# Support runbook pointers

This page is the public/support handoff index. **M6 freeze:** treat this index as
frozen for soft launch; add entries only via M7+ review.

See also: [`ONCALL_ROTA.md`](../operations/ONCALL_ROTA.md),
[`SOFT_LAUNCH_SOAK_EVIDENCE.md`](../operations/SOFT_LAUNCH_SOAK_EVIDENCE.md),
[`GA_CUTOVER_RUNBOOK.md`](../operations/GA_CUTOVER_RUNBOOK.md),
[`POST_GA_14D_REVIEW.md`](../operations/POST_GA_14D_REVIEW.md),
[`CONTINUOUS_COMPLIANCE_CADENCE.md`](../operations/CONTINUOUS_COMPLIANCE_CADENCE.md),
[`STATUS_SLA_FREEZE.md`](../implementation/STATUS_SLA_FREEZE.md),
[`M9_POST_GA_OPS_READINESS.md`](../implementation/M9_POST_GA_OPS_READINESS.md),
[`M8_GA_CUTOVER_READINESS.md`](../implementation/M8_GA_CUTOVER_READINESS.md),
[`M7_SOAK_GA_READINESS.md`](../implementation/M7_SOAK_GA_READINESS.md).

## Customer-impacting alerts

| Symptom | Primary runbook | Related evidence |
| --- | --- | --- |
| Jobs queue but do not start | `docs/runbooks/queue-age.md` | P34 worker crash/redelivery test, `backend/scripts/p34_failure_drills.py --key worker_crash` |
| DLQ depth grows | `docs/runbooks/dlq-depth.md` | Admin DLQ requeue tests and queue integration tests |
| Completion latency exceeds SLO | `docs/runbooks/completion-latency.md` | P31 observability dashboards and synthetic merge/OCR probes |
| Webhook failure rate rises | `docs/runbooks/webhook-failure-rate.md` | Signed webhook/retry tests and P34 sandbox dispatch |
| Billing sync lag or Stripe outage | `docs/runbooks/billing-sync-lag.md` | P34 live-simulated provider fake and billing tests |
| Storage expiry/lifecycle lag | `docs/runbooks/storage-expiry.md` | Storage retention tests and lifecycle reconciliation docs |
| Scheduled task failure | `docs/runbooks/system-task-failure.md` | `docs/operations/BACKGROUND_TASKS.md` |
| Soft-launch plan/region/live-key block | `error.code=soft_launch_blocked` | `docs/implementation/M6_SOFT_LAUNCH_READINESS.md` |

## Platform and incident docs

- Observability overview: `docs/operations/OBSERVABILITY.md`
- Backup/restore and DR: `docs/operations/BACKUP_RESTORE_DR.md`
- Failure drills: `docs/operations/FAILURE_DRILLS.md`
- On-call rota (M6): `docs/operations/ONCALL_ROTA.md`
- Alert silence / paging dry-run (M5): `docs/operations/ALERT_SILENCE_PAGING_DRYRUN.md`
- Secret rotation and incident response: `docs/implementation/SECRET_ROTATION_AND_IR_RUNBOOKS.md`
- Security evidence map: `docs/implementation/SECURITY_COMPLIANCE_EVIDENCE.md`
- Data inventory, retention and erasure: `docs/implementation/DATA_INVENTORY_RETENTION_ERASURE.md`

## Support boundaries

- Support tooling must use the internal admin boundary; operators should not
  query production databases ad hoc for customer support.
- Document bytes/text are not shown by default in support contexts.
- Billing adjustments use append-only credit ledger entries and require audited
  reasons.
- Webhook replay creates a new delivery and must not mutate source job/usage
  state.
- During soft launch, do not promise Wave F / media-core capabilities (DEFER).