Production Checklist

Pre-deployment checklist for fapilog in production environments.

1. Choose Your Configuration Approach

Preset or Auto-detect or Custom Settings?

Approach	When to Use
`preset="production"`	General backend services
`preset="adaptive"`	High-throughput services
`preset="serverless"`	AWS Lambda, Cloud Functions
`auto_detect=True`	Let fapilog detect environment
Custom `Settings()`	Fine-grained control needed

See: Configuration

2. Enable Observability

Metrics

Enable metrics collection
Configure metrics export (application-level)

export FAPILOG_CORE__ENABLE_METRICS=true

Fapilog records events submitted/dropped, queue high-watermark, backpressure waits, and flush latency. Expose these via your preferred exporter.

See: Metrics

Diagnostics

Understand diagnostics output location (stderr by default)
Consider enabling internal logging for debugging

export FAPILOG_CORE__INTERNAL_LOGGING_ENABLED=true

Diagnostics emit warnings for worker errors, sink errors, backpressure drops, and serialization issues.

See: Diagnostics & Resilience

3. Validate Redaction

Review what gets redacted vs what doesn’t
Test redaction with production-like data
Decide on redaction_fail_mode (warn vs closed)

# Test redaction before deployment
from fapilog import get_logger

logger = get_logger(preset="production")
logger.info("Test", password="secret123", email="user@example.com")
# Verify output shows: password="***", email="***"

Warning: Redaction is pattern-based, not content-aware. PII in message strings is NOT redacted.

See: Redaction Behavior, Testing Redaction

4. Configure Backpressure

Size queue for expected burst load
Decide drop vs wait behavior
Monitor for dropped events

# High-throughput settings
export FAPILOG_CORE__MAX_QUEUE_SIZE=20000
export FAPILOG_CORE__BATCH_MAX_SIZE=256
export FAPILOG_CORE__DROP_ON_FULL=true  # or false to wait

See: Troubleshooting Dropped Logs

5. Sink Configuration

Configure production sinks (file, HTTP, cloud)
Verify sink credentials/endpoints
Test sink connectivity

See: CloudWatch, Loki, PostgreSQL

6. Graceful Shutdown

Ensure application calls logger.drain() or uses context manager
Verify shutdown handlers are installed (automatic with presets)

# Option 1: Context manager (recommended for sync code)
from fapilog import runtime

with runtime() as logger:
    # application code
    pass  # Automatic drain on exit

# Option 2: Async context manager
from fapilog import runtime_async

async with runtime_async() as logger:
    await logger.info("Processing...")
    # Automatic drain on exit

# Option 3: Manual drain
logger = get_logger(preset="production")
# ... application code ...
await logger.drain()

See: Graceful Shutdown

Quick Validation Script

"""Run this script to validate production readiness."""
import asyncio
from fapilog import get_async_logger


async def validate_production_config():
    logger = await get_async_logger(preset="production")

    # Test basic logging
    await logger.info("Production validation test")

    # Test redaction
    await logger.info("Redaction test", password="SHOULD_BE_MASKED")

    # Test error logging
    try:
        raise ValueError("Test exception")
    except Exception:
        await logger.exception("Exception test")

    # Drain and report
    result = await logger.drain()
    print(f"Validation complete: {result.events_flushed} events flushed")


if __name__ == "__main__":
    asyncio.run(validate_production_config())