File size: 23,183 Bytes
4b9d59b | 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 | """
Performance Insights for AI Debug System
Analyzes performance metrics and detects:
- Component latency breakdown (p50, p95, p99)
- Bottleneck identification
- Resource utilization tracking
- Performance degradation over time
Example insights:
- "Agent-456 response time increased by 200%"
- "Database queries taking 95th percentile of 2.5s"
- "CPU utilization at 95% on node-1"
"""
import asyncio
from collections import defaultdict
from datetime import datetime, timedelta
from typing import Any, Dict, List, Optional, Tuple
from sqlalchemy.orm import Session
from sqlalchemy import and_, or_, func
from core.models import (
DebugEvent,
DebugMetric,
DebugInsight,
DebugInsightType,
DebugInsightSeverity,
)
from core.structured_logger import StructuredLogger
class PerformanceInsightGenerator:
"""
Generates insights about system performance.
Analyzes latency, throughput, and resource utilization to
identify performance bottlenecks and degradation.
"""
def __init__(self, db_session: Session):
"""
Initialize performance insight generator.
Args:
db_session: SQLAlchemy database session
"""
self.logger = StructuredLogger(__name__)
self.db = db_session
async def analyze_component_latency(
self,
component_type: str,
component_id: str,
time_range: str = "last_1h",
) -> Optional[DebugInsight]:
"""
Analyze component latency with percentiles.
Args:
component_type: Component type
component_id: Component ID
time_range: Time range to analyze
Returns:
Performance insight or None
"""
try:
time_filter = self._parse_time_range(time_range)
# Get events with duration data
events_with_duration = (
self.db.query(DebugEvent)
.filter(
and_(
DebugEvent.component_type == component_type,
DebugEvent.component_id == component_id,
DebugEvent.timestamp >= time_filter,
DebugEvent.data.isnot(None),
)
)
.all()
)
# Extract duration values
durations = []
for event in events_with_duration:
if event.data and "duration_ms" in event.data:
durations.append(event.data["duration_ms"])
if len(durations) < 10:
return None # Not enough data
# Calculate percentiles
durations.sort()
p50 = durations[int(len(durations) * 0.5)]
p95 = durations[int(len(durations) * 0.95)]
p99 = durations[int(len(durations) * 0.99)]
avg = sum(durations) / len(durations)
# Check for performance issues
if p95 > 5000: # 5 second threshold
return DebugInsight(
insight_type=DebugInsightType.PERFORMANCE.value,
severity=DebugInsightSeverity.WARNING.value,
title=f"High latency detected for {component_type}/{component_id}",
description=f"95th percentile latency is {p95:.0f}ms (avg: {avg:.0f}ms)",
summary=f"P95 latency {p95:.0f}ms exceeds 5s threshold",
evidence={
"component_type": component_type,
"component_id": component_id,
"p50_ms": p50,
"p95_ms": p95,
"p99_ms": p99,
"avg_ms": avg,
"sample_count": len(durations),
},
confidence_score=0.90,
suggestions=[
"Profile hot paths",
"Check for N+1 queries",
"Review external API calls",
"Add caching for frequently accessed data",
],
scope="component",
affected_components=[{"type": component_type, "id": component_id}],
generated_at=datetime.utcnow(),
)
# Latency is acceptable
return DebugInsight(
insight_type=DebugInsightType.PERFORMANCE.value,
severity=DebugInsightSeverity.INFO.value,
title=f"Latency acceptable for {component_type}/{component_id}",
description=f"P95 latency {p95:.0f}ms within acceptable range",
summary=f"Performance: P50={p50:.0f}ms, P95={p95:.0f}ms, P99={p99:.0f}ms",
evidence={
"component_type": component_type,
"component_id": component_id,
"p50_ms": p50,
"p95_ms": p95,
"p99_ms": p99,
},
confidence_score=0.95,
scope="component",
affected_components=[{"type": component_type, "id": component_id}],
generated_at=datetime.utcnow(),
)
except Exception as e:
self.logger.error(
"Failed to analyze component latency",
component_type=component_type,
component_id=component_id,
error=str(e),
)
return None
async def identify_bottlenecks(
self,
correlation_id: str,
) -> Optional[DebugInsight]:
"""
Identify performance bottlenecks in an operation.
