Situation: We need bug reports that let an SRE start debugging immediately without back-and-forth.
Task: Provide a clear, actionable template and an example for an intermittent 502 on /api/checkout.
Action: Use this template whenever filing a triage ticket. Required fields (minimal, prioritized):
- Title: concise, include service/path and symptom (e.g., "Intermittent 502 for /api/checkout")
- Impact: user-visible effect, % requests/errors, affected customers, SLAs hit
- Time window: first seen, last seen, recurrence frequency, timezone
- Severity/Priority: pager/alert level
- Environment: prod/stage, region, cluster, deploy/revision hash
- Steps to reproduce: exact request, headers, payload, auth, rate
- Observed behavior: error messages, HTTP status, response body, timestamps
- Expected behavior: what should happen
- Telemetry & evidence: links to logs (with query), traces, dashboards, Grafana panels, SLO/alert graphs
- Recent changes: deploys, config, infra changes, traffic spikes, feature flags
- Attachments: sample request/response, cURL, HAR, screenshots
- First responder notes: who triaged, initial commands tried, tentative hypotheses
Result (example):
Title: Intermittent 502 for /api/checkout (prod - us-east-1)
Impact: ~0.8% of checkout POSTs failing last 2 hours; checkout conversion down 5%
Time window: 2025-11-22 09:12–11:05 UTC; ~every 3–10 minutes
Env: prod, us-east-1, checkout-service v1.4.7 (git abc123)
Steps to reproduce: POST /api/checkout with cart payload below (cURL attached)
Observed: HTTP 502, body "Bad Gateway"; logs show upstream timeout at gateway LB -> checkout-pod; trace IDs: X (link)
Telemetry: Grafana: gateway-502-rate (link), Jaeger traces (link), logs in Kibana query (link)
Recent changes: canary deploy of payment-proxy 2.2.0 at 08:50 UTC
Attachments: cURL, sample response, heap of relevant logs
First responder: triaged by @alice; hypothesize payment-proxy timeout; increased gateway timeout to 10s for testing.
This gives SREs everything needed to start debugging immediately.
