Event-Driven Local Stack

Use this when your team needs to exercise producer/consumer behavior locally with predictable startup order.

Baseline config

{
  "services": {
    "kafka": {
      "cmd": ["kafka-server-start.sh", "./infra/kafka/server.properties"],
      "port": 9092,
      "ready": {"type": "tcp"}
    },
    "producer": {
      "cmd": ["go", "run", "./cmd/producer"],
      "dependsOn": ["kafka"]
    },
    "consumer": {
      "cmd": ["go", "run", "./cmd/consumer"],
      "dependsOn": ["kafka"]
    }
  }
}

Why it works

• broker readiness gates both producer and consumer
• producer and consumer are independent of each other at startup
• runtime failures stay visible through one session log stream

Operational tips

• keep producer and consumer in the same wave unless one truly blocks the other
• use get_logs with service filtering for each role during incident analysis
• keep seed/fixture publishers as kind: "oneshot" when they should exit after success

When to pick another pattern

• if you need cluster scheduling or production-grade broker operations, this is outside FlowLayer scope
• if infra startup is external, use Infra Bootstrap Gate

For scenario examples, see Kafka Producer Consumer.