1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540
/*
* Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
* SPDX-License-Identifier: Apache-2.0
*/
use std::collections::HashMap;
use std::convert::Infallible;
use std::net::TcpListener as StdTcpListener;
use std::ops::Deref;
use std::process;
use std::sync::{mpsc, Arc};
use std::thread;
use aws_smithy_http_server::{
body::{Body, BoxBody},
routing::IntoMakeService,
};
use http::{Request, Response};
use hyper::server::conn::AddrIncoming;
use parking_lot::Mutex;
use pyo3::{prelude::*, types::IntoPyDict};
use signal_hook::{consts::*, iterator::Signals};
use socket2::Socket;
use tokio::{net::TcpListener, runtime};
use tokio_rustls::TlsAcceptor;
use tower::{util::BoxCloneService, ServiceBuilder};
use crate::{
context::{layer::AddPyContextLayer, PyContext},
tls::{listener::Listener as TlsListener, PyTlsConfig},
util::{error::rich_py_err, func_metadata},
PySocket,
};
/// A Python handler function representation.
///
/// The Python business logic implementation needs to carry some information
/// to be executed properly like the size of its arguments and if it is
/// a coroutine.
#[pyclass]
#[derive(Debug, Clone)]
pub struct PyHandler {
pub func: PyObject,
// Number of args is needed to decide whether handler accepts context as an argument
pub args: usize,
pub is_coroutine: bool,
}
impl Deref for PyHandler {
type Target = PyObject;
fn deref(&self) -> &Self::Target {
&self.func
}
}
// A `BoxCloneService` with default `Request`, `Response` and `Error`.
type Service = BoxCloneService<Request<Body>, Response<BoxBody>, Infallible>;
/// Trait defining a Python application.
///
/// A Python application requires handling of multiple processes, signals and allows to register Python
/// function that will be executed as business logic by the code generated Rust handlers.
/// To properly function, the application requires some state:
/// * `workers`: the list of child Python worker processes, protected by a Mutex.
/// * `context`: the optional Python object that should be passed inside the Rust state struct.
/// * `handlers`: the mapping between an operation name and its [PyHandler] representation.
///
/// Since the Python application is spawning multiple workers, it also requires signal handling to allow the gracefull
/// termination of multiple Hyper servers. The main Rust process is registering signal and using them to understand when it
/// it time to loop through all the active workers and terminate them. Workers registers their own signal handlers and attaches
/// them to the Python event loop, ensuring all coroutines are cancelled before terminating a worker.
///
/// This trait will be implemented by the code generated by the `PythonApplicationGenerator` Kotlin class.
pub trait PyApp: Clone + pyo3::IntoPy<PyObject> {
/// List of active Python workers registered with this application.
fn workers(&self) -> &Mutex<Vec<PyObject>>;
/// Optional Python context object that will be passed as part of the Rust state.
fn context(&self) -> &Option<PyObject>;
/// Mapping between operation names and their `PyHandler` representation.
fn handlers(&mut self) -> &mut HashMap<String, PyHandler>;
/// Build the app's `Service` using given `event_loop`.
fn build_service(&mut self, event_loop: &pyo3::PyAny) -> pyo3::PyResult<Service>;
/// Handle the graceful termination of Python workers by looping through all the
/// active workers and calling `terminate()` on them. If termination fails, this
/// method will try to `kill()` any failed worker.
fn graceful_termination(&self, workers: &Mutex<Vec<PyObject>>) -> ! {
let workers = workers.lock();
for (idx, worker) in workers.iter().enumerate() {
let idx = idx + 1;
Python::with_gil(|py| {
let pid: isize = worker
.getattr(py, "pid")
.map(|pid| pid.extract(py).unwrap_or(-1))
.unwrap_or(-1);
tracing::debug!(idx, pid, "terminating worker");
match worker.call_method0(py, "terminate") {
Ok(_) => {}
Err(e) => {
tracing::error!(error = ?rich_py_err(e), idx, pid, "error terminating worker");
worker
.call_method0(py, "kill")
.map_err(|e| {
tracing::error!(
error = ?rich_py_err(e), idx, pid, "unable to kill kill worker"
);
})
.unwrap();
}
}
});
}
process::exit(0);
}
/// Handler the immediate termination of Python workers by looping through all the
/// active workers and calling `kill()` on them.
