api_endpoints_test.rs 49.4 KB
Newer Older
1
2
mod common;

3
4
5
6
7
use axum::{
    body::Body,
    extract::Request,
    http::{header::CONTENT_TYPE, StatusCode},
};
8
9
10
use common::mock_worker::{HealthStatus, MockWorker, MockWorkerConfig, WorkerType};
use reqwest::Client;
use serde_json::json;
11
use sglang_router_rs::config::{PolicyConfig, RetryConfig, RouterConfig, RoutingMode};
12
13
14
use sglang_router_rs::routers::{RouterFactory, RouterTrait};
use std::sync::Arc;
use tower::ServiceExt;
15
16
17
18

/// Test context that manages mock workers
struct TestContext {
    workers: Vec<MockWorker>,
19
20
21
    router: Arc<dyn RouterTrait>,
    client: Client,
    config: RouterConfig,
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
}

impl TestContext {
    async fn new(worker_configs: Vec<MockWorkerConfig>) -> Self {
        // Create default router config
        let config = RouterConfig {
            mode: RoutingMode::Regular {
                worker_urls: vec![],
            },
            policy: PolicyConfig::Random,
            host: "127.0.0.1".to_string(),
            port: 3002,
            max_payload_size: 256 * 1024 * 1024,
            request_timeout_secs: 600,
            worker_startup_timeout_secs: 1,
            worker_startup_check_interval_secs: 1,
38
            discovery: None,
39
40
            dp_aware: false,
            api_key: None,
41
42
43
            metrics: None,
            log_dir: None,
            log_level: None,
44
            request_id_headers: None,
45
46
            max_concurrent_requests: 64,
            cors_allowed_origins: vec![],
47
            retry: RetryConfig::default(),
48
49
50
51
52
        };

        Self::new_with_config(config, worker_configs).await
    }

53
54
55
56
    async fn new_with_config(
        mut config: RouterConfig,
        worker_configs: Vec<MockWorkerConfig>,
    ) -> Self {
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
        let mut workers = Vec::new();
        let mut worker_urls = Vec::new();

        // Start mock workers if any
        for worker_config in worker_configs {
            let mut worker = MockWorker::new(worker_config);
            let url = worker.start().await.unwrap();
            worker_urls.push(url);
            workers.push(worker);
        }

        if !workers.is_empty() {
            tokio::time::sleep(tokio::time::Duration::from_millis(200)).await;
        }

72
73
74
75
76
77
78
79
80
81
        // Update config with worker URLs if not already set
        if let RoutingMode::Regular {
            worker_urls: ref mut urls,
        } = config.mode
        {
            if urls.is_empty() {
                *urls = worker_urls.clone();
            }
        }

82
83
84
85
86
        let client = Client::builder()
            .timeout(std::time::Duration::from_secs(config.request_timeout_secs))
            .build()
            .unwrap();

87
88
        // Create app context
        let app_context = common::create_test_context(config.clone());
89
90
91

        // Create router using sync factory in a blocking context
        let router =
92
            tokio::task::spawn_blocking(move || RouterFactory::create_router(&app_context))
93
94
95
96
                .await
                .unwrap()
                .unwrap();
        let router = Arc::from(router);
97

98
99
        // Wait for router to discover workers
        if !workers.is_empty() {
100
101
102
            tokio::time::sleep(tokio::time::Duration::from_millis(500)).await;
        }

103
104
105
106
107
108
        Self {
            workers,
            router,
            client,
            config,
        }
109
110
    }

111
112
113
114
115
    async fn create_app(&self) -> axum::Router {
        common::test_app::create_test_app(
            Arc::clone(&self.router),
            self.client.clone(),
            &self.config,
116
117
118
119
120
121
122
123
124
125
126
127
128
129
        )
    }

    async fn shutdown(mut self) {
        for worker in &mut self.workers {
            worker.stop().await;
        }
    }
}

#[cfg(test)]
mod health_tests {
    use super::*;

130
131
132
133
    #[tokio::test]
    async fn test_liveness_endpoint() {
        let ctx = TestContext::new(vec![]).await;
        let app = ctx.create_app().await;
134

135
136
137
138
139
        let req = Request::builder()
            .method("GET")
            .uri("/liveness")
            .body(Body::empty())
            .unwrap();
140

141
142
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
143

144
        ctx.shutdown().await;
145
146
    }

147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
    #[tokio::test]
    async fn test_readiness_with_healthy_workers() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18001,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("GET")
            .uri("/readiness")
            .body(Body::empty())
            .unwrap();
165

