Spaces:
Sleeping
Sleeping
File size: 21,361 Bytes
036c5c4 0164d71 036c5c4 02588dd 036c5c4 02588dd 036c5c4 8c4055f 02588dd 036c5c4 8c4055f 02588dd 036c5c4 7715603 036c5c4 8c4055f 036c5c4 8c4055f 036c5c4 8c4055f 036c5c4 8c4055f 036c5c4 8c4055f 02588dd 8c4055f 036c5c4 02588dd 036c5c4 02588dd 036c5c4 8c4055f 036c5c4 02588dd 036c5c4 8c4055f 036c5c4 8c4055f 036c5c4 0164d71 036c5c4 0164d71 036c5c4 0164d71 036c5c4 0164d71 036c5c4 0164d71 036c5c4 fc638eb 036c5c4 fc638eb bc8ed4e fc638eb 036c5c4 8c4055f 036c5c4 8c4055f 02588dd 036c5c4 8c4055f 02588dd 036c5c4 8c4055f 02588dd 036c5c4 8c4055f 036c5c4 fc638eb 036c5c4 fc638eb 036c5c4 fc638eb bc8ed4e fc638eb 036c5c4 |
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 |
"""
Modular Priority-Tier Worker Pool
A self-contained, plug-and-play worker pool for processing async jobs
with priority-tier scheduling. Can be used in any Python application.
Usage:
from services.priority_worker_pool import PriorityWorkerPool, WorkerConfig
# Define your job processor function
async def process_my_job(job, session):
# Process job and return updated job
job.status = "completed"
job.output_data = {"result": "done"}
return job
# Configure and start pool
pool = PriorityWorkerPool(
database_url="sqlite+aiosqlite:///./my_db.db",
job_model=MyJobModel,
job_processor=process_my_job,
config=WorkerConfig(fast_workers=5, medium_workers=5, slow_workers=5)
)
await pool.start()
Environment Variables (optional):
FAST_WORKERS: Number of fast workers (default: 5)
MEDIUM_WORKERS: Number of medium workers (default: 5)
SLOW_WORKERS: Number of slow workers (default: 5)
FAST_INTERVAL: Fast tier polling interval in seconds (default: 5)
MEDIUM_INTERVAL: Medium tier polling interval in seconds (default: 30)
SLOW_INTERVAL: Slow tier polling interval in seconds (default: 60)
Dependencies:
sqlalchemy[asyncio]>=2.0.0
aiosqlite (for SQLite) or asyncpg (for PostgreSQL)
Job Model Requirements:
Your job model must have these columns:
- job_id: str (unique identifier)
- status: str (queued, processing, completed, failed, cancelled)
- priority: str (fast, medium, slow)
- next_process_at: datetime (nullable, for rescheduling)
- retry_count: int (default 0)
- created_at: datetime
- started_at: datetime (nullable)
- completed_at: datetime (nullable)
- error_message: str (nullable)
"""
import asyncio
import logging
import os
from abc import ABC, abstractmethod
from dataclasses import dataclass, field
from datetime import datetime, timedelta
from typing import Optional, List, Callable, Any, TypeVar, Generic
from sqlalchemy import select, or_, and_
from sqlalchemy.ext.asyncio import create_async_engine, async_sessionmaker, AsyncSession
logger = logging.getLogger(__name__)
# Generic type for job model
JobType = TypeVar('JobType')
@dataclass
class WorkerConfig:
"""Configuration for the worker pool."""
fast_workers: int = 5
medium_workers: int = 5
slow_workers: int = 5
fast_interval: int = 2 # seconds
medium_interval: int = 10 # seconds
slow_interval: int = 15 # seconds
max_retries: int = 60 # Max retry attempts before failing
job_per_api_key: int = 1 # Max concurrent jobs per API key
@classmethod
def from_env(cls) -> 'WorkerConfig':
"""Create config from environment variables."""
return cls(
fast_workers=int(os.getenv("FAST_WORKERS", "5")),
medium_workers=int(os.getenv("MEDIUM_WORKERS", "5")),
slow_workers=int(os.getenv("SLOW_WORKERS", "5")),
fast_interval=int(os.getenv("FAST_INTERVAL", "5")),
medium_interval=int(os.getenv("MEDIUM_INTERVAL", "30")),
slow_interval=int(os.getenv("SLOW_INTERVAL", "60")),
job_per_api_key=int(os.getenv("JOB_PER_API_KEY", "1")),
)
@dataclass
class PriorityMapping:
"""Maps job types to priority tiers."""
