queuemgr 1.0.20

Full-featured job queue system with multiprocessing support for Linux

Queue Manager

A Python-based job queue system with per-job processes and signal variables.

Features

• Per-Job Processes: Each job runs in its own OS process for isolation
• Signal Variables: Jobs communicate via shared state protected by mutexes
• Registry Persistence: Job states and results are persisted to a registry
• Process Control: Start, stop, and delete jobs with graceful termination
• Progress Tracking: Real-time progress updates and status monitoring
• Error Handling: Comprehensive error handling and recovery mechanisms

Multiprocessing start method

queuemgr creates every job/manager process from its own local multiprocessing context (see queuemgr/mp_context.py) instead of relying on the process-wide default. That context defaults to spawn, which does not fork the calling process and therefore does not duplicate live threads, open sockets, or an in-flight asyncio event loop (e.g. a running asyncio/hypercorn server) into the child.

queuemgr never calls multiprocessing.set_start_method, so it does not read or mutate the global default start method used by your application or other libraries. If you need a different method (for example because your deployment environment cannot use "spawn"), set the QUEUEMGR_MP_START_METHOD environment variable to "spawn", "forkserver", or "fork" before queuemgr creates its first process.

export QUEUEMGR_MP_START_METHOD=forkserver

Installation

pip install queuemgr

Quick Start

from queuemgr.queue.job_queue import JobQueue
from queuemgr.core.registry import JsonlRegistry
from queuemgr.jobs.base import QueueJobBase
from queuemgr.core.types import JobStatus
from queuemgr.core.ipc import update_job_state

class MyJob(QueueJobBase):
    def execute(self):
        # Your job logic here
        for i in range(100):
            update_job_state(
                self._shared_state,
                progress=i,
                description=f"Processing item {i}"
            )
    
    def on_start(self):
        update_job_state(self._shared_state, description="Job started")
    
    def on_stop(self):
        update_job_state(self._shared_state, description="Job stopped")
    
    def on_end(self):
        update_job_state(self._shared_state, description="Job completed")
    
    def on_error(self, exc):
        update_job_state(self._shared_state, description=f"Job failed: {exc}")

# Create queue and job
registry = JsonlRegistry("my_registry.jsonl")
queue = JobQueue(registry)
job = MyJob("my-job-1", {"param": "value"})

# Add and start job
queue.add_job(job)
queue.start_job("my-job-1")

# Monitor progress
status = queue.get_job_status("my-job-1")
print(f"Status: {status.status}, Progress: {status.progress}%")

Examples

See the examples/ directory for complete examples:

• simple_job.py - Basic job creation and execution
• error_handling_job.py - Error handling and recovery
• progress_job.py - Long-running job with progress updates
• registry_example.py - Registry functionality and job history

Development

# Install development dependencies
pip install -e ".[dev]"

# Run tests
pytest

# Run tests with coverage
pytest --cov=queuemgr

# Format code
black queuemgr tests

# Lint code
flake8 queuemgr tests

# Type check
mypy queuemgr

License

MIT License - see LICENSE file for details.

Author

Vasiliy Zdanovskiy - [e-mail]

vvz-queuemgr