166
167
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
168

169
        ctx.shutdown().await;
170
171
    }

172
173
174
    #[tokio::test]
    async fn test_readiness_with_unhealthy_workers() {
        let ctx = TestContext::new(vec![]).await;
175

176
        let app = ctx.create_app().await;
177

178
179
180
181
182
        let req = Request::builder()
            .method("GET")
            .uri("/readiness")
            .body(Body::empty())
            .unwrap();
183

184
185
186
        let resp = app.oneshot(req).await.unwrap();
        // With no workers, readiness should return SERVICE_UNAVAILABLE
        assert_eq!(resp.status(), StatusCode::SERVICE_UNAVAILABLE);
187

188
        ctx.shutdown().await;
189
190
    }

191
192
193
194
195
196
197
198
199
200
201
202
    #[tokio::test]
    async fn test_health_endpoint_details() {
        let ctx = TestContext::new(vec![
            MockWorkerConfig {
                port: 18003,
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
            },
            MockWorkerConfig {
                port: 18004,
203
204
205
206
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
207
208
209
            },
        ])
        .await;
210

211
        let app = ctx.create_app().await;
212

213
214
215
216
217
        let req = Request::builder()
            .method("GET")
            .uri("/health")
            .body(Body::empty())
            .unwrap();
218

219
220
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
221

222
223
224
225
226
227
        // The health endpoint returns plain text, not JSON
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_str = String::from_utf8_lossy(&body);
        assert!(body_str.contains("All servers healthy"));
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
        ctx.shutdown().await;
    }

    #[tokio::test]
    async fn test_health_generate_endpoint() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18005,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("GET")
            .uri("/health_generate")
            .body(Body::empty())
            .unwrap();

        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert!(body_json.is_object());

        ctx.shutdown().await;
261
262
263
264
265
266
267
    }
}

#[cfg(test)]
mod generation_tests {
    use super::*;

268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
    #[tokio::test]
    async fn test_generate_success() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18101,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "text": "Hello, world!",
            "stream": false
        });
285

286
287
288
289
290
291
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
292

293
294
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
295

296
297
298
299
300
301
302
303
304
305
306
307
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert!(body_json.get("text").is_some());
        assert!(body_json.get("meta_info").is_some());
        let meta_info = &body_json["meta_info"];
        assert!(meta_info.get("finish_reason").is_some());
        assert_eq!(meta_info["finish_reason"]["type"], "stop");

        ctx.shutdown().await;
    }
308

309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
    #[tokio::test]
    async fn test_generate_streaming() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18102,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "text": "Stream test",
            "stream": true
325
326
        });

327
328
329
330
331
332
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
333

334
335
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
336

337
338
339
340
        // For streaming responses, the router might use chunked encoding or other streaming mechanisms
        // The exact content-type can vary based on the router implementation
        // Just verify we got a successful response
        // Note: In a real implementation, we'd check for text/event-stream or appropriate streaming headers
341

342
343
        ctx.shutdown().await;
    }
344

345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
    #[tokio::test]
    async fn test_generate_with_worker_failure() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18103,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 1.0, // Always fail
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "text": "This should fail",
            "stream": false
        });
362

363
364
365
366
367
368
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
369

370
371
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::INTERNAL_SERVER_ERROR);
372

373
374
        ctx.shutdown().await;
    }
375

376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
    #[tokio::test]
    async fn test_v1_chat_completions_success() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18104,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "model": "test-model",
            "messages": [
                {"role": "user", "content": "Hello!"}
            ],
            "stream": false
        });
396

397
398
399
400
401
402
        let req = Request::builder()
            .method("POST")
            .uri("/v1/chat/completions")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
403

404
405
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
406

407
408
409
410
411
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert!(body_json.get("choices").is_some());
412

413
        ctx.shutdown().await;
414
    }
415
}
416

417
418
419
#[cfg(test)]
mod model_info_tests {
    use super::*;
420

421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
    #[tokio::test]
    async fn test_get_server_info() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18201,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("GET")
            .uri("/get_server_info")
            .body(Body::empty())
            .unwrap();
439

440
441
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
442

443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert!(body_json.is_object());
        // Check for actual sglang server fields
        assert!(body_json.get("version").is_some());
        assert!(body_json.get("model_path").is_some());
        assert!(body_json.get("tokenizer_path").is_some());
        assert!(body_json.get("port").is_some());
        assert!(body_json.get("max_num_batched_tokens").is_some());
        assert!(body_json.get("schedule_policy").is_some());

        ctx.shutdown().await;
    }
458

459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
    #[tokio::test]
    async fn test_get_model_info() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18202,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("GET")
            .uri("/get_model_info")
            .body(Body::empty())
            .unwrap();
477

