logfire-logging

Logfire Logging

Core Pattern

from helpers.observability import logfire

# Basic logging
logfire.info("Operation completed",
             operation="data_processing",
             item_count=123)

# Spans for duration tracking
with logfire.span("Processing batch", batch_id=batch.id):
    result = process_batch(batch)
    logfire.info("Batch complete", items_processed=result.count)

When to Use Spans

We use spans for operations that have duration to track, nested sub-operations, or need grouped logs for easier debugging.

Span Naming

Create specific, human-readable span names:

with logfire.span(
    f"Processing {strategy_name} for {item.name}",
    item_id=item.id,
    strategy_name=strategy_name,
    operation="strategy_processing"
):
    execute_strategy()

with logfire.span(
    f"Syncing orders for {customer.email}",
    customer_id=customer.id,
    order_count=len(orders),
    operation="order_sync"
):
    sync_all_orders(orders)

with logfire.span(
    f"Generating report for {date.isoformat()}",
    report_type="daily_sales",
    date=date.isoformat()
):
    generate_daily_report(date)

Span names should immediately tell you what's happening. Include key identifiers to make debugging easier.

Logging Levels

• logfire.info() - Normal operations, important events
• logfire.error() - Serious problems
• logfire.warning() - Concerning but recoverable
• logfire.debug() - Detailed info for development

Context Attributes

Include searchable, meaningful attributes:

logfire.info(
    "Order processed successfully",
    order_id=order.id,
    customer_id=customer.id,
    total_amount=float(order.total),
    operation="order_processing"
)

What NOT to Log

We skip sensitive data (API keys, passwords), high-frequency noise (every loop iteration), obvious operations ("Starting function"), and debugging artifacts (use debug level for raw responses).

Best Practices

We make the first line immediately useful, include specific values (not just "processing"), use consistent attribute naming (snake_case), convert Decimals to floats, and stay thoughtful about volume.