Monitoring and alerts
Grid’s console Monitoring and Alerts pages read from host Prometheus and Alertmanager. Those processes are not the Compose app containers.
Operator install
See Monitoring (install) and grid-core/install/monitoring.
Required production env:
GRID_PROMETHEUS_URL=…
GRID_ALERTMANAGER_URL=…
GRID_MONITORING_CONFIG_DIR=…
Console behavior
| Page | Role |
|---|---|
| Monitoring | Unit-oriented metrics / health using control-plane join keys |
| Alerts | Firings from Alertmanager |
| Settings (where exposed) | Write rules + alertmanager.yml, then reload |
Collectors on managed infra can be bootstrapped via API (monitoring/setup/:id) to remote_write into the host Prometheus.
Join key
Metrics are correlated with inventory using grid_unit_id (path-derived UUID by default). Stable paths ⇒ stable ids across control-plane reinstalls.