478
479
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
480

481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert!(body_json.is_object());
        // Check for actual sglang model info fields
        assert_eq!(
            body_json.get("model_path").and_then(|v| v.as_str()),
            Some("mock-model-path")
        );
        assert_eq!(
            body_json.get("tokenizer_path").and_then(|v| v.as_str()),
            Some("mock-tokenizer-path")
        );
        assert_eq!(
            body_json.get("is_generation").and_then(|v| v.as_bool()),
            Some(true)
        );
        assert!(body_json.get("preferred_sampling_params").is_some());

        ctx.shutdown().await;
502
503
    }

504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
    #[tokio::test]
    async fn test_v1_models() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18203,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("GET")
            .uri("/v1/models")
            .body(Body::empty())
            .unwrap();
522

523
524
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
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
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        assert!(body_json.get("object").is_some());
        assert_eq!(
            body_json.get("object").and_then(|v| v.as_str()),
            Some("list")
        );

        let data = body_json.get("data").and_then(|v| v.as_array());
        assert!(data.is_some());

        let models = data.unwrap();
        assert!(!models.is_empty());

        let first_model = &models[0];
        assert_eq!(
            first_model.get("id").and_then(|v| v.as_str()),
            Some("mock-model")
        );
        assert_eq!(
            first_model.get("object").and_then(|v| v.as_str()),
            Some("model")
        );
        assert!(first_model.get("created").is_some());
        assert_eq!(
            first_model.get("owned_by").and_then(|v| v.as_str()),
            Some("organization-owner")
        );

        ctx.shutdown().await;
    }
559

560
561
562
563
    #[tokio::test]
    async fn test_model_info_with_no_workers() {
        let ctx = TestContext::new(vec![]).await;
        let app = ctx.create_app().await;
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
613
614
615
616
        // Test server info with no workers
        let req = Request::builder()
            .method("GET")
            .uri("/get_server_info")
            .body(Body::empty())
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        // Router may return various error codes when no workers
        assert!(
            resp.status() == StatusCode::OK
                || resp.status() == StatusCode::SERVICE_UNAVAILABLE
                || resp.status() == StatusCode::NOT_FOUND
                || resp.status() == StatusCode::INTERNAL_SERVER_ERROR,
            "Unexpected status code: {:?}",
            resp.status()
        );

        // Test model info with no workers
        let req = Request::builder()
            .method("GET")
            .uri("/get_model_info")
            .body(Body::empty())
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        // Router may return various error codes when no workers
        assert!(
            resp.status() == StatusCode::OK
                || resp.status() == StatusCode::SERVICE_UNAVAILABLE
                || resp.status() == StatusCode::NOT_FOUND
                || resp.status() == StatusCode::INTERNAL_SERVER_ERROR,
            "Unexpected status code: {:?}",
            resp.status()
        );

        // Test v1/models with no workers
        let req = Request::builder()
            .method("GET")
            .uri("/v1/models")
            .body(Body::empty())
            .unwrap();
        let resp = app.oneshot(req).await.unwrap();
        // Router may return various error codes when no workers
        assert!(
            resp.status() == StatusCode::OK
                || resp.status() == StatusCode::SERVICE_UNAVAILABLE
                || resp.status() == StatusCode::NOT_FOUND
                || resp.status() == StatusCode::INTERNAL_SERVER_ERROR,
            "Unexpected status code: {:?}",
            resp.status()
        );

        ctx.shutdown().await;
617
618
    }

619
620
621
622
623
624
625
626
627
628
629
630
    #[tokio::test]
    async fn test_model_info_with_multiple_workers() {
        let ctx = TestContext::new(vec![
            MockWorkerConfig {
                port: 18204,
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
            },
            MockWorkerConfig {
                port: 18205,
631
632
633
634
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
635
636
637
            },
        ])
        .await;
638

639
        let app = ctx.create_app().await;
640

641
642
643
644
        // Test that model info is consistent across workers
        for _ in 0..5 {
            let req = Request::builder()
                .method("GET")
645
                .uri("/get_model_info")
646
647
                .body(Body::empty())
                .unwrap();
648

649
            let resp = app.clone().oneshot(req).await.unwrap();
650
651
            assert_eq!(resp.status(), StatusCode::OK);

652
653
654
655
            let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
                .await
                .unwrap();
            let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
656
            assert_eq!(
657
                body_json.get("model_path").and_then(|v| v.as_str()),
658
659
                Some("mock-model-path")
            );
660
        }
661