Args:
correlation_id: Operation to analyze
Returns:
Performance insight or None
"""
try:
# Get all events for the operation
events = (
self.db.query(DebugEvent)
.filter(DebugEvent.correlation_id == correlation_id)
.order_by(DebugEvent.timestamp)
.all()
)
if not events:
return None
# Find slowest steps
slow_steps = []
for event in events:
if event.data and "duration_ms" in event.data:
slow_steps.append({
"component": f"{event.component_type}/{event.component_id}",
"message": event.message,
"duration_ms": event.data["duration_ms"],
})
if not slow_steps:
return None
# Sort by duration
slow_steps.sort(key=lambda x: x["duration_ms"], reverse=True)
total_duration = sum(s["duration_ms"] for s in slow_steps)
# Check if any step dominates the execution time
if slow_steps:
slowest = slow_steps[0]
slowest_percentage = (slowest["duration_ms"] / total_duration * 100) if total_duration > 0 else 0
if slowest_percentage > 50: # One step takes >50% of time
return DebugInsight(
insight_type=DebugInsightType.PERFORMANCE.value,
severity=DebugInsightSeverity.WARNING.value,
title="Performance bottleneck identified",
description=f"{slowest['component']} - {slowest['message']} "
f"takes {slowest_percentage:.1f}% of total execution time",
summary=f"Bottleneck: {slowest['duration_ms']}ms ({slowest_percentage:.1f}% of total)",
evidence={
"correlation_id": correlation_id,
"slowest_step": slowest,
"slowest_percentage": slowest_percentage,
"total_duration_ms": total_duration,
"all_steps": slow_steps[:5], # Top 5 steps
},
confidence_score=0.88,
suggestions=[
f"Optimize {slowest['component']} operation",
"Check for inefficient queries",
"Review algorithmic complexity",
"Consider parallelization",
],
scope="component",
affected_components=[
{"type": e.component_type, "id": e.component_id} for e in events
],
generated_at=datetime.utcnow(),
)
return None
except Exception as e:
self.logger.error(
"Failed to identify bottlenecks",
correlation_id=correlation_id,
error=str(e),
)
return None
async def track_resource_utilization(
self,
time_range: str = "last_1h",
) -> List[DebugInsight]:
"""
Track resource utilization across components.
Args:
time_range: Time range to analyze
Returns:
List of performance insights
"""
try:
insights = []
time_filter = self._parse_time_range(time_range)
# Get CPU usage metrics
cpu_metrics = (
self.db.query(DebugMetric)
.filter(
and_(
DebugMetric.metric_name == "cpu_usage",
DebugMetric.timestamp >= time_filter,
)
)
.all()
)
# Group by component and calculate averages
cpu_by_component = defaultdict(list)
for metric in cpu_metrics:
key = f"{metric.component_type}/{metric.component_id}"
cpu_by_component[key].append(metric.value)
# Check for high CPU usage
for comp_key, values in cpu_by_component.items():
avg_cpu = sum(values) / len(values)
max_cpu = max(values)
if avg_cpu > 80: # 80% average threshold
component_type, component_id = comp_key.split("/")
insights.append(
DebugInsight(
insight_type=DebugInsightType.PERFORMANCE.value,
severity=DebugInsightSeverity.WARNING.value,
title=f"High CPU usage on {comp_key}",
description=f"Average CPU usage {avg_cpu:.1f}% (peak: {max_cpu:.1f}%)",
summary=f"CPU utilization at {avg_cpu:.1f}% on average",
evidence={
"component_type": component_type,
"component_id": component_id,
"avg_cpu_percent": avg_cpu,
"max_cpu_percent": max_cpu,
"sample_count": len(values),
},
confidence_score=0.92,
suggestions=[
"Scale up resources",
"Optimize code efficiency",
"Review for infinite loops",
"Check for background processes",
],
scope="component",
affected_components=[{"type": component_type, "id": component_id}],
generated_at=datetime.utcnow(),
)
)
# Get memory usage metrics
memory_metrics = (
self.db.query(DebugMetric)
.filter(
and_(
DebugMetric.metric_name == "memory_usage",
DebugMetric.timestamp >= time_filter,
)
)
.all()
)
memory_by_component = defaultdict(list)
for metric in memory_metrics:
key = f"{metric.component_type}/{metric.component_id}"
