// the find
smartloli/EFAK
A AI-Driven, Distributed and high-performance monitoring system, for comprehensive monitoring and management of kafka cluster.
EFAK (Eagle for Apache Kafka) is a web-based monitoring and management dashboard for Kafka clusters, now in v5 with an AI chat assistant bolted on. It targets ops teams running Kafka who want a self-hosted alternative to paid observability tools like Confluent Cloud or Datadog's Kafka integration.
The function-calling implementation is genuinely useful — the AI can query live broker metrics, consumer lag, and topic stats rather than just answering generic Kafka questions. Multi-cluster support is built in from the start, not an afterthought. The distributed scheduler using Redis for sharding is a thoughtful choice for HA deployments, and it gracefully degrades to single-node mode. Docker Compose setup brings up MySQL, Redis, and the app together with one command, which is the right default for a tool this complex.
The 'AI-driven' label is doing heavy lifting — this is a chat widget over standard function calling, not anything that autonomously detects or remediates problems. The README is written almost entirely in Chinese for a project advertising itself internationally, which will frustrate adoption outside China. MySQL as the primary store is a strange choice when Kafka shops almost universally run Postgres or TimescaleDB for time-series metrics — you're adding a third database dependency alongside Kafka and Redis. No mention of RBAC or multi-tenancy, so in any org with more than one team, everyone shares the admin password.