662
        ctx.shutdown().await;
663
664
    }

665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
    #[tokio::test]
    async fn test_model_info_with_unhealthy_worker() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18206,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 1.0, // Always fail
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("GET")
            .uri("/get_model_info")
            .body(Body::empty())
            .unwrap();
683

684
685
686
687
688
689
690
691
        let resp = app.oneshot(req).await.unwrap();
        // Worker with fail_rate: 1.0 should always return an error status
        assert!(
            resp.status() == StatusCode::INTERNAL_SERVER_ERROR
                || resp.status() == StatusCode::SERVICE_UNAVAILABLE,
            "Expected error status for always-failing worker, got: {:?}",
            resp.status()
        );
692

693
694
695
        ctx.shutdown().await;
    }
}
696

697
698
699
#[cfg(test)]
mod worker_management_tests {
    use super::*;
700

701
702
703
704
705
706
707
708
709
710
711
712
    #[tokio::test]
    async fn test_add_new_worker() {
        let ctx = TestContext::new(vec![]).await;
        let app = ctx.create_app().await;

        // Start a mock worker
        let mut worker = MockWorker::new(MockWorkerConfig {
            port: 18301,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
713
        });
714
        let url = worker.start().await.unwrap();
715

716
717
718
719
720
721
        // Add the worker
        let req = Request::builder()
            .method("POST")
            .uri(&format!("/add_worker?url={}", url))
            .body(Body::empty())
            .unwrap();
722

723
724
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
725

726
727
728
729
730
731
        // List workers to verify
        let req = Request::builder()
            .method("GET")
            .uri("/list_workers")
            .body(Body::empty())
            .unwrap();
732

733
734
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
735

736
737
738
739
740
741
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        let workers = body_json["urls"].as_array().unwrap();
        assert!(workers.iter().any(|w| w.as_str().unwrap() == url));
742

743
744
        worker.stop().await;
        ctx.shutdown().await;
745
746
    }

747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
    #[tokio::test]
    async fn test_remove_existing_worker() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18302,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        // Get the worker URL
        let req = Request::builder()
            .method("GET")
            .uri("/list_workers")
            .body(Body::empty())
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        let workers = body_json["urls"].as_array().unwrap();
        let worker_url = workers[0].as_str().unwrap();

        // Remove the worker
        let req = Request::builder()
            .method("POST")
            .uri(&format!("/remove_worker?url={}", worker_url))
            .body(Body::empty())
            .unwrap();
780

781
782
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
783

784
785
786
787
788
789
790
791
792
793
794
795
796
        // Verify it's removed
        let req = Request::builder()
            .method("GET")
            .uri("/list_workers")
            .body(Body::empty())
            .unwrap();
        let resp = app.oneshot(req).await.unwrap();
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
        let workers = body_json["urls"].as_array().unwrap();
        assert!(workers.is_empty());
797

798
        ctx.shutdown().await;
799
800
    }

801
802
803
804
    #[tokio::test]
    async fn test_add_worker_invalid_url() {
        let ctx = TestContext::new(vec![]).await;
        let app = ctx.create_app().await;
805

806
807
808
809
810
811
        // Invalid URL format
        let req = Request::builder()
            .method("POST")
            .uri("/add_worker?url=not-a-valid-url")
            .body(Body::empty())
            .unwrap();
812

813
814
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::BAD_REQUEST);
815

816
817
818
819
820
821
        // Missing URL parameter
        let req = Request::builder()
            .method("POST")
            .uri("/add_worker")
            .body(Body::empty())
            .unwrap();
822

823
824
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::BAD_REQUEST);
825

826
827
828
829
830
831
        // Empty URL
        let req = Request::builder()
            .method("POST")
            .uri("/add_worker?url=")
            .body(Body::empty())
            .unwrap();
832

833
834
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::BAD_REQUEST);
835

836
        ctx.shutdown().await;
837
838
    }

839
840
841
842
843
844
845
846
847
848
849
    #[tokio::test]
    async fn test_add_duplicate_worker() {
        // Start a mock worker
        let mut worker = MockWorker::new(MockWorkerConfig {
            port: 18303,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        });
        let url = worker.start().await.unwrap();
850

851
852
        let ctx = TestContext::new(vec![]).await;
        let app = ctx.create_app().await;
853

854
855
856
857
858
859
860
861
        // Add worker first time
        let req = Request::builder()
            .method("POST")
            .uri(&format!("/add_worker?url={}", url))
            .body(Body::empty())
            .unwrap();
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
862

863
        tokio::time::sleep(tokio::time::Duration::from_millis(500)).await;
864

