A small IT team usually finds out about a problem from the person it happened to. Monitoring exists, but it covers whatever someone remembered to set up, the network gear sits in a separate tool or none at all, and the logs that would explain the problem are spread across a dozen machines. ResorsIT treats monitoring as part of the platform rather than an add-on, so coverage grows as the Site does.
Collecting From Everything
Telegraf runs on each monitored host and writes metrics to InfluxDB, the time-series store. When a package installs, it declares the endpoints it wants scraped and the installer adds them to Telegraf’s configuration, so a service is monitored from its first minute. Fluent Bit collects logs from the same services and hosts and writes them to OpenSearch, which gives the Site one searchable place for everything its services and hosts emit.
Network Devices on the Same Footing
Switches, firewalls and access points run their own firmware and cannot host an agent, so they are monitored from the discovery host that already reaches them. Interface and health metrics are polled over SNMP into InfluxDB, each device’s syslog stream goes through Fluent Bit into OpenSearch, and its SNMP traps feed the same alerting path as everything else. A failing switch port and a failing server raise alerts the same way.
Dashboards and Alerts
Grafana sits over both stores, so one dashboard can show a service’s metrics beside its logs, and packages bring their own dashboards when they install. AlertManager decides what becomes of each alert: routing it by label, grouping related alerts into one notification, silencing alerts during known work, and delivering them to chat or a notification service. Karma gives one view of what is firing, with acknowledgement built in. On-call schedules and escalation are left to a dedicated service, PagerDuty commercially or GoAlert in open source, so each Site chooses the one that fits its rota.
For simple uptime checks of websites and services, ResorsIT also offers Uptime Kuma.