277 lines
10 KiB
JSON
277 lines
10 KiB
JSON
{
|
||
"schema_version": "backup_notification_policy_v1",
|
||
"generated_at": "2026-06-04T21:42:18+08:00",
|
||
"source_readiness_matrix_ref": "docs/evaluations/backup_dr_readiness_matrix_2026-06-04.json",
|
||
"source_refs": [
|
||
"docs/runbooks/BACKUP-STATUS.md",
|
||
"docs/evaluations/backup_dr_readiness_matrix_2026-06-04.json",
|
||
"docs/workplans/2026-06-04-reboot-cold-start-backup-recovery-workplan.md",
|
||
"scripts/backup/backup-status.sh",
|
||
"scripts/ops/backup-alert-label-contract-check.py",
|
||
"scripts/ops/backup-health-textfile-exporter.py"
|
||
],
|
||
"program_status": {
|
||
"overall_completion_percent": 100,
|
||
"current_priority": "P1",
|
||
"current_task_id": "P1-103",
|
||
"next_task_id": "P1-104",
|
||
"read_only_mode": true
|
||
},
|
||
"rollups": {
|
||
"total_rules": 8,
|
||
"by_decision": {
|
||
"suppress_immediate_success": 2,
|
||
"escalate_immediate": 4,
|
||
"create_action_required": 2
|
||
},
|
||
"immediate_escalation_rule_ids": [
|
||
"backup_warning_stale",
|
||
"backup_failed",
|
||
"offsite_verify_failure",
|
||
"backup_status_core_blocker"
|
||
],
|
||
"suppressed_success_rule_ids": [
|
||
"scheduled_backup_success",
|
||
"offsite_sync_success"
|
||
]
|
||
},
|
||
"notification_channels": [
|
||
{
|
||
"channel_id": "awooop_operator_event",
|
||
"purpose": "承載需要人工處理、incident 或批准證據的 operator-visible event。",
|
||
"immediate_allowed": true,
|
||
"success_immediate_allowed": false,
|
||
"requires_operator_action": true
|
||
},
|
||
{
|
||
"channel_id": "telegram_ops",
|
||
"purpose": "只承載 failure、warning 或 action-required 即時升級；正常成功不得即時送出。",
|
||
"immediate_allowed": true,
|
||
"success_immediate_allowed": false,
|
||
"requires_operator_action": true
|
||
},
|
||
{
|
||
"channel_id": "prometheus_textfile",
|
||
"purpose": "承載成功、失敗與新鮮度證據，供每日摘要與 alert rule 讀取。",
|
||
"immediate_allowed": false,
|
||
"success_immediate_allowed": false,
|
||
"requires_operator_action": false
|
||
},
|
||
{
|
||
"channel_id": "daily_status_summary",
|
||
"purpose": "每日 06:05 台北時間摘要成功狀態、警告、阻擋與下一步。",
|
||
"immediate_allowed": false,
|
||
"success_immediate_allowed": false,
|
||
"requires_operator_action": false
|
||
}
|
||
],
|
||
"policy_rules": [
|
||
{
|
||
"rule_id": "scheduled_backup_success",
|
||
"event_kind": "backup_job_completed",
|
||
"backup_state": "success",
|
||
"severity": "info",
|
||
"decision": "suppress_immediate_success",
|
||
"channels": ["prometheus_textfile", "daily_status_summary"],
|
||
"owner_agent": "hermes",
|
||
"requires_incident": false,
|
||
"requires_approval_record": false,
|
||
"message_contract": "成功只寫入 metrics / textfile 與每日 06:05 摘要；不得送 Telegram / AwoooP 即時成功訊息。",
|
||
"evidence_refs": ["docs/runbooks/BACKUP-STATUS.md"]
|
||
},
|
||
{
|
||
"rule_id": "offsite_sync_success",
|
||
"event_kind": "offsite_verify_completed",
|
||
"backup_state": "success",
|
||
"severity": "info",
|
||
"decision": "suppress_immediate_success",
|
||
"channels": ["prometheus_textfile", "daily_status_summary"],
|
||
"owner_agent": "hermes",
|
||
"requires_incident": false,
|
||
"requires_approval_record": false,
|
||