865
866
867
868
869
870
871
872
873
        // Try to add same worker again
        let req = Request::builder()
            .method("POST")
            .uri(&format!("/add_worker?url={}", url))
            .body(Body::empty())
            .unwrap();
        let resp = app.oneshot(req).await.unwrap();
        // Should return error for duplicate
        assert_eq!(resp.status(), StatusCode::BAD_REQUEST);
874

875
876
877
        worker.stop().await;
        ctx.shutdown().await;
    }
878

879
880
881
882
883
884
885
886
887
888
889
    #[tokio::test]
    async fn test_add_unhealthy_worker() {
        // Start unhealthy worker
        let mut worker = MockWorker::new(MockWorkerConfig {
            port: 18304,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Unhealthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        });
        let url = worker.start().await.unwrap();
890

891
892
        let ctx = TestContext::new(vec![]).await;
        let app = ctx.create_app().await;
893

894
895
896
897
898
899
900
        // Try to add unhealthy worker
        let req = Request::builder()
            .method("POST")
            .uri(&format!("/add_worker?url={}", url))
            .body(Body::empty())
            .unwrap();
        let resp = app.oneshot(req).await.unwrap();
901

902
903
904
905
906
        // Router should reject unhealthy workers
        assert!(
            resp.status() == StatusCode::BAD_REQUEST
                || resp.status() == StatusCode::SERVICE_UNAVAILABLE
        );
907

908
909
        worker.stop().await;
        ctx.shutdown().await;
910
    }
911
912
913
914
915
}

#[cfg(test)]
mod router_policy_tests {
    use super::*;
916

917
918
919
920
921
    #[tokio::test]
    async fn test_random_policy() {
        let ctx = TestContext::new(vec![
            MockWorkerConfig {
                port: 18801,
922
923
924
925
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
926
927
928
            },
            MockWorkerConfig {
                port: 18802,
929
                worker_type: WorkerType::Regular,
930
                health_status: HealthStatus::Healthy,
931
932
                response_delay_ms: 0,
                fail_rate: 0.0,
933
934
935
936
937
938
939
940
941
942
943
            },
        ])
        .await;

        // Send multiple requests and verify they succeed
        let app = ctx.create_app().await;

        for i in 0..10 {
            let payload = json!({
                "text": format!("Request {}", i),
                "stream": false
944
945
            });

946
947
948
949
950
951
            let req = Request::builder()
                .method("POST")
                .uri("/generate")
                .header(CONTENT_TYPE, "application/json")
                .body(Body::from(serde_json::to_string(&payload).unwrap()))
                .unwrap();
952

953
954
955
            let resp = app.clone().oneshot(req).await.unwrap();
            assert_eq!(resp.status(), StatusCode::OK);
        }
956

957
958
        ctx.shutdown().await;
    }
959

960
961
962
963
964
965
966
967
968
969
970
971
972
973
    #[tokio::test]
    async fn test_worker_selection() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18203,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let _payload = json!({
            "text": "Test selection",
            "stream": false
974
        });
975
976
977
978
979
980
981

        // Check that router has the worker
        let worker_urls = ctx.router.get_worker_urls();
        assert_eq!(worker_urls.len(), 1);
        assert!(worker_urls[0].contains("18203"));

        ctx.shutdown().await;
982
983
984
985
986
987
988
    }
}

#[cfg(test)]
mod error_tests {
    use super::*;

989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
    #[tokio::test]
    async fn test_404_not_found() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18401,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        // Test unknown endpoint
        let req = Request::builder()
            .method("GET")
            .uri("/unknown_endpoint")
            .body(Body::empty())
            .unwrap();
1008

1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::NOT_FOUND);

        // Test POST to unknown endpoint
        let req = Request::builder()
            .method("POST")
            .uri("/api/v2/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(
                serde_json::to_string(&json!({"text": "test"})).unwrap(),
            ))
            .unwrap();
1021

1022
1023
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::NOT_FOUND);
1024

1025
1026
        ctx.shutdown().await;
    }
1027

1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
    #[tokio::test]
    async fn test_method_not_allowed() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18402,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        // GET request to POST-only endpoint
        let req = Request::builder()
            .method("GET")
            .uri("/generate")
            .body(Body::empty())
            .unwrap();
1047

1048
1049
1050
        let resp = app.clone().oneshot(req).await.unwrap();
        // Note: Axum returns 405 for wrong methods on matched routes
        assert_eq!(resp.status(), StatusCode::METHOD_NOT_ALLOWED);
1051

1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
        // POST request to GET-only endpoint
        let req = Request::builder()
            .method("POST")
            .uri("/health")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from("{}"))
            .unwrap();

