version: 1
id: observability-sli-recording-rules
title: Plan SLI Recording Rules
summary: Define availability and latency SLI recording rules that feed burn-rate alerts without scraping raw series in every query.
difficulty: intermediate
estimatedMinutes: 25
prerequisites: [observability-metrics-alerts, slo-definition-basics]
image: alpine:3.21
shell: /bin/sh
setup:
  - "mkdir -p /workspace/sli /workspace/docs"
  - "printf 'Ticket: payments-api SLOs need recording rules.\\nAvailability + latency SLIs, then burn-rate ready names.\\n' > /workspace/sli/TICKET.md"
  - "printf 'slo_availability: 99.9\\nslo_latency_p99_ms: 300\\n' > /workspace/sli/slo.env"
tasks:
  - id: recording-rules
    title: Author availability and latency rules
    description: 'Create /workspace/sli/recording-rules.yaml mentioning job:payments_api:availability:ratio and job:payments_api:latency:p99. Include record: at least twice. Mention http_requests_total (or requests) for availability and histogram or bucket or quantile for latency. File must be at least 6 non-empty lines.'
    hints:
      - 'Tip code RECORD_SLI: precompute SLIs so burn-rate alerts stay cheap and stable.'
      - 'Use the exact metric names job:payments_api:availability:ratio and job:payments_api:latency:p99.'
      - 'Availability usually divides successful requests by total; latency uses histogram quantiles.'
    checks:
      - type: file
        name: Availability rule named
        path: /workspace/sli/recording-rules.yaml
        value: "job:payments_api:availability:ratio"
      - type: file
        name: Latency rule named
        path: /workspace/sli/recording-rules.yaml
        value: "job:payments_api:latency:p99"
      - type: command
        name: Uses record key
        command: "[ $(grep -c 'record:' /workspace/sli/recording-rules.yaml) -ge 2 ]"
      - type: command
        name: Availability sources requests
        command: "grep -Eqi 'http_requests_total|requests_total|request' /workspace/sli/recording-rules.yaml"
      - type: command
        name: Latency sources histogram
        command: "grep -Eqi 'histogram|bucket|quantile|le=' /workspace/sli/recording-rules.yaml"
      - type: command
        name: Rules file has depth
        command: "[ $(grep -cve '^[[:space:]]*$' /workspace/sli/recording-rules.yaml) -ge 6 ]"
  - id: burn-ready
    title: Document burn-rate inputs
    description: 'Write /workspace/docs/SLI.md with headings Availability, Latency, and Burn. Availability must mention 99.9 or ratio. Latency must mention 300 or p99. Burn must mention burn or error budget. Write /workspace/sli/STATUS.txt containing SLI_OK.'
    hints:
      - 'Tip code BURN_INPUT: burn-rate alerts should consume recording-rule SLIs, not ad-hoc scrape math.'
      - 'Pull 99.9 and 300 from slo.env.'
      - 'STATUS.txt should include SLI_OK.'
    checks:
      - type: file
        name: Availability heading present
        path: /workspace/docs/SLI.md
        value: Availability
      - type: file
        name: Latency heading present
        path: /workspace/docs/SLI.md
        value: Latency
      - type: file
        name: Burn heading present
        path: /workspace/docs/SLI.md
        value: Burn
      - type: command
        name: Availability mentions target
        command: "grep -A3 '^#* *Availability' /workspace/docs/SLI.md | grep -Eqi '99\\.9|ratio'"
      - type: command
        name: Latency mentions target
        command: "grep -A3 '^#* *Latency' /workspace/docs/SLI.md | grep -Eqi '300|p99'"
      - type: command
        name: Burn mentions budget math
        command: "grep -A3 '^#* *Burn' /workspace/docs/SLI.md | grep -Eqi 'burn|error budget|budget'"
      - type: file
        name: SLI status ok
        path: /workspace/sli/STATUS.txt
        value: SLI_OK
limits: {cpus: "0.5", memory: 128m, pids: 64, timeout: 1800, network: false}