mappings: dict = field(default_factory=dict)
def get_priority(self, job_type: str, default: str = "fast") -> str:
"""Get priority for a job type."""
return self.mappings.get(job_type, default)
def get_interval(self, priority: str, config: WorkerConfig) -> int:
"""Get polling interval for a priority tier."""
if priority == "fast":
return config.fast_interval
elif priority == "medium":
return config.medium_interval
else:
return config.slow_interval
class JobProcessor(ABC, Generic[JobType]):
"""Abstract base class for job processors."""
@abstractmethod
async def process(self, job: JobType, session: AsyncSession) -> JobType:
"""
Process a job and return the updated job.
Args:
job: The job to process
session: Database session for updates
Returns:
The updated job with new status/output
"""
pass
@abstractmethod
async def check_status(self, job: JobType, session: AsyncSession) -> JobType:
"""
Check status of an in-progress job (for async third-party operations).
Args:
job: The job to check
session: Database session for updates
Returns:
The updated job. Set next_process_at to reschedule if not done.
"""
pass
class PriorityWorker(Generic[JobType]):
"""Worker that processes jobs of a specific priority tier."""
def __init__(
self,
worker_id: int,
priority: str,
poll_interval: int,
session_maker: async_sessionmaker,
job_model: type,
job_processor: JobProcessor[JobType],
max_retries: int = 60,
wake_event: Optional[asyncio.Event] = None,
config: Optional[WorkerConfig] = None
):
self.worker_id = worker_id
self.priority = priority
self.poll_interval = poll_interval
self.session_maker = session_maker
self.job_model = job_model
self.job_processor = job_processor
self.max_retries = max_retries
self._running = False
self._current_job_id: Optional[str] = None
self._wake_event = wake_event # Event to wake worker immediately when new jobs arrive
self._config = config or WorkerConfig.from_env()
async def start(self):
"""Start the worker polling loop."""
self._running = True
logger.debug(f"Worker {self.worker_id} ({self.priority}) started, polling every {self.poll_interval}s")
asyncio.create_task(self._poll_loop())
async def stop(self):
"""Stop the worker."""
self._running = False
logger.info(f"Worker {self.worker_id} ({self.priority}) stopped")
async def _poll_loop(self):
"""Main polling loop with optimized scheduling.
Optimizations:
- When no jobs are found, sleep for poll_interval before checking again
- When a job is processed, immediately check for the next job (no waiting)
- This ensures first job starts immediately when queue was empty
- This ensures next job starts immediately after current job finishes
"""
while self._running:
job_found = False
try:
job_found = await self._process_one_job()
except Exception as e:
logger.error(f"Worker {self.worker_id}: Error in poll loop: {e}")
# Only sleep if no job was found - otherwise immediately look for next job
if not job_found:
# Wait on event with timeout - allows immediate wake-up when new job arrives
if self._wake_event:
try:
# Wait for event or timeout (whichever comes first)
await asyncio.wait_for(
self._wake_event.wait(),
timeout=self.poll_interval
)
# Clear event after waking (we'll check for jobs)
self._wake_event.clear()
except asyncio.TimeoutError:
pass # Normal timeout, check for jobs
else:
await asyncio.sleep(self.poll_interval)
async def _process_one_job(self) -> bool:
"""Find and process one job.