        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::METHOD_NOT_ALLOWED);

        ctx.shutdown().await;
1064
1065
    }

1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
    #[tokio::test]
    async fn test_payload_too_large() {
        // Create context with small payload limit
        let config = RouterConfig {
            mode: RoutingMode::Regular {
                worker_urls: vec![],
            },
            policy: PolicyConfig::Random,
            host: "127.0.0.1".to_string(),
            port: 3010,
            max_payload_size: 1024, // 1KB limit
            request_timeout_secs: 600,
            worker_startup_timeout_secs: 1,
            worker_startup_check_interval_secs: 1,
            dp_aware: false,
            api_key: None,
            discovery: None,
            metrics: None,
            log_dir: None,
            log_level: None,
            request_id_headers: None,
            max_concurrent_requests: 64,
            cors_allowed_origins: vec![],
1089
            retry: RetryConfig::default(),
1090
1091
1092
1093
1094
1095
        };

        let ctx = TestContext::new_with_config(
            config,
            vec![MockWorkerConfig {
                port: 18403,
1096
1097
1098
1099
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
1100
1101
1102
            }],
        )
        .await;
1103

1104
1105
1106
        // Note: The server would have payload size middleware configured
        // but we cannot test it directly through the test app
        // This test is kept for documentation purposes
1107

1108
1109
        ctx.shutdown().await;
    }
1110

1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
    #[tokio::test]
    async fn test_invalid_json_payload() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18404,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        // Send invalid JSON
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from("{invalid json}"))
            .unwrap();
1131

1132
1133
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::BAD_REQUEST);
1134

1135
1136
1137
1138
1139
1140
1141
        // Send empty body
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::empty())
            .unwrap();
1142

1143
1144
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::BAD_REQUEST);
1145

1146
1147
        ctx.shutdown().await;
    }
1148

1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
    #[tokio::test]
    async fn test_missing_required_fields() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18405,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        // Missing messages in chat completion
        let payload = json!({
            "model": "test-model"
            // missing "messages"
1166
        });
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179

        let req = Request::builder()
            .method("POST")
            .uri("/v1/chat/completions")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.oneshot(req).await.unwrap();
        // Axum validates JSON schema - returns 422 for validation errors
        assert_eq!(resp.status(), StatusCode::UNPROCESSABLE_ENTITY);

        ctx.shutdown().await;
1180
1181
    }

1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
    #[tokio::test]
    async fn test_invalid_model() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18406,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let payload = json!({
            "model": "invalid-model-name-that-does-not-exist",
            "messages": [{"role": "user", "content": "Hello"}],
            "stream": false
        });
1200

1201
1202
1203
1204
1205
1206
        let req = Request::builder()
            .method("POST")
            .uri("/v1/chat/completions")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
1207

1208
1209
1210
        let resp = app.oneshot(req).await.unwrap();
        // Mock worker accepts any model, but real implementation might return 400
        assert!(resp.status().is_success() || resp.status() == StatusCode::BAD_REQUEST);
1211

1212
1213
1214
        ctx.shutdown().await;
    }
}
1215

1216
1217
1218
#[cfg(test)]
mod cache_tests {
    use super::*;
1219

1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
    #[tokio::test]
    async fn test_flush_cache() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18501,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        let req = Request::builder()
            .method("POST")
            .uri("/flush_cache")
            .body(Body::empty())
            .unwrap();
1238

1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        // The response might be empty or contain a message
        let body_bytes = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        if !body_bytes.is_empty() {
            if let Ok(body) = serde_json::from_slice::<serde_json::Value>(&body_bytes) {
                // Check that we got a successful response with expected fields
                assert!(body.is_object());
                assert!(body.get("message").is_some() || body.get("status").is_some());
            }
        }

        ctx.shutdown().await;
1255
1256
    }

1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
    #[tokio::test]
    async fn test_get_loads() {
        let ctx = TestContext::new(vec![
            MockWorkerConfig {
                port: 18502,
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
            },
            MockWorkerConfig {
                port: 18503,
1269
1270
1271
1272
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
1273
1274
1275
            },
        ])
        .await;
1276

1277
        let app = ctx.create_app().await;
1278

1279
1280
1281
1282
1283
        let req = Request::builder()
            .method("GET")
            .uri("/get_loads")
            .body(Body::empty())
            .unwrap();
1284

1285
1286
        let resp = app.oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
1287

1288
1289
1290
1291
        let body = axum::body::to_bytes(resp.into_body(), usize::MAX)
            .await
            .unwrap();
        let body_json: serde_json::Value = serde_json::from_slice(&body).unwrap();
1292

1293
1294
1295
1296
        // Verify the response contains load information
        assert!(body_json.is_object());
        // The exact structure depends on the implementation
        // but should contain worker load information
1297

