fail_callback)

    # Setup KV Caches and update CacheConfig after profiling.
    num_gpu_blocks, num_cpu_blocks, kv_cache_config = \
        self._initialize_kv_caches(vllm_config)

    vllm_config.cache_config.num_gpu_blocks = num_gpu_blocks
    vllm_config.cache_config.num_cpu_blocks = num_cpu_blocks
    self.collective_rpc("initialize_cache",
                        args=(num_gpu_blocks, num_cpu_blocks))

    self.structured_output_manager = StructuredOutputManager(vllm_config)

    # Setup scheduler.
    if isinstance(vllm_config.scheduler_config.scheduler_cls, str):
        Scheduler = resolve_obj_by_qualname(
            vllm_config.scheduler_config.scheduler_cls)
    else:
        Scheduler = vllm_config.scheduler_config.scheduler_cls

    # This warning can be removed once the V1 Scheduler interface is
    # finalized and we can maintain support for scheduler classes that
    # implement it
    if Scheduler is not V1Scheduler:
        logger.warning(
            "Using configured V1 scheduler class %s. "
            "This scheduler interface is not public and "
            "compatibility may not be maintained.",
            vllm_config.scheduler_config.scheduler_cls)

    self.scheduler: SchedulerInterface = Scheduler(
        vllm_config=vllm_config,
        kv_cache_config=kv_cache_config,
        structured_output_manager=self.structured_output_manager,
        include_finished_set=vllm_config.parallel_config.data_parallel_size
        > 1,
        log_stats=self.log_stats,
    )

    # Setup MM Input Mapper.
    self.mm_input_cache_server = MirroredProcessingCache(
        vllm_config.model_config)

    # Setup batch queue for pipeline parallelism.
    # Batch queue for scheduled batches. This enables us to asynchronously
    # schedule and execute batches, and is required by pipeline parallelism
    # to eliminate pipeline bubbles.
    self.batch_queue_size = self.model_executor.max_concurrent_batches
    self.batch_queue: Optional[queue.Queue[tuple[Future[ModelRunnerOutput],
                                                 SchedulerOutput]]] = None
    if self.batch_queue_size > 1:
        logger.info("Batch queue is enabled with size %d",
                    self.batch_queue_size)
        self.batch_queue = queue.Queue(self.batch_queue_size)

_initialize_kv_caches

_initialize_kv_caches(
    vllm_config: VllmConfig,
) -> tuple[int, int, KVCacheConfig]

Source code in vllm/v1/engine/core.py

def _initialize_kv_caches(
        self, vllm_config: VllmConfig) -> tuple[int, int, KVCacheConfig]:
    start = time.time()

    # Get all kv cache needed by the model
    kv_cache_specs = self.model_executor.get_kv_cache_specs()

    # Profiles the peak memory usage of the model to determine how much
    # memory can be allocated for kv cache.
    available_gpu_memory = self.model_executor.determine_available_memory()

    assert len(kv_cache_specs) == len(available_gpu_memory)
    # Get the kv cache tensor size
    kv_cache_configs = [
        get_kv_cache_config(vllm_config, kv_cache_spec_one_worker,
                            available_gpu_memory_one_worker)
        for kv_cache_spec_one_worker, available_gpu_memory_one_worker in
        zip(kv_cache_specs, available_gpu_memory)
    ]

    # Since we use a shared centralized controller, we need the
    # `kv_cache_config` to be consistent across all workers to make sure
    # all the memory operators can be applied to all workers.
    unify_kv_cache_configs(kv_cache_configs)

    # All workers have the same kv_cache_config except layer names, so use
    # an arbitrary one to initialize the scheduler.
    assert all([
        cfg.num_blocks == kv_cache_configs[0].num_blocks
        for cfg in kv_cache_configs
    ])
    num_gpu_blocks = kv_cache_configs[0].num_blocks
    num_cpu_blocks = 0
    scheduler_kv_cache_config = kv_cache_configs[0]

    # Initialize kv cache and warmup the execution
    self.model_executor.initialize_from_config(kv_cache_configs)

    elapsed = time.time() - start
    logger.info(("init engine (profile, create kv cache, "
                 "warmup model) took %.2f seconds"), elapsed)
    return num_gpu_blocks, num_cpu_blocks, scheduler_kv_cache_config

abort_requests

abort_requests(request_ids: list[str])

Abort requests from the scheduler.

Source code in vllm/v1/engine/core.py

def abort_requests(self, request_ids: list[str]):
    """Abort requests from the scheduler."""

    # TODO: The scheduler doesn't really need to know the
    # specific finish reason, TBD whether we propagate that
    # (i.e. client-aborted vs stop criteria met).
    self.scheduler.finish_requests(request_ids,
                                   RequestStatus.FINISHED_ABORTED)

add_lora

add_lora(lora_request: LoRARequest) -> bool

Source code in vllm/v1/engine/core.py

def add_lora(self, lora_request: LoRARequest) -> bool:
    return self.model_executor.add_lora(lora_request)

add_request

add_request(request: EngineCoreRequest)

Add request to the scheduler.

Source code in vllm/v1/engine/core.py

def add_request(self, request: EngineCoreRequest):
    """Add request to the scheduler."""

    if request.mm_hashes is not None:
        # Here, if hash exists for a multimodal input, then it will be
        # fetched from the cache, else it will be added to the cache.
        # Note that the cache here is mirrored with the client cache, so
        # anything that has a hash must have a HIT cache entry here
        # as well.
        assert request.mm_inputs is not None
        request.mm_inputs = self.mm_input_cache_server.get_and_update_p1(
            request.mm_inputs, request.mm_hashes)

    req = Request.from_engine_core_request(request)
    if req.use_structured_output:
        # Start grammar compilation asynchronously
        self.structured_output_manager.grammar_init(req)

    if req.kv_transfer_params is not None and (
            not self.scheduler.get_kv_connector()):
        logger.warning("Got kv_transfer_params, but no KVConnector found. "
                       "Disabling KVTransfer for this request.")

    self.scheduler.add_request(req)

collective_rpc

collective_rpc(
    method: Union[str, Callable[..., _R]],
    timeout: Optional[float] = None,
    args: tuple = (),
    kwargs: Optional[dict[str, Any]] = None,
) -> list[_R]

