rama-http-backend 0.3.0

error types and utilities for rama
Documentation 
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
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612

//! HTTP connection pool: connection identity and connector assembly.

use std::time::Duration;

use rama_core::Layer;
use rama_core::error::{BoxError, BoxErrorExt as _};
use rama_core::extensions::ExtensionsRef;
use rama_http_types::Request;
use rama_net::address::{HostWithOptPort, ProxyAddress};
use rama_net::client::pool::{ConnID, MultiplexPool, MuxSelection, PooledConnector, ReqToConnID};
use rama_net::client::{ConnectorService, ConnectorTarget};
use rama_net::{AuthorityInputExt, Protocol, ProtocolInputExt};

use super::{BindBodyToConnLayer, BindBodyToConnector};

#[derive(Clone, Debug, Default)]
#[non_exhaustive]
/// [`BasicHttpConnIdentifier`] can be used together with a [`Pool`](rama_net::client::pool::Pool)
/// to create a basic http connection pool.
pub struct BasicHttpConnIdentifier;

#[derive(Clone, Debug, PartialEq, Eq)]
/// Connection Identifier which will match inputs that have the exact same
/// protocol, authority, proxy address and connector target
pub struct BasicHttpConId {
    pub protocol: Option<Protocol>,
    pub authority: HostWithOptPort,
    pub proxy_address: Option<ProxyAddress>,
    pub connector_target: Option<ConnectorTarget>,
}

impl ConnID for BasicHttpConId {
    #[cfg(feature = "opentelemetry")]
    fn attributes(&self) -> impl Iterator<Item = rama_core::telemetry::opentelemetry::KeyValue> {
        self.protocol
            .as_ref()
            .map(|protocol| {
                rama_core::telemetry::opentelemetry::KeyValue::new("protocol", protocol.to_string())
            })
            .into_iter()
            .chain([rama_core::telemetry::opentelemetry::KeyValue::new(
                "authority",
                self.authority.to_string(),
            )])
    }
}

impl<Body> ReqToConnID<Request<Body>> for BasicHttpConnIdentifier {
    type ID = BasicHttpConId;

    fn id(&self, req: &Request<Body>) -> Result<Self::ID, BoxError> {
        let authority = req
            .authority()
            .ok_or_else(|| BoxError::from_static_str("no authority found in http request"))?;
        let protocol = req.protocol().cloned();

        Ok(BasicHttpConId {
            protocol,
            authority,
            proxy_address: req.extensions().get_ref().cloned(),
            connector_target: req.extensions().get_ref().cloned(),
        })
    }
}

#[derive(Debug, Clone)]
/// Config used to create a multiplexing http connection pool ([`MultiplexPool`]).
///
/// The per-connection concurrency comes from the connection's
/// [`MaxConcurrency`](rama_net::conn::MaxConcurrency) extension (set by the http
/// connectors: 1 for http/1, the stream capacity for http/2), clamped to
/// `max_concurrent_streams` as an upper bound.
pub struct HttpPooledConnectorConfig {
    /// Set the max amount of connections that this connection pool will contain
    ///
    /// This is the sum of active connections and idle connections. When this limit
    /// is hit idle connections will be replaced with new ones.
    pub max_total: usize,
    /// Upper bound on the concurrent requests a single connection may serve.
    ///
    /// Acts as a ceiling for each connection, each connection also figures
    /// it's own max concurrency out by itself
    pub max_concurrent_streams: usize,
    /// How a connection is chosen among several that can serve a request.
    pub selection: MuxSelection,
    /// If connections have been idle (no active streams) for longer than this
    /// timeout they are dropped. Only checked when a connection is requested.
    pub idle_timeout: Option<Duration>,
    /// How long to wait for the pool to hand out a connection before timing out.
    pub wait_for_pool_timeout: Option<Duration>,
}

impl Default for HttpPooledConnectorConfig {
    fn default() -> Self {
        Self {
            max_total: 50,
            max_concurrent_streams: 100,
            selection: MuxSelection::default(),
            idle_timeout: Some(Duration::from_secs(300)),
            wait_for_pool_timeout: Some(Duration::from_secs(120)),
        }
    }
}

