{ "schema": "pg36-ch24-alert-candidates-v1", "service_id": "pg36_shop", "accepted": [ { "id": "PG36ShopAvailabilityFastBurn", "class": "symptom", "objective_id": "SLO-AVAILABILITY", "route": "page", "severity": "SEV-1", "expression_contract": "bad_ratio_1h > 14.4 * 0.001 AND bad_ratio_5m > 14.4 * 0.001", "long_window": "1h", "short_window": "5m", "burn_rate": 14.4, "for": "2m", "user_impact": "order attempts are failing or have unreconciled outcomes fast enough to spend 2 percent of the 28-day budget in one hour", "owner_function": "service", "route_id": "route://shop-oncall", "runbook_id": "RB-USER-SYMPTOM", "first_safe_action": "freeze the latest risky release and reconcile unknown order outcomes before retrying writes", "verification": "the long and short windows both recover and sampled idempotency tokens reconcile", "dashboard": "dashboard://pg36-shop-slo", "missing_semantics": "unknown plus metamonitoring alert", "silence_policy": "only a bounded incident silence with owner, reason, end time, and replacement observation", "test_id": "ALERTTEST-AVAILABILITY-FAST", "expires_for_review": "2026-10-29" }, { "id": "PG36ShopAvailabilitySlowBurn", "class": "symptom", "objective_id": "SLO-AVAILABILITY", "route": "page", "severity": "SEV-2", "expression_contract": "bad_ratio_6h > 6 * 0.001 AND bad_ratio_30m > 6 * 0.001", "long_window": "6h", "short_window": "30m", "burn_rate": 6, "for": "5m", "user_impact": "sustained failed or unreconciled order attempts threaten 5 percent of the 28-day budget in six hours", "owner_function": "service", "route_id": "route://shop-oncall", "runbook_id": "RB-USER-SYMPTOM", "first_safe_action": "stop concurrent risky changes and segment failures by operation and release", "verification": "both burn windows recover and the responsible failure class is bounded", "dashboard": "dashboard://pg36-shop-slo", "missing_semantics": "unknown plus metamonitoring alert", "silence_policy": "bounded incident silence only", "test_id": "ALERTTEST-AVAILABILITY-SLOW", "expires_for_review": "2026-10-29" }, { "id": "PG36ShopAvailabilityBudgetTicket", "class": "symptom", "objective_id": "SLO-AVAILABILITY", "route": "ticket", "severity": "ticket", "expression_contract": "bad_ratio_3d > 1 * 0.001 AND bad_ratio_6h > 1 * 0.001", "long_window": "3d", "short_window": "6h", "burn_rate": 1, "for": "15m", "user_impact": "the current error rate would consume the full budget over the objective window if sustained", "owner_function": "service", "route_id": "route://shop-reliability-backlog", "runbook_id": "RB-BUDGET-REVIEW", "first_safe_action": "open an owned budget review and identify the largest eligible failure class", "verification": "ticket contains event counts, windows, release markers, and a due date", "dashboard": "dashboard://pg36-shop-slo", "missing_semantics": "unknown plus metamonitoring ticket", "silence_policy": "do not silence; deduplicate into the existing owned ticket", "test_id": "ALERTTEST-AVAILABILITY-TICKET", "expires_for_review": "2026-10-29" }, { "id": "PG36ShopCorrectnessMismatch", "class": "integrity", "objective_id": "CTRL-CORRECTNESS", "route": "page", "severity": "SEV-1", "expression_contract": "pg36_shop_reconciliation_mismatches > 0", "long_window": "current-reconciliation", "short_window": "current-reconciliation", "burn_rate": null, "for": "0m", "user_impact": "acknowledged order state may be duplicated, cross-tenant, invalid, or missing", "owner_function": "data", "route_id": "route://shop-oncall", "runbook_id": "RB-CORRECTNESS", "first_safe_action": "freeze affected writes and preserve the reconciliation boundary before repair", "verification": "an independently reviewed reconciliation returns zero unexplained mismatches", "dashboard": "dashboard://pg36-shop-correctness", "missing_semantics": "stale or missing reconciliation is a separate control failure", "silence_policy": "no routine silence; incident commander must record an alternative detector", "test_id": "ALERTTEST-CORRECTNESS", "expires_for_review": "2026-10-29" }, { "id": "PG36ShopFreshnessFastBurn", "class": "symptom", "objective_id": "SLO-FRESHNESS", "route": "page", "severity": "SEV-2", "expression_contract": "freshness_bad_ratio_1h > 14.4 * 0.01 AND freshness_bad_ratio_5m > 14.4 * 0.01", "long_window": "1h", "short_window": "5m", "burn_rate": 14.4, "for": "2m", "user_impact": "commit-correlated reads are missing the declared five-second visibility bound", "owner_function": "service", "route_id": "route://shop-oncall", "runbook_id": "RB-READ-FRESHNESS", "first_safe_action": "route the affected read-after-write journey to the primary path while preserving probe tokens", "verification": "new commit tokens become visible within five seconds on the declared path", "dashboard": "dashboard://pg36-shop-freshness", "missing_semantics": "probe absence is unknown, never fresh", "silence_policy": "bounded incident silence with an active primary-path probe", "test_id": "ALERTTEST-FRESHNESS", "expires_for_review": "2026-10-29" }, { "id": "PG36ShopCapacityHorizon", "class": "capacity", "objective_id": null, "route": "ticket", "severity": "ticket", "expression_contract": "forecast_days_to_capacity < 14 AND forecast_confidence = reviewed", "long_window": "28d-trend", "short_window": "7d-trend", "burn_rate": null, "for": "1h", "user_impact": "no present impact; forecast indicates an avoidable future saturation risk", "owner_function": "platform", "route_id": "route://db-platform-capacity", "runbook_id": "RB-CAPACITY", "first_safe_action": "validate demand, headroom, retention, and query mix before changing capacity", "verification": "an owned plan restores the reviewed headroom horizon", "dashboard": "dashboard://pg36-shop-capacity", "missing_semantics": "insufficient history produces no forecast and an evidence-quality ticket", "silence_policy": "deduplicate into a capacity plan; do not page", "test_id": "ALERTTEST-CAPACITY", "expires_for_review": "2026-10-29" }, { "id": "PG36MonitoringPathBroken", "class": "metamonitoring", "objective_id": null, "route": "page", "severity": "SEV-2", "expression_contract": "expected_probe_missing OR rule_evaluation_failure OR notification_canary_overdue", "long_window": "15m", "short_window": "5m", "burn_rate": null, "for": "5m", "user_impact": "service health is unknown because the observation or notification path cannot prove it is working", "owner_function": "platform", "route_id": "route://db-platform-oncall", "runbook_id": "RB-METAMONITORING", "first_safe_action": "establish an independent blackbox view before changing the monitored database", "verification": "a canary traverses probe, ingestion, rule evaluation, routing, and receipt", "dashboard": "dashboard://pigsty-observability", "missing_semantics": "failure by definition", "silence_policy": "only with an independent replacement observation path", "test_id": "ALERTTEST-METAMONITORING", "expires_for_review": "2026-10-29" } ], "diagnostic_only": [ { "id": "PG36ReplicaReplayDistance", "class": "cause", "route": "dashboard", "reason": "WAL distance helps explain freshness but is neither user-visible staleness nor read-your-writes proof", "page_when": "only after a correlated freshness or imminent durability symptom" }, { "id": "PG36PoolQueueDepth", "class": "cause", "route": "dashboard", "reason": "a queue can be healthy backpressure while end-to-end latency remains within objective", "page_when": "only through the user-latency symptom or an independently defined overload invariant" }, { "id": "PG36HostCpuHigh", "class": "cause", "route": "dashboard", "reason": "CPU utilization alone does not prove user harm", "page_when": "never by itself" } ], "rejected": [ { "id": "PostgresInstanceDownWithoutAction", "class": "cause", "proposed_route": "page", "decision": "rejected", "reason": "one instance can be intentionally offline or safely covered by the service; the candidate supplies neither user impact nor a safe operator action", "user_impact": null, "owner_function": null, "runbook_id": null, "first_safe_action": null, "replacement": "correlate endpoint and user-event symptoms; retain instance state on the topology dashboard" } ] }