version: 1
id: redis-ha-failover-plan
title: Plan Redis HA Failover
summary: Document a payments Redis primary/replica topology, failover trigger, and client cutover checklist without data-loss windows.
difficulty: intermediate
estimatedMinutes: 25
prerequisites: [postgresql-pitr, observability-metrics-alerts]
image: alpine:3.21
shell: /bin/sh
setup:
  - "mkdir -p /workspace/redis /workspace/docs"
  - "printf 'Ticket: payments needs Redis HA plan.\\nPrimary/replica, failover trigger, client cutover.\\n' > /workspace/redis/TICKET.md"
tasks:
  - id: topology
    title: Sketch primary/replica topology
    description: 'Create /workspace/redis/topology.txt with exactly four non-empty lines covering: (1) primary role, (2) at least one replica, (3) replication link or async/sync note, (4) sentinel or cluster or failover manager name. Include the words primary and replica.'
    hints:
      - 'Tip code REDIS_TOPO: name primary, replica, and who promotes on failure.'
      - 'Exactly four non-empty lines.'
      - 'Mention Sentinel, cluster, or an equivalent failover manager.'
    checks:
      - type: file
        name: Topology present
        path: /workspace/redis/topology.txt
        value: primary
      - type: file
        name: Replica mentioned
        path: /workspace/redis/topology.txt
        value: replica
      - type: command
        name: Failover manager named
        command: "grep -Eqi 'sentinel|cluster|failover' /workspace/redis/topology.txt"
      - type: command
        name: Exactly four non-empty lines
        command: "[ $(grep -cve '^[[:space:]]*$' /workspace/redis/topology.txt) -eq 4 ]"
  - id: failover-runbook
    title: Write failover and cutover notes
    description: 'Write /workspace/docs/FAILOVER.md with headings Trigger, Promote, and Clients. Trigger must mention health or lag or down. Promote must mention replica. Clients must mention reconnect or DNS or endpoint. Write /workspace/redis/STATUS.txt containing REDIS_HA_OK.'
    hints:
      - 'Tip code FAILOVER_TRIGGER: define objective failure signals before humans argue.'
      - 'Tip code CLIENT_CUTOVER: clients must learn the new primary — reconnect/DNS/endpoint.'
      - 'STATUS.txt should include REDIS_HA_OK.'
    checks:
      - type: file
        name: Trigger section present
        path: /workspace/docs/FAILOVER.md
        value: Trigger
      - type: file
        name: Promote section present
        path: /workspace/docs/FAILOVER.md
        value: Promote
      - type: file
        name: Clients section present
        path: /workspace/docs/FAILOVER.md
        value: Clients
      - type: command
        name: Trigger mentions health signals
        command: "grep -A3 '^#* *Trigger' /workspace/docs/FAILOVER.md | grep -Eqi 'health|lag|down'"
      - type: command
        name: Promote mentions replica
        command: "grep -A3 '^#* *Promote' /workspace/docs/FAILOVER.md | grep -Eqi 'replica'"
      - type: command
        name: Clients mention cutover
        command: "grep -A3 '^#* *Clients' /workspace/docs/FAILOVER.md | grep -Eqi 'reconnect|DNS|endpoint'"
      - type: file
        name: Redis HA status ok
        path: /workspace/redis/STATUS.txt
        value: REDIS_HA_OK
limits: {cpus: "0.5", memory: 128m, pids: 64, timeout: 1800, network: false}