"message_contract": "異地 verify 成功不即時洗版；只進 latest-only freshness 證據與每日摘要。",
|
||
"evidence_refs": [
|
||
"docs/runbooks/BACKUP-STATUS.md",
|
||
"scripts/backup/verify-offsite-full-sync.sh"
|
||
]
|
||
},
|
||
{
|
||
"rule_id": "backup_warning_stale",
|
||
"event_kind": "backup_freshness_warning",
|
||
"backup_state": "warning",
|
||
"severity": "warning",
|
||
"decision": "escalate_immediate",
|
||
"channels": ["awooop_operator_event", "telegram_ops", "prometheus_textfile"],
|
||
"owner_agent": "openclaw",
|
||
"requires_incident": true,
|
||
"requires_approval_record": false,
|
||
"message_contract": "警告必須帶 target、freshness、last_success_at、evidence ref 與下一個 read-only check；不得夾帶 secret。",
|
||
"evidence_refs": [
|
||
"docs/runbooks/BACKUP-STATUS.md",
|
||
"scripts/backup/backup-status.sh"
|
||
]
|
||
},
|
||
{
|
||
"rule_id": "backup_failed",
|
||
"event_kind": "backup_job_failed",
|
||
"backup_state": "failed",
|
||
"severity": "critical",
|
||
"decision": "escalate_immediate",
|
||
"channels": ["awooop_operator_event", "telegram_ops", "prometheus_textfile"],
|
||
"owner_agent": "openclaw",
|
||
"requires_incident": true,
|
||
"requires_approval_record": false,
|
||
"message_contract": "失敗立即升級，必須包含 target、job、exit code、last success、log evidence ref 與人工處置入口。",
|
||
"evidence_refs": [
|
||
"docs/runbooks/BACKUP-STATUS.md",
|
||
"scripts/backup/backup-status.sh"
|
||
]
|
||
},
|
||
{
|
||
"rule_id": "offsite_verify_failure",
|
||
"event_kind": "offsite_verify_failed",
|
||
"backup_state": "failed",
|
||
"severity": "critical",
|
||
"decision": "escalate_immediate",
|
||
"channels": ["awooop_operator_event", "telegram_ops", "prometheus_textfile"],
|
||
"owner_agent": "openclaw",
|
||
"requires_incident": true,
|
||
"requires_approval_record": false,
|
||
"message_contract": "異地 verify 失敗必須升級並保留 local / remote repo、latest-only 狀態與 retry 建議；不得自動 sync。",
|
||
"evidence_refs": [
|
||
"docs/evaluations/backup_dr_readiness_matrix_2026-06-04.json",
|
||
"scripts/backup/sync-offsite-backups.sh",
|
||
"scripts/backup/verify-offsite-full-sync.sh"
|
||
]
|
||
},
|
||
{
|
||
"rule_id": "backup_status_core_blocker",
|
||
"event_kind": "backup_core_blocker_detected",
|
||
"backup_state": "action_required",
|
||
"severity": "critical",
|
||
"decision": "escalate_immediate",
|
||
"channels": ["awooop_operator_event", "telegram_ops", "prometheus_textfile"],
|
||
"owner_agent": "openclaw",
|
||
"requires_incident": true,
|
||
"requires_approval_record": true,
|
||
"message_contract": "核心阻擋必須連到 incident / approval / evidence；Agent 不得自行 restore、prune、寫 marker 或改排程。",
|
||
"evidence_refs": [
|
||
"docs/evaluations/backup_dr_readiness_matrix_2026-06-04.json",
|
||
"docs/runbooks/BACKUP-STATUS.md"
|
||
]
|
||
},
|
||
{
|
||
"rule_id": "credential_escrow_missing_markers",
|
||
"event_kind": "credential_escrow_gap",
|
||
"backup_state": "blocked",
|
||
"severity": "high",
|
||
"decision": "create_action_required",
|
||
"channels": ["awooop_operator_event", "daily_status_summary"],
|
||
"owner_agent": "openclaw",
|
||
"requires_incident": false,
|
||
"requires_approval_record": true,
|
||
"message_contract": "缺 escrow marker 必須維持 action-required；不得自動寫 marker、不得輸出 credential 或要求 Agent 讀 secret。",