Source code in vllm/v1/engine/core.py

def collective_rpc(self,
                   method: Union[str, Callable[..., _R]],
                   timeout: Optional[float] = None,
                   args: tuple = (),
                   kwargs: Optional[dict[str, Any]] = None) -> list[_R]:
    return self.model_executor.collective_rpc(method, timeout, args,
                                              kwargs)

execute_dummy_batch

execute_dummy_batch()

Source code in vllm/v1/engine/core.py

def execute_dummy_batch(self):
    self.model_executor.collective_rpc("execute_dummy_batch")

execute_model

execute_model(scheduler_output: SchedulerOutput)

Source code in vllm/v1/engine/core.py

def execute_model(self, scheduler_output: SchedulerOutput):
    try:
        return self.model_executor.execute_model(scheduler_output)
    except Exception as err:
        # We do not want to catch BaseException here since we're only
        # interested in dumping info when the exception is due to an
        # error from execute_model itself.

        # NOTE: This method is exception-free
        dump_engine_exception(self.vllm_config, scheduler_output,
                              self.scheduler.make_stats())
        raise err

is_sleeping

is_sleeping() -> bool

Source code in vllm/v1/engine/core.py

def is_sleeping(self) -> bool:
    return self.model_executor.is_sleeping

list_loras

list_loras() -> set[int]

Source code in vllm/v1/engine/core.py

def list_loras(self) -> set[int]:
    return self.model_executor.list_loras()

pin_lora

pin_lora(lora_id: int) -> bool

Source code in vllm/v1/engine/core.py

def pin_lora(self, lora_id: int) -> bool:
    return self.model_executor.pin_lora(lora_id)

profile

profile(is_start: bool = True)

Source code in vllm/v1/engine/core.py

def profile(self, is_start: bool = True):
    self.model_executor.profile(is_start)

remove_lora

remove_lora(lora_id: int) -> bool

Source code in vllm/v1/engine/core.py

def remove_lora(self, lora_id: int) -> bool:
    return self.model_executor.remove_lora(lora_id)

reset_mm_cache

reset_mm_cache()

Source code in vllm/v1/engine/core.py

def reset_mm_cache(self):
    # NOTE: Since this is mainly for debugging, we don't attempt to
    # re-sync the internal caches (P0 processor, P0 mirror, P1 mirror)
    if self.scheduler.has_unfinished_requests():
        logger.warning("Resetting the multi-modal cache when requests are "
                       "in progress may lead to desynced internal caches.")

    self.mm_input_cache_server.reset()

reset_prefix_cache

reset_prefix_cache()

Source code in vllm/v1/engine/core.py

def reset_prefix_cache(self):
    self.scheduler.reset_prefix_cache()

save_sharded_state

save_sharded_state(
    path: str,
    pattern: Optional[str] = None,
    max_size: Optional[int] = None,
) -> None

Source code in vllm/v1/engine/core.py

def save_sharded_state(
    self,
    path: str,
    pattern: Optional[str] = None,
    max_size: Optional[int] = None,
) -> None:
    self.model_executor.save_sharded_state(path=path,
                                           pattern=pattern,
                                           max_size=max_size)

save_tensorized_model

save_tensorized_model(tensorizer_config) -> None

Source code in vllm/v1/engine/core.py

def save_tensorized_model(
    self,
    tensorizer_config,
) -> None:
    self.model_executor.save_tensorized_model(
        tensorizer_config=tensorizer_config, )

shutdown

shutdown()

Source code in vllm/v1/engine/core.py

def shutdown(self):
    self.structured_output_manager.clear_backend()
    if self.model_executor:
        self.model_executor.shutdown()
    if self.scheduler:
        self.scheduler.shutdown()

sleep

sleep(level: int = 1)

Source code in vllm/v1/engine/core.py

def sleep(self, level: int = 1):
    self.model_executor.sleep(level)

step

step() -> tuple[dict[int, EngineCoreOutputs], bool]

Schedule, execute, and make output.

Returns tuple of outputs and a flag indicating whether the model was executed.

Source code in vllm/v1/engine/core.py

def step(self) -> tuple[dict[int, EngineCoreOutputs], bool]:
    """Schedule, execute, and make output.

    Returns tuple of outputs and a flag indicating whether the model
    was executed.
    """

    # Check for any requests remaining in the scheduler - unfinished,
    # or finished and not yet removed from the batch.
    if not self.scheduler.has_requests():
        return {}, False
    scheduler_output = self.scheduler.schedule()
    model_output = self.execute_model(scheduler_output)
    engine_core_outputs = self.scheduler.update_from_output(
        scheduler_output, model_output)  # type: ignore

    return (engine_core_outputs,
            scheduler_output.total_num_scheduled_tokens > 0)

step_with_batch_queue

step_with_batch_queue() -> tuple[
    Optional[dict[int, EngineCoreOutputs]], bool
]

Schedule and execute batches with the batch queue. Note that if nothing to output in this step, None is returned.

The execution flow is as follows: 1. Try to schedule a new batch if the batch queue is not full. If a new batch is scheduled, directly return an empty engine core output. In other words, fulfilling the batch queue has a higher priority than getting model outputs. 2. If there is no new scheduled batch, meaning that the batch queue is full or no other requests can be scheduled, we block until the first batch in the job queue is finished. 3. Update the scheduler from the output.

Source code in vllm/v1/engine/core.py

def step_with_batch_queue(
        self) -> tuple[Optional[dict[int, EngineCoreOutputs]], bool]:
    """Schedule and execute batches with the batch queue.
    Note that if nothing to output in this step, None is returned.