memory_by_component[key].append(metric.value)
# Check for high memory usage
for comp_key, values in memory_by_component.items():
avg_mem = sum(values) / len(values)
max_mem = max(values)
if avg_mem > 80: # 80% average threshold
component_type, component_id = comp_key.split("/")
insights.append(
DebugInsight(
insight_type=DebugInsightType.PERFORMANCE.value,
severity=DebugInsightSeverity.WARNING.value,
title=f"High memory usage on {comp_key}",
description=f"Average memory usage {avg_mem:.1f}% (peak: {max_mem:.1f}%)",
summary=f"Memory utilization at {avg_mem:.1f}% on average",
evidence={
"component_type": component_type,
"component_id": component_id,
"avg_memory_percent": avg_mem,
"max_memory_percent": max_mem,
"sample_count": len(values),
},
confidence_score=0.92,
suggestions=[
"Check for memory leaks",
"Review caching strategy",
"Optimize data structures",
"Consider memory profiling",
],
scope="component",
affected_components=[{"type": component_type, "id": component_id}],
generated_at=datetime.utcnow(),
)
)
return insights
except Exception as e:
self.logger.error("Failed to track resource utilization", error=str(e))
return []
async def detect_performance_degradation(
self,
component_type: str,
component_id: str,
time_range: str = "last_24h",
) -> Optional[DebugInsight]:
"""
Detect performance degradation over time.
Args:
component_type: Component type
component_id: Component ID
time_range: Time range to analyze
Returns:
Performance insight or None
"""
try:
time_filter = self._parse_time_range(time_range)
# Get events with duration data
events = (
self.db.query(DebugEvent)
.filter(
and_(
DebugEvent.component_type == component_type,
DebugEvent.component_id == component_id,
DebugEvent.timestamp >= time_filter,
DebugEvent.data.isnot(None),
)
)
.order_by(DebugEvent.timestamp)
.all()
)
# Split into first half and second half
mid_point = len(events) // 2
first_half = events[:mid_point]
second_half = events[mid_point:]
if len(first_half) < 10 or len(second_half) < 10:
return None
# Calculate average duration for each half
first_avg = 0
first_count = 0
for event in first_half:
if event.data and "duration_ms" in event.data:
first_avg += event.data["duration_ms"]
first_count += 1
second_avg = 0
second_count = 0
for event in second_half:
if event.data and "duration_ms" in event.data:
second_avg += event.data["duration_ms"]
second_count += 1
if first_count > 0 and second_count > 0:
first_avg /= first_count
second_avg /= second_count
# Check for degradation (>20% slower)
if second_avg > first_avg * 1.2:
degradation_percent = ((second_avg - first_avg) / first_avg) * 100
return DebugInsight(
insight_type=DebugInsightType.PERFORMANCE.value,
severity=DebugInsightSeverity.WARNING.value,
title=f"Performance degradation detected for {component_type}/{component_id}",
description=f"Average response time increased by {degradation_percent:.1f}% "
f"({first_avg:.0f}ms → {second_avg:.0f}ms)",
summary=f"Performance degraded by {degradation_percent:.1f}% over time range",
evidence={
"component_type": component_type,
"component_id": component_id,
"first_half_avg_ms": first_avg,
"second_half_avg_ms": second_avg,
"degradation_percent": degradation_percent,
"time_range": time_range,
},
confidence_score=0.85,
suggestions=[
"Review recent changes",
"Check for increased load",
"Analyze database query plans",
"Review external API performance",
"Consider recent deployments",
],
scope="component",
affected_components=[{"type": component_type, "id": component_id}],
generated_at=datetime.utcnow(),
)
return None
except Exception as e:
self.logger.error(
"Failed to detect performance degradation",
component_type=component_type,
component_id=component_id,
error=str(e),
)
return None
async def analyze_throughput(
self,
component_type: str,
time_range: str = "last_1h",
) -> Optional[DebugInsight]:
"""
Analyze throughput metrics for a component type.
