Kanchi

Kanchi is a self-hosted operations platform designed for managing Celery task queues, with a focus on recovery, triage, and automation after failures occur. It addresses the operational challenges teams face when dealing with failed, orphaned, retrying, and running Celery tasks by providing real-time visibility into task state and enabling deliberate, auditable recovery actions. Kanchi is intended for operators and teams responsible for distributed task queues who need to quickly detect, inspect, and resolve incidents in their Celery workflows.

The tool connects to RabbitMQ or Redis brokers and can be deployed alongside existing Celery workers without requiring changes to application code. Kanchi captures live broker events, worker heartbeat context, and offers filters that align with incident investigation workflows. Operators can inspect detailed task payloads, including arguments, keyword arguments, traceback, queue, worker, retry chains, rerun lineage, and progress steps, all consolidated in a single interface. Recovery actions such as reruns are handled through a review process that checks payloads, allows for input repair, skips unsafe items, and records outcomes, ensuring that reruns are deliberate and traceable rather than blind replays.

Automation features in Kanchi let teams build no-code workflows to handle repeat incidents, using triggers, conditions, Slack or webhook notifications, retries, cooldowns, and circuit breakers to avoid creating retry loops. The platform supports bulk actions, guarded reruns, and provides retention controls to manage task history, keeping successful task records lean while preserving data on failed, retried, and orphaned tasks for analysis. Every operational action—such as resolve, unresolve, rerun, skip, failure, and payload edits—is auditable, with a history that survives restarts thanks to storage in SQLite, Postgres, or MySQL databases.

Kanchi includes authentication features such as basic auth, OAuth, sessions, host and origin controls, and email allowlists, catering to teams that require restricted dashboard access. The platform is MIT licensed and supports integration with Prometheus for metrics. Its design emphasizes making task state recoverable, explainable, and safe to automate, supporting teams throughout the operational lifecycle of Celery task management.