    The execution flow is as follows:
    1. Try to schedule a new batch if the batch queue is not full.
    If a new batch is scheduled, directly return an empty engine core
    output. In other words, fulfilling the batch queue has a higher priority
    than getting model outputs.
    2. If there is no new scheduled batch, meaning that the batch queue
    is full or no other requests can be scheduled, we block until the first
    batch in the job queue is finished.
    3. Update the scheduler from the output.
    """
    assert self.batch_queue is not None

    engine_core_outputs = None
    scheduler_output = None
    # Try to schedule a new batch if the batch queue is not full, but
    # the scheduler may return an empty batch if all requests are scheduled.
    # Note that this is not blocking.
    if not self.batch_queue.full():
        scheduler_output = self.scheduler.schedule()
        if scheduler_output.total_num_scheduled_tokens > 0:
            future = self.model_executor.execute_model(scheduler_output)
            self.batch_queue.put_nowait(
                (future, scheduler_output))  # type: ignore

    scheduled_batch = (scheduler_output is not None
                       and scheduler_output.total_num_scheduled_tokens > 0)

    # If no more requests can be scheduled and the job queue is not empty,
    # block until the first batch in the job queue is finished.
    # TODO(comaniac): Ideally we should peek the first batch in the
    # job queue to check if it's finished before scheduling a new batch,
    # but peeking the first element in a queue is not thread-safe,
    # so we need more work.
    if not scheduled_batch and not self.batch_queue.empty():
        future, scheduler_output = self.batch_queue.get_nowait()
        # Blocking until the first result is available.
        model_output = future.result()
        self.batch_queue.task_done()
        engine_core_outputs = (self.scheduler.update_from_output(
            scheduler_output, model_output))

    return engine_core_outputs, scheduled_batch

wake_up

wake_up(tags: Optional[list[str]] = None)

Source code in vllm/v1/engine/core.py

def wake_up(self, tags: Optional[list[str]] = None):
    self.model_executor.wake_up(tags)

EngineCoreProc

Bases: EngineCore

ZMQ-wrapper for running EngineCore in background process.

Source code in vllm/v1/engine/core.py

class EngineCoreProc(EngineCore):
    """ZMQ-wrapper for running EngineCore in background process."""

    ENGINE_CORE_DEAD = b'ENGINE_CORE_DEAD'

    def __init__(
        self,
        vllm_config: VllmConfig,
        local_client: bool,
        handshake_address: str,
        executor_class: type[Executor],
        log_stats: bool,
        client_handshake_address: Optional[str] = None,
        engine_index: int = 0,
    ):
        self.input_queue = queue.Queue[tuple[EngineCoreRequestType, Any]]()
        self.output_queue = queue.Queue[Union[tuple[int, EngineCoreOutputs],
                                              bytes]]()
        executor_fail_callback = lambda: self.input_queue.put_nowait(
            (EngineCoreRequestType.EXECUTOR_FAILED, b''))

        self.engine_index = engine_index
        identity = self.engine_index.to_bytes(length=2, byteorder="little")
        self.engines_running = False

        with self._perform_handshakes(handshake_address, identity,
                                      local_client, vllm_config,
                                      client_handshake_address) as addresses:
            self.client_count = len(addresses.outputs)

            # Set up data parallel environment.
            self.has_coordinator = addresses.coordinator_output is not None
            self.frontend_stats_publish_address = (
                addresses.frontend_stats_publish_address)
            # Only publish request queue stats to coordinator for "internal"
            # LB mode.
            self.publish_dp_lb_stats = (
                self.has_coordinator
                and not vllm_config.parallel_config.data_parallel_external_lb)

            self._init_data_parallel(vllm_config)

            super().__init__(vllm_config, executor_class, log_stats,
                             executor_fail_callback)

        self.step_fn = (self.step if self.batch_queue is None else
                        self.step_with_batch_queue)

        # Background Threads and Queues for IO. These enable us to
        # overlap ZMQ socket IO with GPU since they release the GIL,
        # and to overlap some serialization/deserialization with the
        # model forward pass.
        # Threads handle Socket <-> Queues and core_busy_loop uses Queue.
        threading.Thread(target=self.process_input_sockets,
                         args=(addresses.inputs, addresses.coordinator_input,
                               identity),
                         daemon=True).start()
        self.output_thread = threading.Thread(
            target=self.process_output_sockets,
            args=(addresses.outputs, addresses.coordinator_output,
                  self.engine_index),
            daemon=True)
        self.output_thread.start()

    @contextmanager
    def _perform_handshakes(
        self,
        handshake_address: str,
        identity: bytes,
        local_client: bool,
        vllm_config: VllmConfig,
        client_handshake_address: Optional[str],
    ) -> Generator[EngineZmqAddresses, None, None]:
        """
        Perform startup handshakes.

        For DP=1 or offline mode, this is with the colocated front-end process.

        For DP>1 with internal loadbalancing this is with the shared front-end
        process which may reside on a different node.

        For DP>1 with external loadbalancing, two handshakes are performed:
            - With the rank 0 front-end process which retrieves the
              DP Coordinator ZMQ addresses and DP process group address.
            - With the colocated front-end process which retrieves the
              client input/output socket addresses.
        with the exception of the rank 0 engine itself which doesn't require
        the second handshake.

        Here, "front-end" process can mean the process containing the engine
        core client (which is the API server process in the case the API
        server is not scaled out), OR the launcher process running the
        run_multi_api_server() function in serve.py.
        """
        input_ctx = zmq.Context()
        is_local = local_client and client_handshake_address is None
        handshake = self._perform_handshake(input_ctx, handshake_address,
                                            identity, is_local, vllm_config,
                                            vllm_config.parallel_config)
        if client_handshake_address is None:
            with handshake as addresses:
                yield addresses
        else:
            local_handshake = self._perform_handshake(
                input_ctx, client_handshake_address, identity, local_client,
                vllm_config)
            with handshake as addresses, local_handshake as client_addresses:
                addresses.inputs = client_addresses.inputs
                addresses.outputs = client_addresses.outputs
                yield addresses

        # Update config which may have changed from the handshake
        vllm_config.__post_init__()

    @contextmanager
    def _perform_handshake(
        self,
        ctx: zmq.Context,
        handshake_address: str,
        identity: bytes,
        local_client: bool,
        vllm_config: VllmConfig,
        parallel_config_to_update: Optional[ParallelConfig] = None,
    ) -> Generator[EngineZmqAddresses, None, None]:
        with make_zmq_socket(ctx,
                             handshake_address,
                             zmq.DEALER,
                             identity=identity,
                             linger=5000,
                             bind=False) as handshake_socket:
            # Register engine with front-end.
            addresses = self.startup_handshake(handshake_socket, local_client,
                                               parallel_config_to_update)
            yield addresses

