Architecture

• 基础设施包括平行链节点，容器，中心化数据库通过对应的exporter走prometheus统一接入,再由AlertManager将告警接入slack通知，这种数据可视化主要运维使用，定制化要求不高，有些开源的grafana dashboard模板可以直接拿过来用

• 业务数据分析可以定义一些数据库层面的视图函数，引入了一个postgraphile组件可以基于数据库自动生成graphql查询接口供前端使用

• 业务统计数据的趋势分析，可以定期存储下来做成时序数据存储到相应的历史数据表，也可以通过postgresql-exporter组件expose到prometheus

• 引入了一个pgsync组件实时同步pg数据到es

• 时序业务数据可以直接在grafana上做数据可视化，通过iframe嵌入App。定制化要求高数据面板视图前端走es和graphql接口自行构建

• Since there are many components(Subql app, Postgresql, Elasticsearch, Grafana, Pgsync, Postgraphile and Prometheus, ...) in this stack we may use docker-compose to deploy and move to some container orchestration tool like k8s for future

• some predefined dashboards will be automatic provisioned in Grafana accessed and viewed via the Grafana UI

• pgsync

we need to define rules to mapping data in postgresql to elasticsearch https://github.com/bifrost-finance/bifrost-subql/blob/pgsync/pgsync/schema/salp.json

• postgraphile

we need to define views&functions in postgresql and postgraphile will turn it into graphql endpoints automaticlly

• postgreql-exporter

we need to define rules to mapping data in postgresql to prometheus

• alert rules

we need to define alert rules in prometheums

we can also define alert rules in grafana directly

• forward alert to slack

we need to add a slack channel and with the predefind notifiers alert will be forwarded to slack channel

• cadvisor

this component is for monitoring the docker containers

Deploy

• build

make build

• start

make start