mock_worker.rs 21.9 KB
Newer Older
1
2
3
// Mock worker for testing - these functions are used by integration tests
#![allow(dead_code)]

4
5
6
7
8
9
10
11
12
use axum::{
    extract::{Json, State},
    http::StatusCode,
    response::sse::{Event, KeepAlive},
    response::{IntoResponse, Response, Sse},
    routing::{get, post},
    Router,
};
use futures_util::stream::{self, StreamExt};
13
use serde_json::json;
14
use std::convert::Infallible;
15
16
17
use std::sync::Arc;
use std::time::{SystemTime, UNIX_EPOCH};
use tokio::sync::RwLock;
18
use uuid::Uuid;
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

/// Configuration for mock worker behavior
#[derive(Clone)]
pub struct MockWorkerConfig {
    pub port: u16,
    pub worker_type: WorkerType,
    pub health_status: HealthStatus,
    pub response_delay_ms: u64,
    pub fail_rate: f32,
}

#[derive(Clone, Debug)]
pub enum WorkerType {
    Regular,
    Prefill,
    Decode,
}

#[derive(Clone, Debug)]
pub enum HealthStatus {
    Healthy,
    Unhealthy,
    Degraded,
}

/// Mock worker server for testing
pub struct MockWorker {
    config: Arc<RwLock<MockWorkerConfig>>,
47
48
    shutdown_handle: Option<tokio::task::JoinHandle<()>>,
    shutdown_tx: Option<tokio::sync::oneshot::Sender<()>>,
49
50
51
52
53
54
}

impl MockWorker {
    pub fn new(config: MockWorkerConfig) -> Self {
        Self {
            config: Arc::new(RwLock::new(config)),
55
56
            shutdown_handle: None,
            shutdown_tx: None,
57
58
59
60
61
62
63
64
        }
    }

    /// Start the mock worker server
    pub async fn start(&mut self) -> Result<String, Box<dyn std::error::Error>> {
        let config = self.config.clone();
        let port = config.read().await.port;

65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
        // If port is 0, find an available port
        let port = if port == 0 {
            let listener = std::net::TcpListener::bind("127.0.0.1:0")?;
            let port = listener.local_addr()?.port();
            drop(listener);
            config.write().await.port = port;
            port
        } else {
            port
        };

        let app = Router::new()
            .route("/health", get(health_handler))
            .route("/health_generate", get(health_generate_handler))
            .route("/get_server_info", get(server_info_handler))
            .route("/get_model_info", get(model_info_handler))
            .route("/generate", post(generate_handler))
            .route("/v1/chat/completions", post(chat_completions_handler))
            .route("/v1/completions", post(completions_handler))
84
            .route("/v1/rerank", post(rerank_handler))
85
            .route("/v1/responses", post(responses_handler))
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
            .route("/flush_cache", post(flush_cache_handler))
            .route("/v1/models", get(v1_models_handler))
            .with_state(config);

        let (shutdown_tx, shutdown_rx) = tokio::sync::oneshot::channel::<()>();
        self.shutdown_tx = Some(shutdown_tx);

        // Spawn the server in a separate task
        let handle = tokio::spawn(async move {
            let listener = match tokio::net::TcpListener::bind(("127.0.0.1", port)).await {
                Ok(l) => l,
                Err(e) => {
                    eprintln!("Failed to bind to port {}: {}", port, e);
                    return;
                }
            };
102

103
104
105
106
107
108
109
110
            let server = axum::serve(listener, app).with_graceful_shutdown(async move {
                let _ = shutdown_rx.await;
            });

            if let Err(e) = server.await {
                eprintln!("Server error: {}", e);
            }
        });
111

112
        self.shutdown_handle = Some(handle);
113

114
115
116
117
118
        // Wait for the server to start
        tokio::time::sleep(tokio::time::Duration::from_millis(100)).await;

        let url = format!("http://127.0.0.1:{}", port);
        Ok(url)
119
120
121
122
    }

    /// Stop the mock worker server
    pub async fn stop(&mut self) {
123
124
125
126
127
128
129
        if let Some(shutdown_tx) = self.shutdown_tx.take() {
            let _ = shutdown_tx.send(());
        }

        if let Some(handle) = self.shutdown_handle.take() {
            // Wait for the server to shut down
            let _ = tokio::time::timeout(tokio::time::Duration::from_secs(5), handle).await;
130
131
        }
    }
132
}
133

134
135
136
137
138
139
impl Drop for MockWorker {
    fn drop(&mut self) {
        // Clean shutdown when dropped
        if let Some(shutdown_tx) = self.shutdown_tx.take() {
            let _ = shutdown_tx.send(());
        }
140
141
142
143
144
    }
}