fn immediate_termination(&self, workers: &Mutex<Vec<PyObject>>) -> ! {
let workers = workers.lock();
for (idx, worker) in workers.iter().enumerate() {
let idx = idx + 1;
Python::with_gil(|py| {
let pid: isize = worker
.getattr(py, "pid")
.map(|pid| pid.extract(py).unwrap_or(-1))
.unwrap_or(-1);
tracing::debug!(idx, pid, "killing worker");
worker
.call_method0(py, "kill")
.map_err(|e| {
tracing::error!(error = ?rich_py_err(e), idx, pid, "unable to kill kill worker");
})
.unwrap();
});
}
process::exit(0);
}
/// Register and handler signals of the main Rust thread. Signals not registered
/// in this method are ignored.
///
/// Signals supported:
/// * SIGTERM|SIGQUIT - graceful termination of all workers.
/// * SIGINT - immediate termination of all workers.
///
/// Other signals are NOOP.
fn block_on_rust_signals(&self) {
let mut signals =
Signals::new([SIGINT, SIGHUP, SIGQUIT, SIGTERM, SIGUSR1, SIGUSR2, SIGWINCH])
.expect("Unable to register signals");
for sig in signals.forever() {
match sig {
SIGINT => {
tracing::info!(
sig = %sig, "termination signal received, all workers will be immediately terminated"
);
self.immediate_termination(self.workers());
}
SIGTERM | SIGQUIT => {
tracing::info!(
sig = %sig, "termination signal received, all workers will be gracefully terminated"
);
self.graceful_termination(self.workers());
}
_ => {
tracing::debug!(sig = %sig, "signal is ignored by this application");
}
}
}
}
/// Register and handle termination of all the tasks on the Python asynchronous event loop.
/// We only register SIGQUIT and SIGINT since the main signal handling is done by Rust.
fn register_python_signals(&self, py: Python, event_loop: PyObject) -> PyResult<()> {
let locals = [("event_loop", event_loop)].into_py_dict(py);
py.run(
r#"
import asyncio
import logging
import functools
import signal
async def shutdown(sig, event_loop):
# reimport asyncio and logging to be sure they are available when
# this handler runs on signal catching.
import asyncio
import logging
logging.info(f"Caught signal {sig.name}, cancelling tasks registered on this loop")
tasks = [task for task in asyncio.all_tasks() if task is not
asyncio.current_task()]
list(map(lambda task: task.cancel(), tasks))
results = await asyncio.gather(*tasks, return_exceptions=True)
logging.debug(f"Finished awaiting cancelled tasks, results: {results}")
event_loop.stop()
event_loop.add_signal_handler(signal.SIGTERM,
functools.partial(asyncio.ensure_future, shutdown(signal.SIGTERM, event_loop)))
event_loop.add_signal_handler(signal.SIGINT,
functools.partial(asyncio.ensure_future, shutdown(signal.SIGINT, event_loop)))
"#,
None,
Some(locals),
)?;
Ok(())
}
/// Start a single worker with its own Tokio and Python async runtime and provided shared socket.
///
/// Python asynchronous loop needs to be started and handled during the lifetime of the process and
/// it is passed to this method by the caller, which can use
/// [configure_python_event_loop](#method.configure_python_event_loop) to properly setup it up.
///
/// We retrieve the Python context object, if setup by the user calling [PyApp::context] method,
/// generate the state structure and build the [aws_smithy_http_server::routing::Router], filling
/// it with the functions generated by `PythonServerOperationHandlerGenerator.kt`.
/// At last we get a cloned reference to the underlying [socket2::Socket].
///
/// Now that all the setup is done, we can start the two runtimes and run the [hyper] server.
/// We spawn a thread with a new [tokio::runtime], setup the middlewares and finally block the
/// thread on Hyper serve() method.
/// The main process continues and at the end it is blocked on Python `loop.run_forever()`.