1298
1299
        ctx.shutdown().await;
    }
1300

1301
1302
1303
    #[tokio::test]
    async fn test_flush_cache_no_workers() {
        let ctx = TestContext::new(vec![]).await;
1304

1305
        let app = ctx.create_app().await;
1306

1307
1308
1309
1310
1311
        let req = Request::builder()
            .method("POST")
            .uri("/flush_cache")
            .body(Body::empty())
            .unwrap();
1312

1313
1314
1315
1316
1317
        let resp = app.oneshot(req).await.unwrap();
        // Should either succeed (no-op) or return service unavailable
        assert!(
            resp.status() == StatusCode::OK || resp.status() == StatusCode::SERVICE_UNAVAILABLE
        );
1318

1319
        ctx.shutdown().await;
1320
1321
1322
1323
    }
}

#[cfg(test)]
1324
mod load_balancing_tests {
1325
1326
    use super::*;

1327
1328
1329
1330
1331
1332
    #[tokio::test]
    async fn test_request_distribution() {
        // Create multiple workers
        let ctx = TestContext::new(vec![
            MockWorkerConfig {
                port: 18601,
1333
1334
1335
1336
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
            },
            MockWorkerConfig {
                port: 18602,
                worker_type: WorkerType::Regular,
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
            },
        ])
        .await;
1347

1348
        let app = ctx.create_app().await;
1349

1350
1351
1352
1353
1354
1355
1356
        // Send multiple requests and track distribution
        let mut request_count = 0;
        for i in 0..10 {
            let payload = json!({
                "text": format!("Request {}", i),
                "stream": false
            });
1357

1358
1359
1360
1361
1362
1363
            let req = Request::builder()
                .method("POST")
                .uri("/generate")
                .header(CONTENT_TYPE, "application/json")
                .body(Body::from(serde_json::to_string(&payload).unwrap()))
                .unwrap();
1364

1365
1366
1367
            let resp = app.clone().oneshot(req).await.unwrap();
            if resp.status() == StatusCode::OK {
                request_count += 1;
1368
            }
1369
        }
1370

1371
1372
        // With random policy, all requests should succeed
        assert_eq!(request_count, 10);
1373

1374
1375
1376
        ctx.shutdown().await;
    }
}
1377

1378
1379
1380
#[cfg(test)]
mod pd_mode_tests {
    use super::*;
1381

1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
    #[tokio::test]
    async fn test_pd_mode_routing() {
        // Create PD mode configuration with prefill and decode workers
        let mut prefill_worker = MockWorker::new(MockWorkerConfig {
            port: 18701,
            worker_type: WorkerType::Prefill,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        });
1392

1393
1394
1395
1396
1397
1398
        let mut decode_worker = MockWorker::new(MockWorkerConfig {
            port: 18702,
            worker_type: WorkerType::Decode,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
1399
1400
        });

1401
1402
        let prefill_url = prefill_worker.start().await.unwrap();
        let decode_url = decode_worker.start().await.unwrap();
1403

1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
        tokio::time::sleep(tokio::time::Duration::from_millis(200)).await;

        // Extract port from prefill URL
        let prefill_port = prefill_url
            .split(':')
            .last()
            .and_then(|p| p.trim_end_matches('/').parse::<u16>().ok())
            .unwrap_or(9000);

        let config = RouterConfig {
            mode: RoutingMode::PrefillDecode {
                prefill_urls: vec![(prefill_url, Some(prefill_port))],
                decode_urls: vec![decode_url],
                prefill_policy: None,
                decode_policy: None,
            },
            policy: PolicyConfig::Random,
            host: "127.0.0.1".to_string(),
            port: 3011,
            max_payload_size: 256 * 1024 * 1024,
            request_timeout_secs: 600,
            worker_startup_timeout_secs: 1,
            worker_startup_check_interval_secs: 1,
            discovery: None,
            metrics: None,
            log_dir: None,
            dp_aware: false,
            api_key: None,
            log_level: None,
            request_id_headers: None,
            max_concurrent_requests: 64,
            cors_allowed_origins: vec![],
1436
            retry: RetryConfig::default(),
1437
1438
        };

1439
1440
1441
        // Create app context
        let app_context = common::create_test_context(config);

1442
1443
        // Create router - this might fail due to health check issues
        let router_result =
1444
            tokio::task::spawn_blocking(move || RouterFactory::create_router(&app_context))
1445
1446
1447
1448
1449
1450
1451
1452
1453
                .await
                .unwrap();

        // Clean up workers
        prefill_worker.stop().await;
        decode_worker.stop().await;