            # Send ready message.
            num_gpu_blocks = vllm_config.cache_config.num_gpu_blocks
            # We pass back the coordinator stats update address here for the
            # external LB case for our colocated front-end to use (coordinator
            # only runs with rank 0).
            dp_stats_address = self.frontend_stats_publish_address
            handshake_socket.send(
                msgspec.msgpack.encode({
                    "status": "READY",
                    "local": local_client,
                    "num_gpu_blocks": num_gpu_blocks,
                    "dp_stats_address": dp_stats_address,
                }))

    @staticmethod
    def startup_handshake(
        handshake_socket: zmq.Socket,
        local_client: bool,
        parallel_config: Optional[ParallelConfig] = None,
    ) -> EngineZmqAddresses:

        # Send registration message.
        handshake_socket.send(
            msgspec.msgpack.encode({
                "status": "HELLO",
                "local": local_client,
            }))

        # Receive initialization message.
        logger.info("Waiting for init message from front-end.")
        if not handshake_socket.poll(timeout=HANDSHAKE_TIMEOUT_MINS * 60_000):
            raise RuntimeError("Did not receive response from front-end "
                               f"process within {HANDSHAKE_TIMEOUT_MINS} "
                               f"minutes")
        init_bytes = handshake_socket.recv()
        init_message: EngineHandshakeMetadata = msgspec.msgpack.decode(
            init_bytes, type=EngineHandshakeMetadata)
        logger.debug("Received init message: %s", init_message)

        if parallel_config is not None:
            for key, value in init_message.parallel_config.items():
                setattr(parallel_config, key, value)

        return init_message.addresses

    @staticmethod
    def run_engine_core(*args,
                        dp_rank: int = 0,
                        local_dp_rank: int = 0,
                        **kwargs):
        """Launch EngineCore busy loop in background process."""

        # Signal handler used for graceful termination.
        # SystemExit exception is only raised once to allow this and worker
        # processes to terminate without error
        shutdown_requested = False

        # Ensure we can serialize transformer config after spawning
        maybe_register_config_serialize_by_value()

        def signal_handler(signum, frame):
            nonlocal shutdown_requested
            if not shutdown_requested:
                shutdown_requested = True
                raise SystemExit()

        # Either SIGTERM or SIGINT will terminate the engine_core
        signal.signal(signal.SIGTERM, signal_handler)
        signal.signal(signal.SIGINT, signal_handler)

        engine_core: Optional[EngineCoreProc] = None
        try:
            parallel_config: ParallelConfig = kwargs[
                "vllm_config"].parallel_config
            if parallel_config.data_parallel_size > 1 or dp_rank > 0:
                # Set data parallel rank for this engine process.
                parallel_config.data_parallel_rank = dp_rank
                parallel_config.data_parallel_rank_local = local_dp_rank
                engine_core = DPEngineCoreProc(*args, **kwargs)
            else:
                engine_core = EngineCoreProc(*args, **kwargs)

            engine_core.run_busy_loop()

        except SystemExit:
            logger.debug("EngineCore exiting.")
            raise
        except Exception as e:
            if engine_core is None:
                logger.exception("EngineCore failed to start.")
            else:
                logger.exception("EngineCore encountered a fatal error.")
                engine_core._send_engine_dead()
            raise e
        finally:
            if engine_core is not None:
                engine_core.shutdown()

    def _init_data_parallel(self, vllm_config: VllmConfig):
        pass

    def run_busy_loop(self):
        """Core busy loop of the EngineCore."""

        # Loop until process is sent a SIGINT or SIGTERM
        while True:
            # 1) Poll the input queue until there is work to do.
            self._process_input_queue()
            # 2) Step the engine core and return the outputs.
            self._process_engine_step()

    def _process_input_queue(self):
        """Exits when an engine step needs to be performed."""

        waited = False
        while not self.engines_running and not self.scheduler.has_requests():
            if logger.isEnabledFor(DEBUG) and self.input_queue.empty():
                logger.debug("EngineCore waiting for work.")
                waited = True
            req = self.input_queue.get()
            self._handle_client_request(*req)

        if waited:
            logger.debug("EngineCore loop active.")

        # Handle any more client requests.
        while not self.input_queue.empty():
            req = self.input_queue.get_nowait()
            self._handle_client_request(*req)

    def _process_engine_step(self) -> bool:
        """Called only when there are unfinished local requests."""

        # Step the engine core.
        outputs, model_executed = self.step_fn()
        # Put EngineCoreOutputs into the output queue.
        for output in (outputs.items() if outputs else ()):
            self.output_queue.put_nowait(output)

        return model_executed

    def _handle_client_request(self, request_type: EngineCoreRequestType,
                               request: Any) -> None:
        """Dispatch request from client."""

        if request_type == EngineCoreRequestType.ADD:
            self.add_request(request)
        elif request_type == EngineCoreRequestType.ABORT:
            self.abort_requests(request)
        elif request_type == EngineCoreRequestType.UTILITY:
            client_idx, call_id, method_name, args = request
            output = UtilityOutput(call_id)
            try:
                method = getattr(self, method_name)
                output.result = method(
                    *self._convert_msgspec_args(method, args))
            except BaseException as e:
                logger.exception("Invocation of %s method failed", method_name)
                output.failure_message = (f"Call to {method_name} method"
                                          f" failed: {str(e)}")
            self.output_queue.put_nowait(
                (client_idx, EngineCoreOutputs(utility_output=output)))
        elif request_type == EngineCoreRequestType.EXECUTOR_FAILED:
            raise RuntimeError("Executor failed.")
        else:
            logger.error("Unrecognized input request type encountered: %s",
                         request_type)

    @staticmethod
    def _convert_msgspec_args(method, args):
        """If a provided arg type doesn't match corresponding target method
         arg type, try converting to msgspec object."""
        if not args:
            return args
        arg_types = signature(method).parameters.values()
        assert len(args) <= len(arg_types)
        return tuple(
            msgspec.convert(v, type=p.annotation) if isclass(p.annotation)
            and issubclass(p.annotation, msgspec.Struct)
            and not isinstance(v, p.annotation) else v
            for v, p in zip(args, arg_types))

    def _send_engine_dead(self):
        """Send EngineDead status to the EngineCoreClient."""