///
/// [uvloop]: https://github.com/MagicStack/uvloop
fn start_hyper_worker(
&mut self,
py: Python,
socket: &PyCell<PySocket>,
event_loop: &PyAny,
service: Service,
worker_number: isize,
tls: Option<PyTlsConfig>,
) -> PyResult<()> {
// Clone the socket.
let borrow = socket.try_borrow_mut()?;
let held_socket: &PySocket = &borrow;
let raw_socket = held_socket.get_socket()?;
// Register signals on the Python event loop.
self.register_python_signals(py, event_loop.to_object(py))?;
// Spawn a new background [std::thread] to run the application.
// This is needed because `asyncio` doesn't work properly if it doesn't control the main thread.
// At the end of this function you can see we are calling `event_loop.run_forever()` to
// yield execution of main thread to `asyncio` runtime.
// For more details: https://docs.rs/pyo3-asyncio/latest/pyo3_asyncio/#pythons-event-loop-and-the-main-thread
tracing::trace!("start the tokio runtime in a background task");
thread::spawn(move || {
// The thread needs a new [tokio] runtime.
let rt = runtime::Builder::new_multi_thread()
.enable_all()
.thread_name(format!("smithy-rs-tokio[{worker_number}]"))
.build()
.expect("unable to start a new tokio runtime for this process");
rt.block_on(async move {
let addr = addr_incoming_from_socket(raw_socket);
if let Some(config) = tls {
let (acceptor, acceptor_rx) = tls_config_reloader(config);
let listener = TlsListener::new(acceptor, addr, acceptor_rx);
let server =
hyper::Server::builder(listener).serve(IntoMakeService::new(service));
tracing::trace!("started tls hyper server from shared socket");
// Run forever-ish...
if let Err(err) = server.await {
tracing::error!(error = ?err, "server error");
}
} else {
let server = hyper::Server::builder(addr).serve(IntoMakeService::new(service));
tracing::trace!("started hyper server from shared socket");
// Run forever-ish...
if let Err(err) = server.await {
tracing::error!(error = ?err, "server error");
}
}
});
});
// Block on the event loop forever.
tracing::trace!("run and block on the python event loop until a signal is received");
event_loop.call_method0("run_forever")?;
Ok(())
}
/// Register a Python function to be executed inside the Smithy Rust handler.
///
/// There are some information needed to execute the Python code from a Rust handler,
/// such has if the registered function needs to be awaited (if it is a coroutine) and
/// the number of arguments available, which tells us if the handler wants the state to be
/// passed or not.
fn register_operation(&mut self, py: Python, name: &str, func: PyObject) -> PyResult<()> {
let func_metadata = func_metadata(py, &func)?;
let handler = PyHandler {
func,
is_coroutine: func_metadata.is_coroutine,
args: func_metadata.num_args,
};
tracing::info!(
name,
is_coroutine = handler.is_coroutine,
args = handler.args,
"registering handler function",
);
// Insert the handler in the handlers map.
self.handlers().insert(name.to_string(), handler);
Ok(())
}
/// Configure the Python asyncio event loop.
///
/// First of all we install [uvloop] as the main Python event loop. Thanks to libuv, uvloop
/// performs ~20% better than Python standard event loop in most benchmarks, while being 100%
/// compatible. If [uvloop] is not available as a dependency, we just fall back to the standard
/// Python event loop.
///
/// [uvloop]: https://github.com/MagicStack/uvloop
fn configure_python_event_loop<'py>(&self, py: Python<'py>) -> PyResult<&'py PyAny> {
let asyncio = py.import("asyncio")?;
match py.import("uvloop") {
Ok(uvloop) => {
uvloop.call_method0("install")?;
tracing::trace!("setting up uvloop for current process");
}
Err(_) => {
tracing::warn!("uvloop not found, using python standard event loop, which could have worse performance than uvloop");
}
}
let event_loop = asyncio.call_method0("new_event_loop")?;
asyncio.call_method1("set_event_loop", (event_loop,))?;
Ok(event_loop)
}
/// Main entrypoint: start the server on multiple workers.
