At a glance
Connections In Use is the live count of backend connections currently open to your PostgreSQL instance, read in real time. It is the raw numerator behind connection-pool saturation: how many of your finite max_connections slots are occupied right now. Watching it against your configured ceiling tells a platform team how much connection headroom is left before clients start being turned away.
What it tracks
The card reports Connections In Use for the selected period, sourced frompg_stat_activity (one row per backend) measured against the instance’s max_connections setting. Every active query, idle session, idle-in-transaction backend, and replication connection occupies a slot, so this figure is the live occupancy of your connection capacity. It is a real-time (RT) reading rather than an aggregated window, so it reflects the count at the moment of sampling. Because the number on its own does not say whether you are close to the limit, read it next to Connection Pool Saturation % (the same count expressed as a percentage of max_connections) and Idle-in-Transaction Backends (the slots being wasted by stuck transactions). A steadily climbing in-use count with flat traffic is a classic sign of a client leaking connections, and a count near the ceiling is what precedes Connection Errors (24h). This is a Capacity card with no alert threshold of its own; the alerting lives on the saturation and error cards.
Reconciling against the source
Reconcile against PostgreSQL’s own tooling withSELECT count(*) FROM pg_stat_activity; compared to SHOW max_connections;. On managed services, cross-check the RDS / Aurora DatabaseConnections CloudWatch metric, the Cloud SQL database/postgresql/num_backends metric, or the Azure Monitor active_connections metric.