impl HttpPooledConnectorConfig {
    /// Build a pooled http connector around `inner`.
    ///
    /// The returned connector wraps each pooled connection in
    /// [`BindBodyToConn`](super::BindBodyToConn), so the pool only frees/reuses a
    /// connection once its response body has been consumed, not at response
    /// headers.
    ///
    /// Warning: the connection returned by this pool should only be used for a single
    /// request. Every request should go through the connector stack again, and will
    /// receive a new or resused connection (maybe multiplexed) of its own from.
    pub fn build_connector<S>(
        self,
        inner: S,
    ) -> Result<
        BindBodyToConnector<
            PooledConnector<
                S,
                MultiplexPool<S::Connection, BasicHttpConId>,
                BasicHttpConnIdentifier,
            >,
        >,
        BoxError,
    >
    where
        S: ConnectorService<Request>,
    {
        let pool = MultiplexPool::try_new(self.max_concurrent_streams, self.max_total)?
            .with_selection(self.selection)
            .maybe_with_idle_timeout(self.idle_timeout);

        let connector = PooledConnector::new(inner, pool, BasicHttpConnIdentifier)
            .maybe_with_wait_for_pool_timeout(self.wait_for_pool_timeout);

        Ok(BindBodyToConnLayer::new().into_layer(connector))
    }
}

#[cfg(test)]
mod tests {
    use std::convert::Infallible;
    use std::sync::Arc;
    use std::sync::atomic::{AtomicUsize, Ordering};
    use std::time::Duration;

    use rama_core::error::BoxError;
    use rama_core::extensions::ExtensionsRef;
    use rama_core::rt::Executor;
    use rama_core::service::service_fn;
    use rama_core::{Layer, Service};
    use rama_http_types::body::util::BodyExt as _;
    use rama_http_types::{Body, HeaderValue, Request, Response, Version};
    use rama_net::client::ConnectorService;
    use rama_net::test_utils::client::MockConnectorService;
    use rama_utils::octets::kib;
    use tokio::time::sleep;

    use super::HttpPooledConnectorConfig;
    use crate::client::HttpConnectorLayer;
    use crate::server::HttpServer;

    fn create_test_request(version: Version) -> Request {
        Request::builder()
            .uri("https://www.example.com")
            .version(version)
            .body(Body::from("a random request body"))
            .unwrap()
    }

    /// A mock connector whose every backend connection runs an `HttpServer` that
    /// tags each response with `x-conn-id` (which backend connection served it) and
    /// `x-resp-id` (how many requests that connection has served so far). The
    /// per-connection id is read from a header, so it can be asserted without
    /// draining the (possibly still in-flight) response body.
    fn tagging_mock_connector() -> impl ConnectorService<
        Request,
        Connection: Service<Request, Output = Response, Error = BoxError> + ExtensionsRef,
    > {
        let conns = Arc::new(AtomicUsize::new(0));
        HttpConnectorLayer::default().into_layer(MockConnectorService::new(move || {
            let conn_id = conns.fetch_add(1, Ordering::Relaxed);
            let resps = Arc::new(AtomicUsize::new(0));
            HttpServer::auto(Executor::default()).service(service_fn(move |_req: Request| {
                let resps = resps.clone();
                async move {
                    let resp_id = resps.fetch_add(1, Ordering::Relaxed);
                    let mut resp = Response::new(Body::from("ok"));
                    let headers = resp.headers_mut();
                    headers.insert("x-conn-id", HeaderValue::from(conn_id as u64));
                    headers.insert("x-resp-id", HeaderValue::from(resp_id as u64));
                    Ok::<_, Infallible>(resp)
                }
            }))
        }))
    }

    fn conn_id(resp: &Response) -> u64 {
        resp.headers()
            .get("x-conn-id")
            .unwrap()
            .to_str()
            .unwrap()
            .parse()
            .unwrap()
    }