        // For now, just verify the configuration was attempted
        assert!(router_result.is_err() || router_result.is_ok());
1454
1455
1456
1457
    }
}

#[cfg(test)]
1458
mod request_id_tests {
1459
1460
    use super::*;

1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
    #[tokio::test]
    async fn test_request_id_generation() {
        let ctx = TestContext::new(vec![MockWorkerConfig {
            port: 18901,
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
        }])
        .await;

        let app = ctx.create_app().await;

        // Test 1: Request without any request ID header should generate one
        let payload = json!({
            "text": "Test request",
            "stream": false
        });
1479

1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        // Check that response has x-request-id header
        let request_id = resp.headers().get("x-request-id");
        assert!(
            request_id.is_some(),
            "Response should have x-request-id header"
        );

        let id_value = request_id.unwrap().to_str().unwrap();
        assert!(
            id_value.starts_with("gnt-"),
            "Generate endpoint should have gnt- prefix"
        );
        assert!(
            id_value.len() > 4,
            "Request ID should have content after prefix"
        );

        // Test 2: Request with custom x-request-id should preserve it
        let custom_id = "custom-request-id-123";
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .header("x-request-id", custom_id)
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);
1519

1520
1521
1522
1523
1524
1525
1526
1527
        let response_id = resp.headers().get("x-request-id");
        assert!(response_id.is_some());
        assert_eq!(response_id.unwrap(), custom_id);

        // Test 3: Different endpoints should have different prefixes
        let chat_payload = json!({
            "messages": [{"role": "user", "content": "Hello"}],
            "model": "test-model"
1528
        });
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564

        let req = Request::builder()
            .method("POST")
            .uri("/v1/chat/completions")
            .header(CONTENT_TYPE, "application/json")
            .body(Body::from(serde_json::to_string(&chat_payload).unwrap()))
            .unwrap();

        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        let request_id = resp.headers().get("x-request-id");
        assert!(request_id.is_some());
        assert!(request_id
            .unwrap()
            .to_str()
            .unwrap()
            .starts_with("chatcmpl-"));

        // Test 4: Alternative request ID headers should be recognized
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .header("x-correlation-id", "correlation-123")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();

        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        let response_id = resp.headers().get("x-request-id");
        assert!(response_id.is_some());
        assert_eq!(response_id.unwrap(), "correlation-123");

        ctx.shutdown().await;
1565
1566
    }

1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
    #[tokio::test]
    async fn test_request_id_with_custom_headers() {
        // Create config with custom request ID headers
        let config = RouterConfig {
            mode: RoutingMode::Regular {
                worker_urls: vec![],
            },
            policy: PolicyConfig::Random,
            host: "127.0.0.1".to_string(),
            port: 3002,
            max_payload_size: 256 * 1024 * 1024,
            request_timeout_secs: 600,
            worker_startup_timeout_secs: 1,
            worker_startup_check_interval_secs: 1,
            discovery: None,
            metrics: None,
            dp_aware: false,
            api_key: None,
            log_dir: None,
            log_level: None,
            request_id_headers: Some(vec!["custom-id".to_string(), "trace-id".to_string()]),
            max_concurrent_requests: 64,
            cors_allowed_origins: vec![],
1590
            retry: RetryConfig::default(),
1591
        };
1592

1593
1594
1595
1596
1597
        let ctx = TestContext::new_with_config(
            config,
            vec![MockWorkerConfig {
                port: 18902,
                worker_type: WorkerType::Regular,
1598
1599
1600
                health_status: HealthStatus::Healthy,
                response_delay_ms: 0,
                fail_rate: 0.0,
1601
1602
1603
            }],
        )
        .await;
1604

1605
        let app = ctx.create_app().await;
1606

1607
1608
1609
1610
        let payload = json!({
            "text": "Test request",
            "stream": false
        });
1611

1612
1613
1614
1615
1616
1617
1618
1619
        // Test custom header is recognized
        let req = Request::builder()
            .method("POST")
            .uri("/generate")
            .header(CONTENT_TYPE, "application/json")
            .header("custom-id", "my-custom-id")
            .body(Body::from(serde_json::to_string(&payload).unwrap()))
            .unwrap();
1620

1621
1622
1623
1624
1625
1626
1627
1628
        let resp = app.clone().oneshot(req).await.unwrap();
        assert_eq!(resp.status(), StatusCode::OK);

        let response_id = resp.headers().get("x-request-id");
        assert!(response_id.is_some());
        assert_eq!(response_id.unwrap(), "my-custom-id");

        ctx.shutdown().await;
1629
1630
    }
}