// Handler implementations

145
146
147
148
149
/// Check if request should fail based on configured fail_rate
async fn should_fail(config: &MockWorkerConfig) -> bool {
    rand::random::<f32>() < config.fail_rate
}

150
async fn health_handler(State(config): State<Arc<RwLock<MockWorkerConfig>>>) -> Response {
151
152
153
    let config = config.read().await;

    match config.health_status {
154
        HealthStatus::Healthy => Json(json!({
155
156
157
            "status": "healthy",
            "timestamp": SystemTime::now().duration_since(UNIX_EPOCH).unwrap().as_secs(),
            "worker_type": format!("{:?}", config.worker_type),
158
159
160
161
162
163
164
165
166
167
168
        }))
        .into_response(),
        HealthStatus::Unhealthy => (
            StatusCode::SERVICE_UNAVAILABLE,
            Json(json!({
                "status": "unhealthy",
                "error": "Worker is not responding"
            })),
        )
            .into_response(),
        HealthStatus::Degraded => Json(json!({
169
170
            "status": "degraded",
            "warning": "High load detected"
171
172
        }))
        .into_response(),
173
174
175
    }
}

176
async fn health_generate_handler(State(config): State<Arc<RwLock<MockWorkerConfig>>>) -> Response {
177
178
    let config = config.read().await;

179
    if should_fail(&config).await {
180
181
182
183
184
185
186
        return (
            StatusCode::INTERNAL_SERVER_ERROR,
            Json(json!({
                "error": "Random failure for testing"
            })),
        )
            .into_response();
187
188
    }

189
    if matches!(config.health_status, HealthStatus::Healthy) {
190
        Json(json!({
191
192
193
194
            "status": "ok",
            "queue_length": 0,
            "processing_time_ms": config.response_delay_ms
        }))
195
        .into_response()
196
    } else {
197
198
199
200
201
202
203
        (
            StatusCode::SERVICE_UNAVAILABLE,
            Json(json!({
                "error": "Generation service unavailable"
            })),
        )
            .into_response()
204
205
206
    }
}

207
async fn server_info_handler(State(config): State<Arc<RwLock<MockWorkerConfig>>>) -> Response {
208
209
    let config = config.read().await;

210
    if should_fail(&config).await {
211
212
213
214
215
216
217
        return (
            StatusCode::INTERNAL_SERVER_ERROR,
            Json(json!({
                "error": "Random failure for testing"
            })),
        )
            .into_response();
218
219
    }

220
    Json(json!({
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
        "model_path": "mock-model-path",
        "tokenizer_path": "mock-tokenizer-path",
        "port": config.port,
        "host": "127.0.0.1",
        "max_num_batched_tokens": 32768,
        "max_prefill_tokens": 16384,
        "mem_fraction_static": 0.88,
        "tp_size": 1,
        "dp_size": 1,
        "stream_interval": 8,
        "dtype": "float16",
        "device": "cuda",
        "enable_flashinfer": true,
        "enable_p2p_check": true,
        "context_length": 32768,
        "chat_template": null,
        "disable_radix_cache": false,
        "enable_torch_compile": false,
        "trust_remote_code": false,
        "show_time_cost": false,
        "waiting_queue_size": 0,
        "running_queue_size": 0,
        "req_to_token_ratio": 1.2,
        "min_running_requests": 0,
        "max_running_requests": 2048,
        "max_req_num": 8192,
        "max_batch_tokens": 32768,
        "schedule_policy": "lpm",
        "schedule_conservativeness": 1.0,
        "version": "0.3.0",
        "internal_states": [{
            "waiting_queue_size": 0,
            "running_queue_size": 0
        }]
    }))
256
    .into_response()
257
258
}

259
async fn model_info_handler(State(config): State<Arc<RwLock<MockWorkerConfig>>>) -> Response {
260
261
262
    let config = config.read().await;

    if should_fail(&config).await {
263
264
265
266
267
268
269
        return (
            StatusCode::INTERNAL_SERVER_ERROR,
            Json(json!({
                "error": "Random failure for testing"
            })),
        )
            .into_response();
270
271
    }

272
    Json(json!({
273
274
275
276
277
278
279
280
281
282
        "model_path": "mock-model-path",
        "tokenizer_path": "mock-tokenizer-path",
        "is_generation": true,
        "preferred_sampling_params": {
            "temperature": 0.7,
            "top_p": 0.9,
            "top_k": 40,
            "max_tokens": 2048
        }
    }))
283
    .into_response()
284
285
286
}

async fn generate_handler(
287
288
289
    State(config): State<Arc<RwLock<MockWorkerConfig>>>,
    Json(payload): Json<serde_json::Value>,
) -> Response {
290
291
    let config = config.read().await;