|
||
"evidence_refs": [
|
||
"docs/evaluations/backup_dr_readiness_matrix_2026-06-04.json",
|
||
"scripts/backup/mark-credential-escrow-verified.sh",
|
||
"scripts/backup/offsite-escrow-evidence-report.sh"
|
||
]
|
||
},
|
||
{
|
||
"rule_id": "metric_binding_gap",
|
||
"event_kind": "backup_metric_binding_gap",
|
||
"backup_state": "needs_metric_binding",
|
||
"severity": "warning",
|
||
"decision": "create_action_required",
|
||
"channels": ["awooop_operator_event", "daily_status_summary"],
|
||
"owner_agent": "hermes",
|
||
"requires_incident": false,
|
||
"requires_approval_record": false,
|
||
"message_contract": "metric binding gap 只建立 action-required 與 UI 證據缺口；不得直接修改 Prometheus rule 或 exporter。",
|
||
"evidence_refs": [
|
||
"docs/evaluations/backup_dr_readiness_matrix_2026-06-04.json",
|
||
"scripts/ops/backup-alert-label-contract-check.py"
|
||
]
|
||
}
|
||
],
|
||
"daily_summary_contract": {
|
||
"summary_time_taipei": "06:05",
|
||
"success_immediate_notifications_allowed": false,
|
||
"success_signal_sources": [
|
||
"Prometheus / node-exporter textfile metrics",
|
||
"scripts/backup/backup-status.sh --no-notify",
|
||
"Backup / DR readiness matrix"
|
||
],
|
||
"failure_rows_require_action_refs": true,
|
||
"mandatory_sections": [
|
||
"latest successful backup targets",
|
||
"warning / failed targets",
|
||
"blocked DR targets",
|
||
"offsite latest-only verification",
|
||
"credential escrow marker status",
|
||
"next operator action"
|
||
]
|
||
},
|
||
"agent_roles": [
|
||
{
|
||
"agent_id": "openclaw",
|
||
"role": "通知升級仲裁者，判斷 warning / failed / action-required 是否需要 incident、approval 與 operator action。",
|
||
"allowed_actions": [
|
||
"只讀仲裁嚴重度",
|
||
"要求 incident / approval evidence",
|
||
"拒絕成功即時洗版"
|
||
],
|
||
"blocked_actions": [
|
||
"未批准發送正式 Telegram 測試訊息",
|
||
"未批准執行 restore 或 backup",
|
||
"未批准寫 credential marker"
|
||
]
|
||
},
|
||
{
|
||
"agent_id": "hermes",
|
||
"role": "整理 runbook、每日摘要、降噪政策與 UI 可讀文字。",
|
||
"allowed_actions": [
|
||
"只讀整理通知政策",
|
||
"彙整 daily summary 欄位",
|
||
"標示 metric binding gap"
|
||
],
|
||
"blocked_actions": [
|
||
"直接送出 Telegram / AwoooP 訊息",
|
||
"直接改排程或 workflow",
|
||
"直接修改 Prometheus rule"
|
||
]
|
||
},
|
||
{
|
||
"agent_id": "nemotron",
|
||
"role": "可離線比較通知降噪 pattern 與摘要品質，但不是備份通知主控。",
|
||
"allowed_actions": [
|
||
"使用 sanitized evidence 做離線比較",
|
||
"提出摘要品質建議"
|
||
],
|
||
"blocked_actions": [
|
||
"讀取 production secret",
|
||
"送出通知",
|
||
"觸發 backup / restore / offsite sync"
|
||
]
|
||
}
|
||
],
|
||
"operation_boundaries": {
|
||
"read_only_policy_allowed": true,
|
||
"notification_send_allowed": false,
|
||
"backup_execution_allowed": false,
|
||
"restore_execution_allowed": false,
|
||
"offsite_sync_execution_allowed": false,
|
||
"credential_marker_write_allowed": false,
|
||
"schedule_change_allowed": false,
|
||
"workflow_write_allowed": false,
|
||
"telegram_test_message_allowed": false
|
||
},
|
||
"approval_boundaries": {
|
||
"sdk_installation_allowed": false,
|
||
"paid_api_call_allowed": false,
|
||
"shadow_or_canary_allowed": false,
|
||
"production_routing_allowed": false,
|
||
"destructive_operation_allowed": false
|
||
}
|
||
}
|