///
/// The multiprocessing server is achieved using the ability of a Python interpreter
/// to clone and start itself as a new process.
/// The shared sockets is created and Using the [multiprocessing::Process] module, multiple
/// workers with the method `self.start_worker()` as target are started.
///
/// NOTE: this method ends up calling `self.start_worker` from the Python context, forcing
/// the struct implementing this trait to also implement a `start_worker` method.
/// This is done to ensure the Python event loop is started in the right child process space before being
/// passed to `start_hyper_worker`.
///
/// `PythonApplicationGenerator.kt` generates the `start_worker` method:
///
/// ```no_run
/// use std::convert::Infallible;
/// use std::collections::HashMap;
/// use pyo3::prelude::*;
/// use aws_smithy_http_server_python::{PyApp, PyHandler};
/// use aws_smithy_http_server::body::{Body, BoxBody};
/// use parking_lot::Mutex;
/// use http::{Request, Response};
/// use tower::util::BoxCloneService;
///
/// #[pyclass]
/// #[derive(Debug, Clone)]
/// pub struct App {};
///
/// impl PyApp for App {
/// fn workers(&self) -> &Mutex<Vec<PyObject>> { todo!() }
/// fn context(&self) -> &Option<PyObject> { todo!() }
/// fn handlers(&mut self) -> &mut HashMap<String, PyHandler> { todo!() }
/// fn build_service(&mut self, event_loop: &PyAny) -> PyResult<BoxCloneService<Request<Body>, Response<BoxBody>, Infallible>> { todo!() }
/// }
///
/// #[pymethods]
/// impl App {
/// #[pyo3(text_signature = "($self, socket, worker_number, tls)")]
/// pub fn start_worker(
/// &mut self,
/// py: pyo3::Python,
/// socket: &pyo3::PyCell<aws_smithy_http_server_python::PySocket>,
/// worker_number: isize,
/// tls: Option<aws_smithy_http_server_python::tls::PyTlsConfig>,
/// ) -> pyo3::PyResult<()> {
/// let event_loop = self.configure_python_event_loop(py)?;
/// let service = self.build_service(event_loop)?;
/// self.start_hyper_worker(py, socket, event_loop, service, worker_number, tls)
/// }
/// }
/// ```
///
/// [multiprocessing::Process]: https://docs.python.org/3/library/multiprocessing.html
fn run_server(
&mut self,
py: Python,
address: Option<String>,
port: Option<i32>,
backlog: Option<i32>,
workers: Option<usize>,
tls: Option<PyTlsConfig>,
) -> PyResult<()> {
// Setup multiprocessing environment, allowing connections and socket
// sharing between processes.
let mp = py.import("multiprocessing")?;
// https://github.com/python/cpython/blob/f4c03484da59049eb62a9bf7777b963e2267d187/Lib/multiprocessing/context.py#L164
mp.call_method0("allow_connection_pickling")?;
// Starting from Python 3.8, on macOS, the spawn start method is now the default. See bpo-33725.
// This forces the `PyApp` class to be pickled when it is shared between different process,
// which is currently not supported by PyO3 classes.
//
// Forcing the multiprocessing start method to fork is a workaround for it.
// https://github.com/pytest-dev/pytest-flask/issues/104#issuecomment-577908228
#[cfg(target_os = "macos")]
mp.call_method(
"set_start_method",
("fork",),
// We need to pass `force=True` to prevent `context has already been set` exception,
// see https://github.com/pytorch/pytorch/issues/3492
Some(vec![("force", true)].into_py_dict(py)),
)?;
let address = address.unwrap_or_else(|| String::from("127.0.0.1"));
let port = port.unwrap_or(13734);
let socket = PySocket::new(address, port, backlog)?;
// Lock the workers mutex.
let mut active_workers = self.workers().lock();
// Register the main signal handler.