292
    if should_fail(&config).await {
293
294
295
296
297
298
299
        return (
            StatusCode::INTERNAL_SERVER_ERROR,
            Json(json!({
                "error": "Random failure for testing"
            })),
        )
            .into_response();
300
301
302
303
304
305
306
307
308
309
310
311
312
313
    }

    if config.response_delay_ms > 0 {
        tokio::time::sleep(tokio::time::Duration::from_millis(config.response_delay_ms)).await;
    }

    let is_stream = payload
        .get("stream")
        .and_then(|v| v.as_bool())
        .unwrap_or(false);

    if is_stream {
        let stream_delay = config.response_delay_ms;

314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
        // Check if it's a batch request
        let is_batch = payload.get("text").and_then(|t| t.as_array()).is_some();

        let batch_size = if is_batch {
            payload
                .get("text")
                .and_then(|t| t.as_array())
                .map(|arr| arr.len())
                .unwrap_or(1)
        } else {
            1
        };

        let mut events = Vec::new();

        // Generate events for each item in batch
        for i in 0..batch_size {
331
332
333
334
            let timestamp_start = SystemTime::now()
                .duration_since(UNIX_EPOCH)
                .unwrap()
                .as_secs_f64();
335

336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
            let data = json!({
                "text": format!("Mock response {}", i + 1),
                "meta_info": {
                    "prompt_tokens": 10,
                    "completion_tokens": 5,
                    "completion_tokens_wo_jump_forward": 5,
                    "input_token_logprobs": null,
                    "output_token_logprobs": null,
                    "first_token_latency": stream_delay as f64 / 1000.0,
                    "time_to_first_token": stream_delay as f64 / 1000.0,
                    "time_per_output_token": 0.01,
                    "end_time": timestamp_start + (stream_delay as f64 / 1000.0),
                    "start_time": timestamp_start,
                    "finish_reason": {
                        "type": "stop",
                        "reason": "length"
352
                    }
353
354
355
                },
                "stage": "mid"
            });
356

357
358
            events.push(Ok::<_, Infallible>(Event::default().data(data.to_string())));
        }
359

360
361
        // Add [DONE] event
        events.push(Ok(Event::default().data("[DONE]")));
362

363
        let stream = stream::iter(events);
364

365
366
367
        Sse::new(stream)
            .keep_alive(KeepAlive::default())
            .into_response()
368
    } else {
369
370
        Json(json!({
            "text": "This is a mock response.",
371
            "meta_info": {
372
373
374
375
376
377
378
379
                "prompt_tokens": 10,
                "completion_tokens": 5,
                "completion_tokens_wo_jump_forward": 5,
                "input_token_logprobs": null,
                "output_token_logprobs": null,
                "first_token_latency": config.response_delay_ms as f64 / 1000.0,
                "time_to_first_token": config.response_delay_ms as f64 / 1000.0,
                "time_per_output_token": 0.01,
380
381
                "finish_reason": {
                    "type": "stop",
382
383
                    "reason": "length"
                }
384
385
            }
        }))
386
        .into_response()
387
388
389
390
    }
}

async fn chat_completions_handler(
391
392
393
    State(config): State<Arc<RwLock<MockWorkerConfig>>>,
    Json(payload): Json<serde_json::Value>,
) -> Response {
394
395
    let config = config.read().await;

396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
    if should_fail(&config).await {
        return (
            StatusCode::INTERNAL_SERVER_ERROR,
            Json(json!({
                "error": {
                    "message": "Random failure for testing",
                    "type": "internal_error",
                    "code": "internal_error"
                }
            })),
        )
            .into_response();
    }

    if config.response_delay_ms > 0 {
        tokio::time::sleep(tokio::time::Duration::from_millis(config.response_delay_ms)).await;
412
413
414
415
416
417
418
    }

    let is_stream = payload
        .get("stream")
        .and_then(|v| v.as_bool())
        .unwrap_or(false);

419
420
421
422
423
    let timestamp = SystemTime::now()
        .duration_since(UNIX_EPOCH)
        .unwrap()
        .as_secs();

424
    if is_stream {
425
        let request_id = format!("chatcmpl-{}", Uuid::new_v4());
426