        # Put ENGINE_CORE_DEAD in the queue.
        self.output_queue.put_nowait(EngineCoreProc.ENGINE_CORE_DEAD)

        # Wait until msg sent by the daemon before shutdown.
        self.output_thread.join(timeout=5.0)
        if self.output_thread.is_alive():
            logger.fatal("vLLM shutdown signal from EngineCore failed "
                         "to send. Please report this issue.")

    def process_input_sockets(self, input_addresses: list[str],
                              coord_input_address: Optional[str],
                              identity: bytes):
        """Input socket IO thread."""

        # Msgpack serialization decoding.
        add_request_decoder = MsgpackDecoder(EngineCoreRequest)
        generic_decoder = MsgpackDecoder()

        with ExitStack() as stack, zmq.Context() as ctx:
            input_sockets = [
                stack.enter_context(
                    make_zmq_socket(ctx,
                                    input_address,
                                    zmq.DEALER,
                                    identity=identity,
                                    bind=False))
                for input_address in input_addresses
            ]
            if coord_input_address is None:
                coord_socket = None
            else:
                coord_socket = stack.enter_context(
                    make_zmq_socket(ctx,
                                    coord_input_address,
                                    zmq.XSUB,
                                    identity=identity,
                                    bind=False))
                # Send subscription message to coordinator.
                coord_socket.send(b'\x01')

            # Register sockets with poller.
            poller = zmq.Poller()
            for input_socket in input_sockets:
                # Send initial message to each input socket - this is required
                # before the front-end ROUTER socket can send input messages
                # back to us.
                input_socket.send(b'')
                poller.register(input_socket, zmq.POLLIN)
            if coord_socket is not None:
                poller.register(coord_socket, zmq.POLLIN)

            while True:
                for input_socket, _ in poller.poll():
                    # (RequestType, RequestData)
                    type_frame, *data_frames = input_socket.recv_multipart(
                        copy=False)
                    request_type = EngineCoreRequestType(
                        bytes(type_frame.buffer))

                    # Deserialize the request data.
                    decoder = add_request_decoder if (
                        request_type
                        == EngineCoreRequestType.ADD) else generic_decoder
                    request = decoder.decode(data_frames)

                    # Push to input queue for core busy loop.
                    self.input_queue.put_nowait((request_type, request))

    def process_output_sockets(self, output_paths: list[str],
                               coord_output_path: Optional[str],
                               engine_index: int):
        """Output socket IO thread."""

        # Msgpack serialization encoding.
        encoder = MsgpackEncoder()
        # Send buffers to reuse.
        reuse_buffers: list[bytearray] = []
        # Keep references to outputs and buffers until zmq is finished
        # with them (outputs may contain tensors/np arrays whose
        # backing buffers were extracted for zero-copy send).
        pending = deque[tuple[zmq.MessageTracker, Any, bytearray]]()

        # We must set linger to ensure the ENGINE_CORE_DEAD
        # message is sent prior to closing the socket.
        with ExitStack() as stack, zmq.Context() as ctx:
            sockets = [
                stack.enter_context(
                    make_zmq_socket(ctx, output_path, zmq.PUSH, linger=4000))
                for output_path in output_paths
            ]
            coord_socket = stack.enter_context(
                make_zmq_socket(
                    ctx, coord_output_path, zmq.PUSH, bind=False,
                    linger=4000)) if coord_output_path is not None else None
            max_reuse_bufs = len(sockets) + 1

            while True:
                output = self.output_queue.get()
                if output == EngineCoreProc.ENGINE_CORE_DEAD:
                    for socket in sockets:
                        socket.send(output)
                    break
                assert not isinstance(output, bytes)
                client_index, outputs = output
                outputs.engine_index = engine_index

                if client_index == -1:
                    # Don't reuse buffer for coordinator message
                    # which will be very small.
                    assert coord_socket is not None
                    coord_socket.send_multipart(encoder.encode(outputs))
                    continue

                # Reclaim buffers that zmq is finished with.
                while pending and pending[-1][0].done:
                    reuse_buffers.append(pending.pop()[2])

                buffer = reuse_buffers.pop() if reuse_buffers else bytearray()
                buffers = encoder.encode_into(outputs, buffer)
                tracker = sockets[client_index].send_multipart(buffers,
                                                               copy=False,
                                                               track=True)
                if not tracker.done:
                    ref = outputs if len(buffers) > 1 else None
                    pending.appendleft((tracker, ref, buffer))
                elif len(reuse_buffers) < max_reuse_bufs:
                    # Limit the number of buffers to reuse.
                    reuse_buffers.append(buffer)

ENGINE_CORE_DEAD class-attribute instance-attribute

ENGINE_CORE_DEAD = b'ENGINE_CORE_DEAD'

client_count instance-attribute

client_count = len(outputs)

engine_index instance-attribute

engine_index = engine_index

engines_running instance-attribute

engines_running = False

frontend_stats_publish_address instance-attribute

frontend_stats_publish_address = (
    frontend_stats_publish_address
)

has_coordinator instance-attribute

has_coordinator = coordinator_output is not None

input_queue instance-attribute

input_queue = Queue[tuple[EngineCoreRequestType, Any]]()

output_queue instance-attribute

output_queue = Queue[
    Union[tuple[int, EngineCoreOutputs], bytes]
]()

output_thread instance-attribute

output_thread = Thread(
    target=process_output_sockets,
    args=(outputs, coordinator_output, engine_index),
    daemon=True,
)

publish_dp_lb_stats instance-attribute

publish_dp_lb_stats = (
    has_coordinator and not data_parallel_external_lb
)

step_fn instance-attribute

step_fn = (
    step if batch_queue is None else step_with_batch_queue
)

__init__

__init__(
    vllm_config: VllmConfig,
    local_client: bool,
    handshake_address: str,
    executor_class: type[Executor],
    log_stats: bool,
    client_handshake_address: Optional[str] = None,
    engine_index: int = 0,
)