Args:
component_type: Component type to analyze
time_range: Time range to analyze
Returns:
Performance insight or None
"""
try:
time_filter = self._parse_time_range(time_range)
# Count events per minute
events_per_minute = (
self.db.query(
func.strftime("%Y-%m-%d %H:%M", DebugEvent.timestamp).label("minute"),
func.count(DebugEvent.id).label("count"),
)
.filter(
and_(
DebugEvent.component_type == component_type,
DebugEvent.timestamp >= time_filter,
)
)
.group_by(func.strftime("%Y-%m-%d %H:%M", DebugEvent.timestamp))
.all()
)
if not events_per_minute:
return None
counts = [count for _, count in events_per_minute]
avg_throughput = sum(counts) / len(counts)
min_throughput = min(counts)
max_throughput = max(counts)
# Check for low throughput
if avg_throughput < 10: # Less than 10 events/min
return DebugInsight(
insight_type=DebugInsightType.PERFORMANCE.value,
severity=DebugInsightSeverity.INFO.value,
title=f"Low throughput for {component_type}",
description=f"Average {avg_throughput:.1f} events/minute (min: {min_throughput}, max: {max_throughput})",
summary=f"Throughput: {avg_throughput:.1f} events/min",
evidence={
"component_type": component_type,
"avg_throughput_per_min": avg_throughput,
"min_throughput_per_min": min_throughput,
"max_throughput_per_min": max_throughput,
},
confidence_score=0.80,
suggestions=[
"Check if this is expected for the workload",
"Review scaling configuration",
"Consider horizontal scaling",
],
scope="component",
affected_components=[{"type": component_type}],
generated_at=datetime.utcnow(),
)
# High throughput
if avg_throughput > 1000: # More than 1000 events/min
return DebugInsight(
insight_type=DebugInsightType.PERFORMANCE.value,
severity=DebugInsightSeverity.INFO.value,
title=f"High throughput for {component_type}",
description=f"Average {avg_throughput:.1f} events/minute (min: {min_throughput}, max: {max_throughput})",
summary=f"High throughput: {avg_throughput:.1f} events/min",
evidence={
"component_type": component_type,
"avg_throughput_per_min": avg_throughput,
"min_throughput_per_min": min_throughput,
"max_throughput_per_min": max_throughput,
},
confidence_score=0.80,
suggestions=[
"Consider scaling up if resources are constrained",
"Monitor for errors under high load",
],
scope="component",
affected_components=[{"type": component_type}],
generated_at=datetime.utcnow(),
)
return None
except Exception as e:
self.logger.error(
"Failed to analyze throughput",
component_type=component_type,
error=str(e),
)
return None
def _parse_time_range(self, time_range: str) -> datetime:
"""Parse time range string to datetime."""
now = datetime.utcnow()
if time_range == "last_1h":
return now - timedelta(hours=1)
elif time_range == "last_24h":
return now - timedelta(hours=24)
elif time_range == "last_7d":
return now - timedelta(days=7)
else:
return now - timedelta(hours=1)
|