Metrics and system events-AnalyticDB(AnalyticDB)-阿里云帮助中心

AnalyticDB for PostgreSQL integrates with CloudMonitor to expose metrics and system events for monitoring instance health and triggering alerts.

Metrics

The tables below list the metrics available for each storage mode, along with recommended alert settings. To configure alerts based on these metrics, see Configure an alert rule. To reuse alert settings across instances, see Configure an alert template.

Elastic storage mode

Metric	Unit	Description	Recommended alert period	Recommended threshold
`adbpg_conn_count`	N/A	Number of active connections to the instance.	1-minute period, continue for 5 periods	Average >= 90% of the maximum allowed connections. For example, a 2-core 16 GB instance allows up to 550 connections — set the threshold to 495. For maximum connection limits by instance type, see Limits.
`node_cpu_used_percent`	%	CPU usage percentage of a node.	1-minute period, continue for 5 periods	Average >= 90%
`node_disk_iops`	N/A	Disk I/O operations per second of a node.	None	—
`node_disk_used_percent`	%	Disk usage percentage of a node.	1-minute period, continue for 5 periods	Max >= 80%
`node_mem_used_percent`	%	Memory usage percentage of a node.	1-minute period, continue for 5 periods	Average >= 90%

Reserved storage mode

Metric	Unit	Description	Recommended alert period	Recommended threshold
`ConnectionUsage`	%	Connection usage as a percentage of the maximum allowed connections.	1-minute period, continue for 5 periods	Average >= 90%
`CpuUsage`	%	CPU usage percentage of the instance.	1-minute period, continue for 5 periods	Average >= 90%
`DiskUsage`	%	Disk usage percentage of the instance.	1-minute period, continue for 5 periods	Average >= 80%
`IOPSUsage`	%	IOPS usage as a percentage of the maximum IOPS.	1-minute period, continue for 5 periods	Average >= 90%
`MemoryUsage`	%	Memory usage percentage of the instance.	1-minute period, continue for 5 periods	Average >= 90%

System events

All AnalyticDB for PostgreSQL system events are of type Exception, status Failed, and level Critical. Each event triggers when the monitored resource crosses its defined threshold.

Event	Description
`ComputeGroupCpuHigh`	CPU usage of a compute group exceeds 90%.
`ComputeGroupMemoryHigh`	Memory usage of a compute group exceeds 85%.
`HighCoumputeGroupDiskUsed`	Disk usage of a compute group exceeds 80%.
`InstanceDown`	The instance is unavailable.
`LongTransaction`	The number of long-running transactions is greater than or equal to 5.
`MasterCpuHigh`	CPU usage of the master node exceeds 90%.
`MasterMemoryHigh`	Memory usage of the master node exceeds 85%.
`NonSuperConnection`	The proportion of connections with an instance exceeds 80%.