Enforces constraints:
1. Only one job per user can be in processing state at a time
2. Total processing jobs limited to JOB_PER_API_KEY * number of API keys
Returns:
True if a job was found and processed, False if no jobs available
"""
async with self.session_maker() as session:
from sqlalchemy import func
now = datetime.utcnow()
# Get number of API keys for capacity calculation
try:
from services.api_key_manager import get_key_count
num_api_keys = get_key_count()
max_processing = self._config.job_per_api_key * num_api_keys
except ImportError:
max_processing = 10 # Default fallback
# Check if we're at max processing capacity (only for new jobs being picked up)
count_query = select(func.count()).where(
self.job_model.status == "processing"
)
count_result = await session.execute(count_query)
current_processing = count_result.scalar() or 0
# Query for jobs matching this priority tier
query = select(self.job_model).where(
and_(
self.job_model.priority == self.priority,
self.job_model.status.in_(["queued", "processing"]),
or_(
self.job_model.next_process_at.is_(None),
self.job_model.next_process_at <= now
)
)
).order_by(self.job_model.created_at).limit(1)
result = await session.execute(query)
job = result.scalar_one_or_none()
if not job:
return False
# For queued jobs, apply the constraints
if job.status == "queued":
# Constraint 1: Check if this user already has a job in processing
user_processing_query = select(func.count()).where(
and_(
self.job_model.user_id == job.user_id,
self.job_model.status == "processing"
)
)
user_result = await session.execute(user_processing_query)
user_processing_count = user_result.scalar() or 0
if user_processing_count > 0:
logger.debug(f"Worker {self.worker_id}: User {job.user_id} already has a job processing, skipping")
return False
# Constraint 2: Check if we're at max total processing capacity
if current_processing >= max_processing:
logger.debug(f"Worker {self.worker_id}: At max capacity ({current_processing}/{max_processing}), skipping new job")
return False
self._current_job_id = job.job_id
try:
await self._process_job(session, job)
return True
except Exception as e:
logger.error(f"Worker {self.worker_id}: Error processing job {job.job_id}: {e}")
job.status = "failed"
job.error_message = str(e)
job.completed_at = datetime.utcnow()
await session.commit()
return True # Job was found, even though it failed
finally:
self._current_job_id = None
async def _process_job(self, session: AsyncSession, job: JobType):
"""Process a single job."""
logger.info(f"Worker {self.worker_id}: Processing job {job.job_id} (status: {job.status})")
from sqlalchemy import update
if job.status == "queued":
# New job - try to claim it atomically
# Set next_process_at to future to prevent others from picking it up while we process
next_check = datetime.utcnow() + timedelta(seconds=self.poll_interval * 2)
stmt = (
update(self.job_model)
.where(
self.job_model.job_id == job.job_id,
self.job_model.status == "queued"
)
.values(
status="processing",
started_at=datetime.utcnow(),
next_process_at=next_check
)
)
result = await session.execute(stmt)
await session.commit()
if result.rowcount == 0:
logger.info(f"Worker {self.worker_id}: Failed to claim job {job.job_id} (already taken)")
return
# We claimed it. Refresh and process.
await session.refresh(job)
job = await self.job_processor.process(job, session)
else:
# Already processing - try to claim for status check
# Ensure we only pick it up if next_process_at matches (or is null/past)
# But the SELECT already filtered for that.
# We just need to ensure no one else grabbed it between SELECT and UPDATE.
# Update next_process_at to future to lock it for this check
next_check = datetime.utcnow() + timedelta(seconds=self.poll_interval * 2)
stmt = (
update(self.job_model)
.where(
self.job_model.job_id == job.job_id,
or_(
self.job_model.next_process_at.is_(None),
self.job_model.next_process_at <= datetime.utcnow()
)
)
.values(next_process_at=next_check)
)
result = await session.execute(stmt)
await session.commit()
if result.rowcount == 0:
logger.info(f"Worker {self.worker_id}: Failed to claim job {job.job_id} for check (already taken)")
return
await session.refresh(job)
job = await self.job_processor.check_status(job, session)
# Handle retry limit
if job.status == "processing" and job.retry_count > self.max_retries:
job.status = "failed"
job.error_message = f"Max retries ({self.max_retries}) exceeded"
job.completed_at = datetime.utcnow()
# Handle credit finalization for jobs with reserved credits
if job.status in ("completed", "failed", "cancelled"):
await self._handle_job_credits(session, job)
await session.commit()
async def _handle_job_credits(self, session: AsyncSession, job: JobType):
"""Handle credit finalization when job reaches terminal state."""
# Check if job has credits_reserved attribute (credit-enabled jobs)
if not hasattr(job, 'credits_reserved') or job.credits_reserved <= 0:
return
try:
from services.credit_service.credit_manager import handle_job_completion
await handle_job_completion(session, job)
except ImportError:
# Credit service not available - skip
logger.debug(f"Credit service not available for job {job.job_id}")
except Exception as e:
logger.error(f"Error handling credits for job {job.job_id}: {e}")
class PriorityWorkerPool(Generic[JobType]):
"""
Modular priority-tier worker pool.
Can be used with any job model that follows the required schema.