Source code in vllm/v1/engine/core.py

def __init__(
    self,
    vllm_config: VllmConfig,
    local_client: bool,
    handshake_address: str,
    executor_class: type[Executor],
    log_stats: bool,
    client_handshake_address: Optional[str] = None,
    engine_index: int = 0,
):
    self.input_queue = queue.Queue[tuple[EngineCoreRequestType, Any]]()
    self.output_queue = queue.Queue[Union[tuple[int, EngineCoreOutputs],
                                          bytes]]()
    executor_fail_callback = lambda: self.input_queue.put_nowait(
        (EngineCoreRequestType.EXECUTOR_FAILED, b''))

    self.engine_index = engine_index
    identity = self.engine_index.to_bytes(length=2, byteorder="little")
    self.engines_running = False

    with self._perform_handshakes(handshake_address, identity,
                                  local_client, vllm_config,
                                  client_handshake_address) as addresses:
        self.client_count = len(addresses.outputs)

        # Set up data parallel environment.
        self.has_coordinator = addresses.coordinator_output is not None
        self.frontend_stats_publish_address = (
            addresses.frontend_stats_publish_address)
        # Only publish request queue stats to coordinator for "internal"
        # LB mode.
        self.publish_dp_lb_stats = (
            self.has_coordinator
            and not vllm_config.parallel_config.data_parallel_external_lb)

        self._init_data_parallel(vllm_config)

        super().__init__(vllm_config, executor_class, log_stats,
                         executor_fail_callback)

    self.step_fn = (self.step if self.batch_queue is None else
                    self.step_with_batch_queue)

    # Background Threads and Queues for IO. These enable us to
    # overlap ZMQ socket IO with GPU since they release the GIL,
    # and to overlap some serialization/deserialization with the
    # model forward pass.
    # Threads handle Socket <-> Queues and core_busy_loop uses Queue.
    threading.Thread(target=self.process_input_sockets,
                     args=(addresses.inputs, addresses.coordinator_input,
                           identity),
                     daemon=True).start()
    self.output_thread = threading.Thread(
        target=self.process_output_sockets,
        args=(addresses.outputs, addresses.coordinator_output,
              self.engine_index),
        daemon=True)
    self.output_thread.start()

_convert_msgspec_args staticmethod

_convert_msgspec_args(method, args)

If a provided arg type doesn't match corresponding target method arg type, try converting to msgspec object.

Source code in vllm/v1/engine/core.py

@staticmethod
def _convert_msgspec_args(method, args):
    """If a provided arg type doesn't match corresponding target method
     arg type, try converting to msgspec object."""
    if not args:
        return args
    arg_types = signature(method).parameters.values()
    assert len(args) <= len(arg_types)
    return tuple(
        msgspec.convert(v, type=p.annotation) if isclass(p.annotation)
        and issubclass(p.annotation, msgspec.Struct)
        and not isinstance(v, p.annotation) else v
        for v, p in zip(args, arg_types))

_handle_client_request

_handle_client_request(
    request_type: EngineCoreRequestType, request: Any
) -> None

Dispatch request from client.

Source code in vllm/v1/engine/core.py

def _handle_client_request(self, request_type: EngineCoreRequestType,
                           request: Any) -> None:
    """Dispatch request from client."""

    if request_type == EngineCoreRequestType.ADD:
        self.add_request(request)
    elif request_type == EngineCoreRequestType.ABORT:
        self.abort_requests(request)
    elif request_type == EngineCoreRequestType.UTILITY:
        client_idx, call_id, method_name, args = request
        output = UtilityOutput(call_id)
        try:
            method = getattr(self, method_name)
            output.result = method(
                *self._convert_msgspec_args(method, args))
        except BaseException as e:
            logger.exception("Invocation of %s method failed", method_name)
            output.failure_message = (f"Call to {method_name} method"
                                      f" failed: {str(e)}")
        self.output_queue.put_nowait(
            (client_idx, EngineCoreOutputs(utility_output=output)))
    elif request_type == EngineCoreRequestType.EXECUTOR_FAILED:
        raise RuntimeError("Executor failed.")
    else:
        logger.error("Unrecognized input request type encountered: %s",
                     request_type)

_init_data_parallel

_init_data_parallel(vllm_config: VllmConfig)

Source code in vllm/v1/engine/core.py

def _init_data_parallel(self, vllm_config: VllmConfig):
    pass

_perform_handshake

_perform_handshake(
    ctx: Context,
    handshake_address: str,
    identity: bytes,
    local_client: bool,
    vllm_config: VllmConfig,
    parallel_config_to_update: Optional[
        ParallelConfig
    ] = None,
) -> Generator[EngineZmqAddresses, None, None]

Source code in vllm/v1/engine/core.py

@contextmanager
def _perform_handshake(
    self,
    ctx: zmq.Context,
    handshake_address: str,
    identity: bytes,
    local_client: bool,
    vllm_config: VllmConfig,
    parallel_config_to_update: Optional[ParallelConfig] = None,
) -> Generator[EngineZmqAddresses, None, None]:
    with make_zmq_socket(ctx,
                         handshake_address,
                         zmq.DEALER,
                         identity=identity,
                         linger=5000,
                         bind=False) as handshake_socket:
        # Register engine with front-end.
        addresses = self.startup_handshake(handshake_socket, local_client,
                                           parallel_config_to_update)
        yield addresses

        # Send ready message.
        num_gpu_blocks = vllm_config.cache_config.num_gpu_blocks
        # We pass back the coordinator stats update address here for the
        # external LB case for our colocated front-end to use (coordinator
        # only runs with rank 0).
        dp_stats_address = self.frontend_stats_publish_address
        handshake_socket.send(
            msgspec.msgpack.encode({
                "status": "READY",
                "local": local_client,
                "num_gpu_blocks": num_gpu_blocks,
                "dp_stats_address": dp_stats_address,
            }))

_perform_handshakes

_perform_handshakes(
    handshake_address: str,
    identity: bytes,
    local_client: bool,
    vllm_config: VllmConfig,
    client_handshake_address: Optional[str],
) -> Generator[EngineZmqAddresses, None, None]

Perform startup handshakes.

