Observability Standards

Guidelines for structured logging, performance metrics, Celery task monitoring, and audit logging.

Structured Logging Rules

MetaPilot uses Python's standard

logging

library configured via Django

LOGGING

settings.

Logs use a simple structured format:

text

{levelname} {asctime} {module} {message}

Example Output:

text

INFO 2026-07-24 01:15:22 scheduler_service Claimed 1 jobs for processing
ERROR 2026-07-24 01:16:04 websocket [Inbox WS] Redis connection timeout

DEBUG
: Detailed diagnostic info (SQL queries, internal variable dumps).
INFO
: Normal operational events (Tenant created, Job claimed, WS client connected).
WARNING
: Recoverable non-fatal events (Rate limit hit, invalid login attempt).
ERROR
: Actionable errors requiring investigation (Database failures, API exceptions).
CRITICAL
: Emergency failures causing system shutdown.

Critical administrative actions are recorded in the

AuditLog

table (

tenants.models.AuditLog

Audit log attributes include:

tenant

user

action

resource

ip_address

details_json

created_at

Monitor Celery task queues using standard CLI tools:

bash

# View active tasks and worker stats
celery -A core inspect active
celery -A core events

For production UI monitoring, deploy Flower:

bash

celery -A core flower --port=5555