15.259 14.9.19 Alertas
| ID | Alerta | Severidad | Condición | Contrato |
|---|---|---|---|---|
| ALT-01 | CaptureFrameLoss | Critical | Sustained required-stream gap beyond profile tolerance. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-02 | CameraDisconnected | Critical | Camera unavailable while capture session active. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-03 | ClockSkewExceeded | High | Cross-stream residual invalidates temporal claims. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-04 | DiskCritical | Critical | Insufficient capacity to guarantee next capture window. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-05 | DiskForecastWarning | Warning | Predicted time-to-full below operator horizon. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-06 | FilesystemReadOnly | Critical | Writes fail or filesystem remounted read-only. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-07 | MetadataIntegrityFailure | Critical | Database integrity/transaction failures exceed immediate threshold. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-08 | QueueStalled | High | Oldest ready age rises while workers appear available. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-09 | WorkerCrashLoop | High | Restart budget exceeded. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-10 | JetsonThermalThrottle | High | Sustained throttling impacts benchmarked profile. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-11 | JetsonMemoryPressure | High | Memory/swap or OOM trend crosses profile gate. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-12 | InferenceLatencyRegression | Warning | Paired/profile baseline regression beyond change budget. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-13 | QualityUnknownSpike | Warning | Unknown/degraded fraction changes materially by controlled window. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-14 | ArtifactDigestFailure | Critical | Any confirmed content integrity mismatch. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-15 | ModelVerificationFailure | Critical | Unsigned, untrusted or incompatible model rejected. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-16 | TelemetryDropCritical | High | Critical events are being dropped. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-17 | SyncBacklogGrowing | Warning | Eligible backlog grows beyond capacity/policy. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-18 | BackupRestoreOverdue | Warning | No successful restore test within profile interval. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-19 | UpdateRollback | High | Automatic rollback occurred. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
| ALT-20 | UnauthorizedAccess | High | Repeated denied access or policy violation. | deduplication runbookId evidenceLinks cooldown resolvedEvent |
Una alerta sin runbook, evidencia y evento de resolución es un mensaje, no un mecanismo operacional. Los thresholds se derivan de SLI y capacidad observada para reducir flapping.