feat(docker): PP-04 可观测性 MVP — Alertmanager 告警出口 + Grafana provisioning

PP-04 核实属实:11 条告警规则在 prometheus 加载但无 alertmanager(告警
无通知出口),grafana provisioning 目录空,exporter 服务也未部署
("配置齐全运行为零")。

MVP 打通告警链路 + 让 grafana 可用(不依赖 exporter,基于 app metrics):
- docker-compose.production.yml 加 alertmanager 服务 + alertmanager_data 卷
- prometheus.yml 加 alerting 指向 alertmanager:9093
- alertmanager/config.yml 路由(SEV-1 critical 即时通知 + 分组)
- grafana/provisioning/datasources 自动连 prometheus
- grafana/provisioning/dashboards provider 就绪

待办(上线前):① alertmanager 占位 webhook 替换为真实渠道(钉钉/企微/邮件)
② 补 grafana dashboard JSON ③ 部署 postgres/redis/nginx exporter 让 prometheus 抓得到
This commit is contained in:
iven
2026-06-26 09:25:43 +08:00
parent 3351c68d10
commit 6457c53d9c
5 changed files with 86 additions and 0 deletions

View File

@@ -0,0 +1,15 @@
# Grafana dashboard provider
# 自动加载 ./json/ 下的 dashboard JSON 文件PP-04 后续补充 HMS 概览 dashboard
apiVersion: 1
providers:
- name: "HMS Dashboards"
orgId: 1
folder: ""
type: file
disableDeletion: false
editable: true
updateIntervalSeconds: 30
options:
path: /etc/grafana/provisioning/dashboards/json
foldersFromFilesStructure: false

View File

@@ -0,0 +1,13 @@
# Grafana 数据源自动 provisioning
# 启动时自动注册 Prometheus 数据源,无需手动在 UI 配置
apiVersion: 1
datasources:
- name: Prometheus
type: prometheus
access: proxy
url: http://prometheus:9090
isDefault: true
editable: false
jsonData:
timeInterval: "15s"