427
428
429
        let stream = stream::once(async move {
            let chunk = json!({
                "id": request_id,
430
431
                "object": "chat.completion.chunk",
                "created": timestamp,
432
                "model": "mock-model",
433
434
435
                "choices": [{
                    "index": 0,
                    "delta": {
436
                        "content": "This is a mock chat response."
437
438
439
440
441
                    },
                    "finish_reason": null
                }]
            });

442
443
444
            Ok::<_, Infallible>(Event::default().data(chunk.to_string()))
        })
        .chain(stream::once(async { Ok(Event::default().data("[DONE]")) }));
445

446
447
448
        Sse::new(stream)
            .keep_alive(KeepAlive::default())
            .into_response()
449
    } else {
450
451
        Json(json!({
            "id": format!("chatcmpl-{}", Uuid::new_v4()),
452
            "object": "chat.completion",
453
454
            "created": timestamp,
            "model": "mock-model",
455
456
457
458
            "choices": [{
                "index": 0,
                "message": {
                    "role": "assistant",
459
                    "content": "This is a mock chat response."
460
                },
461
                "finish_reason": "stop"
462
463
464
            }],
            "usage": {
                "prompt_tokens": 10,
465
466
                "completion_tokens": 5,
                "total_tokens": 15
467
468
            }
        }))
469
        .into_response()
470
471
472
473
    }
}

async fn completions_handler(
474
475
476
    State(config): State<Arc<RwLock<MockWorkerConfig>>>,
    Json(payload): Json<serde_json::Value>,
) -> Response {
477
478
    let config = config.read().await;

479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
    if should_fail(&config).await {
        return (
            StatusCode::INTERNAL_SERVER_ERROR,
            Json(json!({
                "error": {
                    "message": "Random failure for testing",
                    "type": "internal_error",
                    "code": "internal_error"
                }
            })),
        )
            .into_response();
    }

    if config.response_delay_ms > 0 {
        tokio::time::sleep(tokio::time::Duration::from_millis(config.response_delay_ms)).await;
495
496
497
498
499
500
501
    }

    let is_stream = payload
        .get("stream")
        .and_then(|v| v.as_bool())
        .unwrap_or(false);

502
503
504
505
    let timestamp = SystemTime::now()
        .duration_since(UNIX_EPOCH)
        .unwrap()
        .as_secs();
506
507

    if is_stream {
508
        let request_id = format!("cmpl-{}", Uuid::new_v4());
509

510
511
512
513
514
515
516
517
518
519
520
521
522
        let stream = stream::once(async move {
            let chunk = json!({
                "id": request_id,
                "object": "text_completion",
                "created": timestamp,
                "model": "mock-model",
                "choices": [{
                    "text": "This is a mock completion.",
                    "index": 0,
                    "logprobs": null,
                    "finish_reason": null
                }]
            });
523

524
525
526
            Ok::<_, Infallible>(Event::default().data(chunk.to_string()))
        })
        .chain(stream::once(async { Ok(Event::default().data("[DONE]")) }));
527

528
529
530
        Sse::new(stream)
            .keep_alive(KeepAlive::default())
            .into_response()
531
    } else {
532
533
534
535
536
537
538
539
        Json(json!({
            "id": format!("cmpl-{}", Uuid::new_v4()),
            "object": "text_completion",
            "created": timestamp,
            "model": "mock-model",
            "choices": [{
                "text": "This is a mock completion.",
                "index": 0,
540
541
                "logprobs": null,
                "finish_reason": "stop"
542
            }],
543
            "usage": {
544
545
546
                "prompt_tokens": 10,
                "completion_tokens": 5,
                "total_tokens": 15
547
548
            }
        }))
549
        .into_response()
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
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
async fn responses_handler(
    State(config): State<Arc<RwLock<MockWorkerConfig>>>,
    Json(payload): Json<serde_json::Value>,
) -> Response {
    let config = config.read().await;

    if should_fail(&config).await {
        return (
            StatusCode::INTERNAL_SERVER_ERROR,
            Json(json!({
                "error": {
                    "message": "Random failure for testing",
                    "type": "internal_error",
                    "code": "internal_error"
                }
            })),
        )
            .into_response();
    }

    if config.response_delay_ms > 0 {
        tokio::time::sleep(tokio::time::Duration::from_millis(config.response_delay_ms)).await;
    }

    let is_stream = payload
        .get("stream")
        .and_then(|v| v.as_bool())
        .unwrap_or(false);

    let timestamp = SystemTime::now()
        .duration_since(UNIX_EPOCH)
        .unwrap()
        .as_secs() as i64;

    if is_stream {
        let request_id = format!("resp-{}", Uuid::new_v4());

        let stream = stream::once(async move {
            let chunk = json!({
                "id": request_id,
                "object": "response",
                "created_at": timestamp,
                "model": "mock-model",
                "status": "in_progress",
                "output": [{
                    "type": "message",
                    "role": "assistant",
                    "content": [{
                        "type": "output_text",
                        "text": "This is a mock responses streamed output."
                    }]
                }]
            });
            Ok::<_, Infallible>(Event::default().data(chunk.to_string()))
        })
        .chain(stream::once(async { Ok(Event::default().data("[DONE]")) }));

