Files
awoooi/docs/security/MONITORING-ALERTING-OBSERVABILITY-INVENTORY.md

7.6 KiB
Raw Permalink Blame History

IwoooS Monitoring / Alerting / Observability repo-only 清冊

項目 內容
日期 2026-06-12
狀態 repo_only_inventory_ready
工具 scripts/security/monitoring-alerting-observability-inventory.py
Snapshot docs/security/monitoring-alerting-observability-inventory.snapshot.json
Schema docs/schemas/monitoring_alerting_observability_inventory_v1.schema.json
runtime gate 0

1. 目的

這份清冊把 Prometheus、Alertmanager、Grafana、SigNoz、Sentry、Langfuse、OTEL、Telegram / notification policy、deploy / reload scripts 與 alert chain smoke scripts 集中納入 IwoooS 高價值配置控管。

本階段仍是 repo-only source inventory只讀已提交檔案、計算 SHA256、整理 owner gate 與 live evidence 缺口。不連 live Prometheus、不 reload Alertmanager、不改 Grafana、不套用 SigNoz rule、不部署 Sentry、不發 Telegram、不建立 silence、不 SSH、不 kubectl、不讀 secret value。

2026-06-14 已新增 docs/security/MONITORING-OWNER-REQUEST-DRAFT.mddocs/security/monitoring-owner-request-draft.snapshot.json,將 60 個 surface 轉成 owner request draft。此更新只代表人工送件前欄位包已完成request sent、owner response received / accepted、live evidence、reload、receiver route change、Telegram send、alert chain smoke、runtime gate 與 action button 仍全部為 0 / false

2. 覆蓋摘要

指標 目前值 說明
surface 60 全部為 committed repo source
source exists 60 每個 source path 都存在並有 SHA256
Prometheus config surface 8 基礎設定、remote write、generated target、service registry、exporter query
alert rule surface 13 Prometheus、K8s、SLO、Ollama、app alert rule 與 Grafana alert rule
Alertmanager receiver surface 1 route / receiver / grouping source
Grafana surface 6 alert rule 與 dashboard JSON
SigNoz surface 3 alert rule、log rule、API client
Sentry surface 4 compose、deploy、webhook receiver、API client
Langfuse surface 3 compose、runbook、API client
notification policy surface 4 failure-only policy、backup policy、observability contract、notification matrix
Telegram surface 3 digest policy、receipt package、gateway service
OTEL surface 1 SigNoz OTEL collector
deploy / reload surface 6 Alertmanager / Prometheus / Sentry / exporter deploy 或 reload-capable script
drift guard surface 1 Prometheus rule drift guard
smoke surface 4 live / test alert 與 alert chain smoke script
write-capable surface 11 可能 reload、deploy、send notification、fire alert 或 restart exporter
owner response received / accepted 0 / 0 不得假性拉高
live evidence received 0 尚未驗證 live monitoring truth
reload owner / receiver owner / route smoke accepted 0 / 0 / 0 尚未授權 reload、route change 或 live smoke
runtime gate / action button 0 / 0 不得建立前端執行入口
P1-4 成熟度 56% -> 62% 只代表 repo-only 清冊完成,不代表 live alert chain 通過

3. Write-capable surface

surface 風險 必要 gate
deploy_alertmanager_config_script 可影響 receiver / route / reload owner response、維護窗口、rollback owner、receiver smoke
deploy_prometheus_alerts_script 可影響 alert rules / reload rule test、receiver mapping、reload owner、rollback owner
k8s_deploy_prometheus_config_script 可觸發 K8s apply / reload K8s owner、ArgoCD / kubectl 邊界、rollback owner
api_apply_prometheus_config_script 可由 API side 套用 Prometheus config API deploy owner、config source owner、reload proof
monitoring_exporter_deploy_script 可透過 host deploy 影響 exporter / restart host owner、SSH boundary、restart window
sentry_self_hosted_deploy 可部署 Sentry self-hosted stack deploy owner、backup owner、migration rollback
telegram_gateway_service 可送 Telegram / 寫 delivery path token injection owner、receipt owner、send approval gate
notification_manager_service 可改通知通道與路由 channel owner、receipt owner、failure-only policy
converged_alert_recurrence_notifier 可造成 recurrence escalation 或噪音 noise budget owner、silence boundary、receipt proof
fire_live_alert_script 可向 live alert chain 打告警 allowed receiver、test window、cleanup owner
fire_test_alert_script 可觸發測試告警與通知 dedup proof、receiver owner、noise guard

4. 固定 0 / false 邊界

runtime_execution_authorized=false
host_write_authorized=false
prometheus_reload_authorized=false
alertmanager_reload_authorized=false
grafana_dashboard_apply_authorized=false
signoz_rule_apply_authorized=false
sentry_deploy_authorized=false
langfuse_config_change_authorized=false
otel_collector_reload_authorized=false
receiver_route_change_authorized=false
silence_policy_change_authorized=false
telegram_send_authorized=false
notification_route_change_authorized=false
webhook_receiver_change_authorized=false
remote_write_change_authorized=false
exporter_deploy_authorized=false
live_alert_fire_authorized=false
alert_chain_smoke_authorized=false
ssh_read_authorized=false
ssh_write_authorized=false
kubectl_action_authorized=false
secret_value_collection_allowed=false
active_scan_authorized=false
action_buttons_allowed=false

5. 下一步 owner response

  1. Prometheus owner提供 live config hash、rule diff、reload owner、rollback owner 與 false-green guard。
  2. Alertmanager owner提供 receiver diff、route owner、silence policy owner、reload owner 與 failure-only notification proof。
  3. Grafana owner提供 dashboard UID / folder owner、import owner、rollback ref 與 smoke plan。
  4. SigNoz / OTEL owner提供 pipeline diff、rule apply owner、data export boundary 與 rollback owner。
  5. Sentry / Langfuse owner提供 compose live hash、secret injection owner、upgrade / restart window、backup owner 與 rollback owner。
  6. Telegram / notification owner提供 receiver owner、receipt owner、redaction proof、retry boundary 與 no-secret-value evidence。
  7. Alert chain smoke owner提供 allowed receiver、execution window、expected receipt、noise budget 與 cleanup owner。

6. 完成度

工作 完成度 說明
P1-4 repo-only surface 註冊 100% 60 個 surface 已納入 snapshot
source existence / SHA256 100% 60 / 60 source path 存在
monitoring / alerting 高價值配置成熟度 56% -> 62% 只代表清冊與 guard 準備度
owner request draft artifact 100% 60 份 request draft 已完成;不代表已送件或已授權
owner response 收件 / 接受 0% 尚未收到或接受任何 owner response
live evidence collection 0% 未讀 live monitoring stack
reload / receiver / route smoke gate 0% 未授權、未執行
runtime gate 0% 無前端執行按鈕

7. 邊界

本清冊完成不代表 live Prometheus / Alertmanager / Grafana / SigNoz / Sentry / Langfuse 已一致,也不代表 alert route 已送達或告警必到。不得把 repo source 可見、snapshot、IwoooS UI、AwoooP approval 或 LOGBOOK 當成 Prometheus reload、Alertmanager reload、Grafana import、SigNoz apply、Sentry deploy、Langfuse change、OTEL reload、remote write change、silence change、Telegram send、live alert fire、alert chain smoke、SSH、kubectl、active scan 或 secret collection 授權。