"""
def __init__(
self,
database_url: str,
job_model: type,
job_processor: JobProcessor[JobType],
config: Optional[WorkerConfig] = None
):
"""
Initialize the worker pool.
Args:
database_url: SQLAlchemy async database URL
job_model: Your ORM model class for jobs
job_processor: Instance of JobProcessor to handle jobs
config: Worker configuration (uses env vars if not provided)
"""
self.database_url = database_url
self.job_model = job_model
self.job_processor = job_processor
self.config = config or WorkerConfig.from_env()
self.engine = create_async_engine(database_url, echo=False)
self.session_maker = async_sessionmaker(
self.engine,
class_=AsyncSession,
expire_on_commit=False
)
self.workers: List[PriorityWorker] = []
self._running = False
# Wake events for each priority tier - allows immediate job notification
self._wake_events: dict[str, asyncio.Event] = {
"fast": asyncio.Event(),
"medium": asyncio.Event(),
"slow": asyncio.Event()
}
async def start(self):
"""Start all workers."""
self._running = True
worker_id = 0
# Create fast workers
for i in range(self.config.fast_workers):
worker = PriorityWorker(
worker_id=worker_id,
priority="fast",
poll_interval=self.config.fast_interval,
session_maker=self.session_maker,
job_model=self.job_model,
job_processor=self.job_processor,
max_retries=self.config.max_retries,
wake_event=self._wake_events["fast"],
config=self.config
)
self.workers.append(worker)
await worker.start()
worker_id += 1
# Create medium workers
for i in range(self.config.medium_workers):
worker = PriorityWorker(
worker_id=worker_id,
priority="medium",
poll_interval=self.config.medium_interval,
session_maker=self.session_maker,
job_model=self.job_model,
job_processor=self.job_processor,
max_retries=self.config.max_retries,
wake_event=self._wake_events["medium"],
config=self.config
)
self.workers.append(worker)
await worker.start()
worker_id += 1
# Create slow workers
for i in range(self.config.slow_workers):
worker = PriorityWorker(
worker_id=worker_id,
priority="slow",
poll_interval=self.config.slow_interval,
session_maker=self.session_maker,
job_model=self.job_model,
job_processor=self.job_processor,
max_retries=self.config.max_retries,
wake_event=self._wake_events["slow"],
config=self.config
)
self.workers.append(worker)
await worker.start()
worker_id += 1
total = self.config.fast_workers + self.config.medium_workers + self.config.slow_workers
logger.info(
f"PriorityWorkerPool started with {total} workers: "
f"{self.config.fast_workers} fast, {self.config.medium_workers} medium, {self.config.slow_workers} slow"
)
def notify_new_job(self, priority: str):
"""
Wake sleeping workers of the specified priority tier.
Call this when a new job is created to start processing immediately.
Args:
priority: Priority tier ("fast", "medium", or "slow")
"""
if priority in self._wake_events:
self._wake_events[priority].set()
logger.debug(f"Notified {priority} workers of new job")
async def stop(self):
"""Stop all workers and refund orphaned jobs."""
self._running = False
# Refund credits for any jobs that were processing when server stopped
await self._refund_orphaned_jobs()
for worker in self.workers:
await worker.stop()
logger.info("PriorityWorkerPool stopped")
async def _refund_orphaned_jobs(self):
"""Refund credits for jobs abandoned during shutdown."""
try:
from services.credit_service.credit_manager import refund_orphaned_jobs
async with self.session_maker() as session:
refund_count = await refund_orphaned_jobs(session)
if refund_count > 0:
logger.info(f"Shutdown: Refunded {refund_count} orphaned job(s)")
except ImportError:
logger.debug("Credit service not available for orphaned job refunds")
except Exception as e:
logger.error(f"Error refunding orphaned jobs during shutdown: {e}")
# Convenience functions for priority mapping
def get_priority_for_job_type(job_type: str, mappings: dict) -> str:
"""Get priority tier for a job type using provided mappings."""
return mappings.get(job_type, "fast")
def get_interval_for_priority(priority: str, config: Optional[WorkerConfig] = None) -> int:
"""Get polling interval for a priority tier."""
cfg = config or WorkerConfig.from_env()
if priority == "fast":
return cfg.fast_interval
elif priority == "medium":
return cfg.medium_interval
else:
return cfg.slow_interval
|