Monitoring and Alarms

Monitoring and alarms track time-series data, such as resource utilization and capacity metrics, as well as predefined system events. The notification service (SNS) sends alarm messages to configured endpoints. Supported alarm types include resource alarms and event alarms. Endpoints include the system message center, email, DingTalk, WeCom, Lark, Webhook, SMS, Microsoft Teams, and SNMP Trap. Some resource alarms require an agent.

Prometheus provides time-series monitoring. It aggregates metrics from different sources and collects them through a unified service.

In this architecture, the Prometheus server collects and stores data and provides query access instead of directly generating metrics for each target. Exporters periodically collect samples such as host CPU utilization. ZSphere uses both pull and push modes according to the monitoring target. For hosts and metrics collected outside VMs, Prometheus periodically pulls the data gathered by exporters. Network or security restrictions can prevent Prometheus from directly reaching services inside VMs . In that case, a Pushgateway relays the data: exporters collect the metrics and periodically push them to the Pushgateway, and Prometheus then pulls the data from the Pushgateway for unified storage and querying.