        Sse::new(stream)
            .keep_alive(KeepAlive::default())
            .into_response()
    } else {
        Json(json!({
            "id": format!("resp-{}", Uuid::new_v4()),
            "object": "response",
            "created_at": timestamp,
            "model": "mock-model",
            "output": [{
                "type": "message",
                "role": "assistant",
                "content": [{
                    "type": "output_text",
                    "text": "This is a mock responses output."
                }]
            }],
            "status": "completed",
            "usage": {
                "input_tokens": 10,
                "output_tokens": 5,
                "total_tokens": 15
            }
        }))
        .into_response()
    }
}

638
async fn flush_cache_handler(State(config): State<Arc<RwLock<MockWorkerConfig>>>) -> Response {
639
640
641
    let config = config.read().await;

    if should_fail(&config).await {
642
643
644
645
646
647
648
        return (
            StatusCode::INTERNAL_SERVER_ERROR,
            Json(json!({
                "error": "Random failure for testing"
            })),
        )
            .into_response();
649
650
    }

651
652
    Json(json!({
        "message": "Cache flushed successfully"
653
    }))
654
    .into_response()
655
656
}

657
async fn v1_models_handler(State(config): State<Arc<RwLock<MockWorkerConfig>>>) -> Response {
658
659
660
    let config = config.read().await;

    if should_fail(&config).await {
661
662
663
664
665
666
667
668
669
670
671
        return (
            StatusCode::INTERNAL_SERVER_ERROR,
            Json(json!({
                "error": {
                    "message": "Random failure for testing",
                    "type": "internal_error",
                    "code": "internal_error"
                }
            })),
        )
            .into_response();
672
673
    }

674
675
676
677
678
679
    let timestamp = SystemTime::now()
        .duration_since(UNIX_EPOCH)
        .unwrap()
        .as_secs();

    Json(json!({
680
681
        "object": "list",
        "data": [{
682
            "id": "mock-model",
683
            "object": "model",
684
685
            "created": timestamp,
            "owned_by": "organization-owner"
686
687
        }]
    }))
688
    .into_response()
689
690
}

691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
async fn rerank_handler(
    State(config): State<Arc<RwLock<MockWorkerConfig>>>,
    Json(payload): Json<serde_json::Value>,
) -> impl IntoResponse {
    let config = config.read().await;

    // Simulate response delay
    if config.response_delay_ms > 0 {
        tokio::time::sleep(tokio::time::Duration::from_millis(config.response_delay_ms)).await;
    }

    // Simulate failure rate
    if rand::random::<f32>() < config.fail_rate {
        return (StatusCode::INTERNAL_SERVER_ERROR, "Simulated failure").into_response();
    }

    // Extract documents from the request to create mock results
    let empty_vec = vec![];
    let documents = payload
        .get("documents")
        .and_then(|d| d.as_array())
        .unwrap_or(&empty_vec);

    // Create mock rerank results with scores based on document index
    let mut mock_results = Vec::new();
    for (i, doc) in documents.iter().enumerate() {
        let score = 0.95 - (i as f32 * 0.1); // Decreasing scores
        let result = serde_json::json!({
            "score": score,
            "document": doc.as_str().unwrap_or(""),
            "index": i,
            "meta_info": {
                "confidence": if score > 0.9 { "high" } else { "medium" }
            }
        });
        mock_results.push(result);
    }

    // Sort by score (highest first) to simulate proper ranking
    mock_results.sort_by(|a, b| {
        b["score"]
            .as_f64()
            .unwrap()
            .partial_cmp(&a["score"].as_f64().unwrap())
            .unwrap()
    });

    (StatusCode::OK, Json(mock_results)).into_response()
}

741
742
743
744
impl Default for MockWorkerConfig {
    fn default() -> Self {
        Self {
            port: 0,
745
746
747
748
            worker_type: WorkerType::Regular,
            health_status: HealthStatus::Healthy,
            response_delay_ms: 0,
            fail_rate: 0.0,
749
        }
750
751
    }
}