For DP=1 or offline mode, this is with the colocated front-end process.

For DP>1 with internal loadbalancing this is with the shared front-end process which may reside on a different node.

For DP>1 with external loadbalancing, two handshakes are performed: - With the rank 0 front-end process which retrieves the DP Coordinator ZMQ addresses and DP process group address. - With the colocated front-end process which retrieves the client input/output socket addresses. with the exception of the rank 0 engine itself which doesn't require the second handshake.

Here, "front-end" process can mean the process containing the engine core client (which is the API server process in the case the API server is not scaled out), OR the launcher process running the run_multi_api_server() function in serve.py.

Source code in vllm/v1/engine/core.py

@contextmanager
def _perform_handshakes(
    self,
    handshake_address: str,
    identity: bytes,
    local_client: bool,
    vllm_config: VllmConfig,
    client_handshake_address: Optional[str],
) -> Generator[EngineZmqAddresses, None, None]:
    """
    Perform startup handshakes.

    For DP=1 or offline mode, this is with the colocated front-end process.

    For DP>1 with internal loadbalancing this is with the shared front-end
    process which may reside on a different node.

    For DP>1 with external loadbalancing, two handshakes are performed:
        - With the rank 0 front-end process which retrieves the
          DP Coordinator ZMQ addresses and DP process group address.
        - With the colocated front-end process which retrieves the
          client input/output socket addresses.
    with the exception of the rank 0 engine itself which doesn't require
    the second handshake.

    Here, "front-end" process can mean the process containing the engine
    core client (which is the API server process in the case the API
    server is not scaled out), OR the launcher process running the
    run_multi_api_server() function in serve.py.
    """
    input_ctx = zmq.Context()
    is_local = local_client and client_handshake_address is None
    handshake = self._perform_handshake(input_ctx, handshake_address,
                                        identity, is_local, vllm_config,
                                        vllm_config.parallel_config)
    if client_handshake_address is None:
        with handshake as addresses:
            yield addresses
    else:
        local_handshake = self._perform_handshake(
            input_ctx, client_handshake_address, identity, local_client,
            vllm_config)
        with handshake as addresses, local_handshake as client_addresses:
            addresses.inputs = client_addresses.inputs
            addresses.outputs = client_addresses.outputs
            yield addresses

    # Update config which may have changed from the handshake
    vllm_config.__post_init__()

_process_engine_step

_process_engine_step() -> bool

Called only when there are unfinished local requests.

Source code in vllm/v1/engine/core.py

def _process_engine_step(self) -> bool:
    """Called only when there are unfinished local requests."""

    # Step the engine core.
    outputs, model_executed = self.step_fn()
    # Put EngineCoreOutputs into the output queue.
    for output in (outputs.items() if outputs else ()):
        self.output_queue.put_nowait(output)

    return model_executed

_process_input_queue

_process_input_queue()

Exits when an engine step needs to be performed.

Source code in vllm/v1/engine/core.py

def _process_input_queue(self):
    """Exits when an engine step needs to be performed."""

    waited = False
    while not self.engines_running and not self.scheduler.has_requests():
        if logger.isEnabledFor(DEBUG) and self.input_queue.empty():
            logger.debug("EngineCore waiting for work.")
            waited = True
        req = self.input_queue.get()
        self._handle_client_request(*req)

    if waited:
        logger.debug("EngineCore loop active.")

    # Handle any more client requests.
    while not self.input_queue.empty():
        req = self.input_queue.get_nowait()
        self._handle_client_request(*req)

_send_engine_dead

_send_engine_dead()

Send EngineDead status to the EngineCoreClient.

Source code in vllm/v1/engine/core.py

def _send_engine_dead(self):
    """Send EngineDead status to the EngineCoreClient."""

    # Put ENGINE_CORE_DEAD in the queue.
    self.output_queue.put_nowait(EngineCoreProc.ENGINE_CORE_DEAD)

    # Wait until msg sent by the daemon before shutdown.
    self.output_thread.join(timeout=5.0)
    if self.output_thread.is_alive():
        logger.fatal("vLLM shutdown signal from EngineCore failed "
                     "to send. Please report this issue.")

process_input_sockets

process_input_sockets(
    input_addresses: list[str],
    coord_input_address: Optional[str],
    identity: bytes,
)

Input socket IO thread.

Source code in vllm/v1/engine/core.py

def process_input_sockets(self, input_addresses: list[str],
                          coord_input_address: Optional[str],
                          identity: bytes):
    """Input socket IO thread."""

    # Msgpack serialization decoding.
    add_request_decoder = MsgpackDecoder(EngineCoreRequest)
    generic_decoder = MsgpackDecoder()

    with ExitStack() as stack, zmq.Context() as ctx:
        input_sockets = [
            stack.enter_context(
                make_zmq_socket(ctx,
                                input_address,
                                zmq.DEALER,
                                identity=identity,
                                bind=False))
            for input_address in input_addresses
        ]
        if coord_input_address is None:
            coord_socket = None
        else:
            coord_socket = stack.enter_context(
                make_zmq_socket(ctx,
                                coord_input_address,
                                zmq.XSUB,
                                identity=identity,
                                bind=False))
            # Send subscription message to coordinator.
            coord_socket.send(b'\x01')

        # Register sockets with poller.
        poller = zmq.Poller()
        for input_socket in input_sockets:
            # Send initial message to each input socket - this is required
            # before the front-end ROUTER socket can send input messages
            # back to us.
            input_socket.send(b'')
            poller.register(input_socket, zmq.POLLIN)
        if coord_socket is not None:
            poller.register(coord_socket, zmq.POLLIN)

        while True:
            for input_socket, _ in poller.poll():
                # (RequestType, RequestData)
                type_frame, *data_frames = input_socket.recv_multipart(
                    copy=False)
                request_type = EngineCoreRequestType(
                    bytes(type_frame.buffer))

                # Deserialize the request data.
                decoder = add_request_decoder if (
                    request_type
                    == EngineCoreRequestType.ADD) else generic_decoder
                request = decoder.decode(data_frames)

                # Push to input queue for core busy loop.
                self.input_queue.put_nowait((request_type, request))

process_output_sockets

process_output_sockets(
    output_paths: list[str],
    coord_output_path: Optional[str],
    engine_index: int,
)

Output socket IO thread.