    #[tokio::test]
    async fn pool_keeps_h2_connection_in_use_until_response_body_consumed() {
        let connector = HttpPooledConnectorConfig {
            max_concurrent_streams: 1,
            max_total: 4,
            ..Default::default()
        }
        .build_connector(tagging_mock_connector())
        .unwrap();

        let req = || create_test_request(Version::HTTP_2);

        // Serve req1 but hold its (still-streaming) response body: the connection is
        // moved into that body, so it is logically still in use.
        let est1 = connector.serve(req()).await.unwrap();
        let resp1 = est1.conn.serve(req()).await.unwrap();
        drop(est1);

        // req2 must therefore land on a NEW connection (cap = 1, conn 0 busy).
        let est2 = connector.serve(req()).await.unwrap();
        let resp2 = est2.conn.serve(req()).await.unwrap();

        assert_eq!(conn_id(&resp1), 0);
        assert_eq!(
            conn_id(&resp2),
            1,
            "second request must not reuse a connection whose response body is still in flight"
        );
    }

    #[tokio::test]
    async fn pool_keeps_h1_connection_in_use_until_response_body_consumed() {
        let connector = HttpPooledConnectorConfig {
            max_concurrent_streams: 1,
            max_total: 4,
            ..Default::default()
        }
        .build_connector(tagging_mock_connector())
        .unwrap();

        let req = || create_test_request(Version::HTTP_11);

        let est1 = connector.serve(req()).await.unwrap();
        let resp1 = est1.conn.serve(req()).await.unwrap();
        drop(est1);

        let est2 = connector.serve(req()).await.unwrap();
        let resp2 = est2.conn.serve(req()).await.unwrap();

        assert_eq!(conn_id(&resp1), 0);
        assert_eq!(
            conn_id(&resp2),
            1,
            "h1: second request must not reuse a connection whose response body is still in flight"
        );
    }

    /// An h1 connection the server closed (`Connection: close`) must not be handed
    /// back out by the pool. Multiplex storage keeps the connection after a handout
    /// drops, so this relies on `ConnectionHealthWatcher` being marked broken and
    /// swept, the multiplex equivalent of the exclusive pool's `drop_connection_if_no_response`.
    #[tokio::test(start_paused = true)]
    async fn pool_does_not_reuse_h1_connection_after_server_close() {
        let conns = Arc::new(AtomicUsize::new(0));
        let inner =
            HttpConnectorLayer::default().into_layer(MockConnectorService::new(move || {
                let conn_id = conns.fetch_add(1, Ordering::Relaxed);
                HttpServer::auto(Executor::default()).service(service_fn(
                    move |_req: Request| async move {
                        let mut resp = Response::new(Body::from("ok"));
                        let headers = resp.headers_mut();
                        headers.insert("x-conn-id", HeaderValue::from(conn_id as u64));
                        headers.insert("connection", HeaderValue::from_static("close"));
                        Ok::<_, Infallible>(resp)
                    },
                ))
            }));
        let connector = HttpPooledConnectorConfig {
            max_total: 4,
            ..Default::default()
        }
        .build_connector(inner)
        .unwrap();

        let req = || create_test_request(Version::HTTP_11);

        // Serve and fully drain req1 so the h1 connection processes the close and
        // (should) get marked broken before it could be reused.
        let est1 = connector.serve(req()).await.unwrap();
        let resp1 = est1.conn.serve(req()).await.unwrap();
        drop(est1);
        let id1 = conn_id(&resp1);
        resp1.into_body().collect().await.unwrap();

        // Let the connection task observe the close and update health.
        sleep(Duration::from_millis(50)).await;

        let est2 = connector.serve(req()).await.unwrap();
        let resp2 = est2.conn.serve(req()).await.unwrap();

        assert_eq!(id1, 0);
        assert_ne!(
            conn_id(&resp2),
            id1,
            "must not reuse an h1 connection the server closed"
        );
    }