// TODO(move from num_cpus to thread::available_parallelism after MSRV is 1.60)
// Start all the workers as new Python processes and store the in the `workers` attribute.
for idx in 1..workers.unwrap_or_else(num_cpus::get) + 1 {
let sock = socket.try_clone()?;
let tls = tls.clone();
let process = mp.getattr("Process")?;
let handle = process.call1((
py.None(),
self.clone().into_py(py).getattr(py, "start_worker")?,
format!("smithy-rs-worker[{idx}]"),
(sock.into_py(py), idx, tls.into_py(py)),
))?;
handle.call_method0("start")?;
active_workers.push(handle.to_object(py));
}
// Unlock the workers mutex.
drop(active_workers);
tracing::trace!("rust python server started successfully");
self.block_on_rust_signals();
Ok(())
}
/// Lambda main entrypoint: start the handler on Lambda.
fn run_lambda_handler(&mut self, py: Python) -> PyResult<()> {
use aws_smithy_http_server::routing::LambdaHandler;
let event_loop = self.configure_python_event_loop(py)?;
// Register signals on the Python event loop.
self.register_python_signals(py, event_loop.to_object(py))?;
let service = self.build_and_configure_service(py, event_loop)?;
// Spawn a new background [std::thread] to run the application.
// This is needed because `asyncio` doesn't work properly if it doesn't control the main thread.
// At the end of this function you can see we are calling `event_loop.run_forever()` to
// yield execution of main thread to `asyncio` runtime.
// For more details: https://docs.rs/pyo3-asyncio/latest/pyo3_asyncio/#pythons-event-loop-and-the-main-thread
tracing::trace!("start the tokio runtime in a background task");
thread::spawn(move || {
let rt = runtime::Builder::new_multi_thread()
.enable_all()
.build()
.expect("unable to start a new tokio runtime for this process");
rt.block_on(async move {
let handler = LambdaHandler::new(service);
let lambda = lambda_http::run(handler);
tracing::debug!("starting lambda handler");
if let Err(err) = lambda.await {
tracing::error!(error = %err, "unable to start lambda handler");
}
});
});
// Block on the event loop forever.
tracing::trace!("run and block on the python event loop until a signal is received");
event_loop.call_method0("run_forever")?;
Ok(())
}
// Builds the `Service` and adds necessary layers to it.
fn build_and_configure_service(
&mut self,
py: Python,
event_loop: &pyo3::PyAny,
) -> pyo3::PyResult<Service> {
let service = self.build_service(event_loop)?;
let context = PyContext::new(self.context().clone().unwrap_or_else(|| py.None()))?;
let service = ServiceBuilder::new()
.boxed_clone()
.layer(AddPyContextLayer::new(context))
.service(service);
Ok(service)
}
}
fn addr_incoming_from_socket(socket: Socket) -> AddrIncoming {
let std_listener: StdTcpListener = socket.into();
// StdTcpListener::from_std doesn't set O_NONBLOCK
std_listener
.set_nonblocking(true)
.expect("unable to set `O_NONBLOCK=true` on `std::net::TcpListener`");
let listener = TcpListener::from_std(std_listener)
.expect("unable to create `tokio::net::TcpListener` from `std::net::TcpListener`");
AddrIncoming::from_listener(listener)
.expect("unable to create `AddrIncoming` from `TcpListener`")
}
// Builds `TlsAcceptor` from given `config` and also creates a background task
// to reload certificates and returns a channel to receive new `TlsAcceptor`s.
fn tls_config_reloader(config: PyTlsConfig) -> (TlsAcceptor, mpsc::Receiver<TlsAcceptor>) {
let reload_dur = config.reload_duration();
let (tx, rx) = mpsc::channel();
let acceptor = TlsAcceptor::from(Arc::new(config.build().expect("invalid tls config")));
tokio::spawn(async move {
tracing::trace!(dur = ?reload_dur, "starting timer to reload tls config");
loop {
tokio::time::sleep(reload_dur).await;
tracing::trace!("reloading tls config");
match config.build() {
Ok(config) => {
let new_config = TlsAcceptor::from(Arc::new(config));
// Note on expect: `tx.send` can only fail if the receiver is dropped,
// it probably a bug if that happens
tx.send(new_config).expect("could not send new tls config")
}
Err(err) => {
tracing::error!(error = ?err, "could not reload tls config because it is invalid");
}
}
}
});
(acceptor, rx)
}