Source code in vllm/v1/engine/core.py

def process_output_sockets(self, output_paths: list[str],
                           coord_output_path: Optional[str],
                           engine_index: int):
    """Output socket IO thread."""

    # Msgpack serialization encoding.
    encoder = MsgpackEncoder()
    # Send buffers to reuse.
    reuse_buffers: list[bytearray] = []
    # Keep references to outputs and buffers until zmq is finished
    # with them (outputs may contain tensors/np arrays whose
    # backing buffers were extracted for zero-copy send).
    pending = deque[tuple[zmq.MessageTracker, Any, bytearray]]()

    # We must set linger to ensure the ENGINE_CORE_DEAD
    # message is sent prior to closing the socket.
    with ExitStack() as stack, zmq.Context() as ctx:
        sockets = [
            stack.enter_context(
                make_zmq_socket(ctx, output_path, zmq.PUSH, linger=4000))
            for output_path in output_paths
        ]
        coord_socket = stack.enter_context(
            make_zmq_socket(
                ctx, coord_output_path, zmq.PUSH, bind=False,
                linger=4000)) if coord_output_path is not None else None
        max_reuse_bufs = len(sockets) + 1

        while True:
            output = self.output_queue.get()
            if output == EngineCoreProc.ENGINE_CORE_DEAD:
                for socket in sockets:
                    socket.send(output)
                break
            assert not isinstance(output, bytes)
            client_index, outputs = output
            outputs.engine_index = engine_index

            if client_index == -1:
                # Don't reuse buffer for coordinator message
                # which will be very small.
                assert coord_socket is not None
                coord_socket.send_multipart(encoder.encode(outputs))
                continue

            # Reclaim buffers that zmq is finished with.
            while pending and pending[-1][0].done:
                reuse_buffers.append(pending.pop()[2])

            buffer = reuse_buffers.pop() if reuse_buffers else bytearray()
            buffers = encoder.encode_into(outputs, buffer)
            tracker = sockets[client_index].send_multipart(buffers,
                                                           copy=False,
                                                           track=True)
            if not tracker.done:
                ref = outputs if len(buffers) > 1 else None
                pending.appendleft((tracker, ref, buffer))
            elif len(reuse_buffers) < max_reuse_bufs:
                # Limit the number of buffers to reuse.
                reuse_buffers.append(buffer)

run_busy_loop

run_busy_loop()

Core busy loop of the EngineCore.

Source code in vllm/v1/engine/core.py

def run_busy_loop(self):
    """Core busy loop of the EngineCore."""

    # Loop until process is sent a SIGINT or SIGTERM
    while True:
        # 1) Poll the input queue until there is work to do.
        self._process_input_queue()
        # 2) Step the engine core and return the outputs.
        self._process_engine_step()

run_engine_core staticmethod

run_engine_core(
    *args,
    dp_rank: int = 0,
    local_dp_rank: int = 0,
    **kwargs,
)

Launch EngineCore busy loop in background process.

Source code in vllm/v1/engine/core.py

@staticmethod
def run_engine_core(*args,
                    dp_rank: int = 0,
                    local_dp_rank: int = 0,
                    **kwargs):
    """Launch EngineCore busy loop in background process."""

    # Signal handler used for graceful termination.
    # SystemExit exception is only raised once to allow this and worker
    # processes to terminate without error
    shutdown_requested = False

    # Ensure we can serialize transformer config after spawning
    maybe_register_config_serialize_by_value()

    def signal_handler(signum, frame):
        nonlocal shutdown_requested
        if not shutdown_requested:
            shutdown_requested = True
            raise SystemExit()

    # Either SIGTERM or SIGINT will terminate the engine_core
    signal.signal(signal.SIGTERM, signal_handler)
    signal.signal(signal.SIGINT, signal_handler)

    engine_core: Optional[EngineCoreProc] = None
    try:
        parallel_config: ParallelConfig = kwargs[
            "vllm_config"].parallel_config
        if parallel_config.data_parallel_size > 1 or dp_rank > 0:
            # Set data parallel rank for this engine process.
            parallel_config.data_parallel_rank = dp_rank
            parallel_config.data_parallel_rank_local = local_dp_rank
            engine_core = DPEngineCoreProc(*args, **kwargs)
        else:
            engine_core = EngineCoreProc(*args, **kwargs)

        engine_core.run_busy_loop()

    except SystemExit:
        logger.debug("EngineCore exiting.")
        raise
    except Exception as e:
        if engine_core is None:
            logger.exception("EngineCore failed to start.")
        else:
            logger.exception("EngineCore encountered a fatal error.")
            engine_core._send_engine_dead()
        raise e
    finally:
        if engine_core is not None:
            engine_core.shutdown()

startup_handshake staticmethod

startup_handshake(
    handshake_socket: Socket,
    local_client: bool,
    parallel_config: Optional[ParallelConfig] = None,
) -> EngineZmqAddresses

Source code in vllm/v1/engine/core.py

@staticmethod
def startup_handshake(
    handshake_socket: zmq.Socket,
    local_client: bool,
    parallel_config: Optional[ParallelConfig] = None,
) -> EngineZmqAddresses:

    # Send registration message.
    handshake_socket.send(
        msgspec.msgpack.encode({
            "status": "HELLO",
            "local": local_client,
        }))

    # Receive initialization message.
    logger.info("Waiting for init message from front-end.")
    if not handshake_socket.poll(timeout=HANDSHAKE_TIMEOUT_MINS * 60_000):
        raise RuntimeError("Did not receive response from front-end "
                           f"process within {HANDSHAKE_TIMEOUT_MINS} "
                           f"minutes")
    init_bytes = handshake_socket.recv()
    init_message: EngineHandshakeMetadata = msgspec.msgpack.decode(
        init_bytes, type=EngineHandshakeMetadata)
    logger.debug("Received init message: %s", init_message)

    if parallel_config is not None:
        for key, value in init_message.parallel_config.items():
            setattr(parallel_config, key, value)

    return init_message.addresses