    /// An h1 response body abandoned before end-of-stream leaves the connection
    /// mid-message, the pool must not reuse it. Uses a large body so the response is
    /// genuinely still on the wire when dropped (a tiny fully-buffered body could be
    /// drained and legitimately reused).
    #[tokio::test(start_paused = true)]
    async fn pool_does_not_reuse_h1_connection_after_body_dropped_early() {
        let conns = Arc::new(AtomicUsize::new(0));
        let inner =
            HttpConnectorLayer::default().into_layer(MockConnectorService::new(move || {
                let conn_id = conns.fetch_add(1, Ordering::Relaxed);
                HttpServer::auto(Executor::default()).service(service_fn(
                    move |_req: Request| async move {
                        let mut resp = Response::new(Body::from(vec![0u8; kib(1024)]));
                        resp.headers_mut()
                            .insert("x-conn-id", HeaderValue::from(conn_id as u64));
                        Ok::<_, Infallible>(resp)
                    },
                ))
            }));
        let connector = HttpPooledConnectorConfig {
            max_total: 4,
            ..Default::default()
        }
        .build_connector(inner)
        .unwrap();

        let req = || create_test_request(Version::HTTP_11);

        // Take the response (headers) but drop it without reading the body.
        let est1 = connector.serve(req()).await.unwrap();
        let resp1 = est1.conn.serve(req()).await.unwrap();
        drop(est1);
        let id1 = conn_id(&resp1);
        drop(resp1);

        // Let the connection task observe the abandoned read and update health.
        sleep(Duration::from_millis(50)).await;

        let est2 = connector.serve(req()).await.unwrap();
        let resp2 = est2.conn.serve(req()).await.unwrap();

        assert_eq!(id1, 0);
        assert_ne!(
            conn_id(&resp2),
            id1,
            "must not reuse an h1 connection whose response body was abandoned mid-stream"
        );
    }

    #[tokio::test]
    async fn pool_reuses_connection_after_body_consumed() {
        let connector = HttpPooledConnectorConfig {
            max_concurrent_streams: 1,
            max_total: 4,
            ..Default::default()
        }
        .build_connector(tagging_mock_connector())
        .unwrap();

        let req = || create_test_request(Version::HTTP_2);

        let resp1 = connector
            .serve(req())
            .await
            .unwrap()
            .conn
            .serve(req())
            .await
            .unwrap();
        assert_eq!(conn_id(&resp1), 0);
        // Drain to end-of-stream: releases connection 0 back to the pool.
        resp1.into_body().collect().await.unwrap();

        let resp2 = connector
            .serve(req())
            .await
            .unwrap()
            .conn
            .serve(req())
            .await
            .unwrap();
        assert_eq!(
            conn_id(&resp2),
            0,
            "connection must be reused once its response body is consumed"
        );
    }

    #[tokio::test]
    async fn pool_multiplexes_on_h2() {
        let connector = HttpPooledConnectorConfig::default()
            .build_connector(tagging_mock_connector())
            .unwrap();

        let req = || create_test_request(Version::HTTP_2);

        // Hold resp1 (body unconsumed) so connection 0 stays in use.
        let resp1 = connector
            .serve(req())
            .await
            .unwrap()
            .conn
            .serve(req())
            .await
            .unwrap();
        let resp2 = connector
            .serve(req())
            .await
            .unwrap()
            .conn
            .serve(req())
            .await
            .unwrap();

        assert_eq!(conn_id(&resp1), 0);
        assert_eq!(
            conn_id(&resp2),
            0,
            "h2: a second in-flight request multiplexes onto the same connection"
        );
    }

    #[tokio::test]
    async fn pool_does_not_multiplex_on_h1() {
        let connector = HttpPooledConnectorConfig::default()
            .build_connector(tagging_mock_connector())
            .unwrap();

        let req = || create_test_request(Version::HTTP_11);

        let resp1 = connector
            .serve(req())
            .await
            .unwrap()
            .conn
            .serve(req())
            .await
            .unwrap();
        let resp2 = connector
            .serve(req())
            .await
            .unwrap()
            .conn
            .serve(req())
            .await
            .unwrap();

        assert_eq!(conn_id(&resp1), 0);
        assert_eq!(
            conn_id(&resp2),
            1,
            "h1 does not multiplex: a second in-flight request needs a new connection"
        );
    }

