Performance Tuning

Optimize Queen server for your workload.

Key Parameters

1. Database Pool Size

DB_POOL_SIZE=150  # Default

Guidelines:

• Light load: 20-50
• Medium load: 50-100
• Heavy load: 100-200

2. Worker Threads

NUM_WORKERS=10  # Default

Guidelines:

• Match CPU cores
• 10-20 for most workloads

3. Poll Workers (ThreadPool-Managed)

# Regular queue poll workers
POLL_WORKER_COUNT=2              # Default for regular long-polling

# Stream poll workers
STREAM_POLL_WORKER_COUNT=2       # Default for stream polling
STREAM_CONCURRENT_CHECKS=10      # Concurrent window checks per worker

# DB ThreadPool size is auto-calculated:
# = POLL_WORKER_COUNT + STREAM_POLL_WORKER_COUNT + 
#   (STREAM_POLL_WORKER_COUNT × STREAM_CONCURRENT_CHECKS) + 
#   DB_THREAD_POOL_SERVICE_THREADS (5)
# Default: 2 + 2 + 20 + 5 = 29 threads

Scaling for Heavy Long-Polling:

POLL_WORKER_COUNT=10             # More regular poll workers
STREAM_POLL_WORKER_COUNT=4       # More stream workers
STREAM_CONCURRENT_CHECKS=15      # More concurrent stream checks

PostgreSQL Tuning

-- Connection limit
max_connections = 200

-- Shared buffers
shared_buffers = 2GB

-- Work memory
work_mem = 64MB

Monitoring

curl http://localhost:6632/metrics

Complete tuning guide