    #[tokio::test]
    async fn pool_respects_max_concurrent_streams() {
        let connector = HttpPooledConnectorConfig {
            max_concurrent_streams: 2,
            max_total: 4,
            ..Default::default()
        }
        .build_connector(tagging_mock_connector())
        .unwrap();

        let req = || create_test_request(Version::HTTP_2);

        // Three in-flight (bodies unconsumed) requests: two fit on connection 0.
        let resp1 = connector
            .serve(req())
            .await
            .unwrap()
            .conn
            .serve(req())
            .await
            .unwrap();
        let resp2 = connector
            .serve(req())
            .await
            .unwrap()
            .conn
            .serve(req())
            .await
            .unwrap();
        let resp3 = connector
            .serve(req())
            .await
            .unwrap()
            .conn
            .serve(req())
            .await
            .unwrap();

        assert_eq!(conn_id(&resp1), 0);
        assert_eq!(conn_id(&resp2), 0, "second request fits on connection 0");
        assert_eq!(
            conn_id(&resp3),
            1,
            "third request exceeds the per-connection limit, needs a new connection"
        );
    }

    /// Like [`tagging_mock_connector`] but each response carries a large body, so
    /// it genuinely streams over multiple frames rather than a single buffered one.
    fn large_body_mock_connector() -> impl ConnectorService<
        Request,
        Connection: Service<Request, Output = Response, Error = BoxError> + ExtensionsRef,
    > {
        let conns = Arc::new(AtomicUsize::new(0));
        HttpConnectorLayer::default().into_layer(MockConnectorService::new(move || {
            let conn_id = conns.fetch_add(1, Ordering::Relaxed);
            HttpServer::auto(Executor::default()).service(service_fn(
                move |_req: Request| async move {
                    let mut resp = Response::new(Body::from(vec![0u8; kib(1024)]));
                    resp.headers_mut()
                        .insert("x-conn-id", HeaderValue::from(conn_id as u64));
                    Ok::<_, Infallible>(resp)
                },
            ))
        }))
    }

    /// A connection stays bound for the whole of a *streaming* response body: it
    /// is not reusable while frames are still arriving, and is released precisely
    /// at end-of-stream (exercising `GuardedBody`'s `poll_frame -> Ready(None)`
    /// over real multi-frame h2 streaming).
    #[tokio::test]
    async fn pool_binds_connection_across_streaming_body() {
        let connector = HttpPooledConnectorConfig {
            max_concurrent_streams: 1,
            max_total: 4,
            ..Default::default()
        }
        .build_connector(large_body_mock_connector())
        .unwrap();

        let req = || create_test_request(Version::HTTP_2);

        // Read the headers and a single body frame: the stream is not yet at
        // end-of-stream, so connection 0 is still in use.
        let resp1 = connector
            .serve(req())
            .await
            .unwrap()
            .conn
            .serve(req())
            .await
            .unwrap();
        let id1 = conn_id(&resp1);
        let mut body1 = resp1.into_body();
        assert!(
            body1.frame().await.is_some(),
            "streaming body should yield at least one frame"
        );

        // Mid-stream: connection 0 must not be reused.
        let resp2 = connector
            .serve(req())
            .await
            .unwrap()
            .conn
            .serve(req())
            .await
            .unwrap();
        assert_eq!(id1, 0);
        assert_ne!(
            conn_id(&resp2),
            id1,
            "a connection still streaming its response body must not be reused"
        );

        // Drain body 1 to end-of-stream. `GuardedBody` releases the connection
        // right here (at `poll_frame -> Ready(None)`), not on drop.
        while let Some(frame) = body1.frame().await {
            frame.unwrap();
        }

        // `body1` is still in scope (not dropped), yet connection 0 is reused.
        // Proving the release happens at end-of-stream, not when the body drops.
        let resp3 = connector
            .serve(req())
            .await
            .unwrap()
            .conn
            .serve(req())
            .await
            .unwrap();
        assert_eq!(
            conn_id(&resp3),
            id1,
            "a connection is reused once its streaming body reaches end-of-stream"
        );
    }
}