Custom metric monitoring
To create a custom view of your database or compare the performance of multiple instances, use a monitoring dashboard. You can select the instances and metrics you want to monitor, compare performance between instances, and correlate different metrics to get a comprehensive overview of your database's health.
Prerequisites
Your ApsaraDB RDS for PostgreSQL instance must be the High-availability Edition.
Database Autonomy Service (DAS) has been rolling out the monitoring dashboard feature for ApsaraDB RDS for PostgreSQL since May 20, 2022.
Add a monitoring dashboard
Log in to the ApsaraDB RDS console.
In the left-side navigation pane, click Maintenance & Monitoring > Performance Monitoring.
On the Performance Monitoring page, click the Monitoring Dashboard tab.
On the tab for your target database engine, click Add Monitoring Dashboard.
Enter a name for the New Monitoring Dashboard, and click OK.
Click Select Instances and Metrics, select the target instances and metrics, and then click
.NoteYou can monitor up to 32 instances.
For more information about the metrics, see Metrics.
In the Select Instances and Metrics dialog box, select the target instances. In the Select Metrics area, select the PostgreSQL metrics to monitor (including host metrics and basic PostgreSQL monitoring). Then click Confirm.
Click Confirm.
NoteTo modify the instances or metrics on the dashboard, click Add Instances and Metrics.
View metric trends
-
Log on to the ApsaraDB RDS console.
-
In the left-side navigation pane, choose Maintenance & Monitoring > Performance Monitoring.
-
Click the Monitoring Dashboard tab.
-
Click the tab for the database engine, select the monitoring dashboard you want to view, then set a time range to display metric trend charts for that period.
The end time must be later than the start time, and the time range cannot exceed seven days.
The following options are available on the dashboard:
| Option | Description |
|---|---|
| Instance filtering | Filter by one or more RDS instances to compare their metrics side by side. |
| Auto Refresh (Every 5 Seconds) | Automatically refresh all trend charts every 5 seconds — useful for monitoring active workloads in real time. |
| Linkage Chart | Synchronize chart cursors so you can see the values of different metrics at the same point in time, making it easier to correlate anomalies across metrics. |
| Chart Layout | Set the number of trend charts displayed per row. |
| Add Instances and Metrics | Add or change the instances and metrics shown in the dashboard. |
| Details (in a trend chart) | Expand the chart and adjust the time range for a closer look at a specific metric. |
| Delete (in a trend chart) | Remove the chart from the dashboard. |
Metrics
Category | Parameter | Description |
host metrics | cpu_sys_ratio | The CPU utilization of the PostgreSQL process in kernel mode. |
cpu_user_ratio | The CPU utilization of the PostgreSQL process in user mode. | |
cpu_usage | The CPU utilization of the PostgreSQL process. | |
mem_usage | The memory usage of the PostgreSQL process. | |
data.r_s | The read IOPS for the disk containing the data directory. | |
data.iops | The IOPS for the disk containing the data directory. | |
data.w_s | The write IOPS for the disk containing the data directory. | |
local_fs_size_total | The total disk space. | |
data.fs.used | The used space on the disk containing the data directory. | |
data.fs.usage | The disk usage of the disk containing the data directory. | |
network_out_io | The outbound network traffic of the PostgreSQL process. | |
network_io | The network traffic of the PostgreSQL process. | |
network_in_io | The inbound network traffic of the PostgreSQL process. | |
basic PostgreSQL monitoring | waiting_connection | The number of waiting connections. |
idle_in_transaction | The number of sessions that are idle within a transaction. | |
active_session | The number of active sessions. | |
total_session | The total number of sessions. | |
tup_inserted | The number of rows inserted per second. | |
tup_deleted | The number of rows deleted per second. | |
tup_updated | The number of rows updated per second. | |
tup_returned | The number of rows returned by sequential scans per second. | |
tup_fetched | The number of rows fetched by index scans per second. | |
tps | The number of transactions per second (TPS). | |
xact_rollback | The number of transactions rolled back per second. | |
xact_commit | The number of transactions committed per second. | |
deadlocks | The number of deadlocks per second. | |
oldest_snapshot | The oldest transaction snapshot point. Dead tuples created before this point can be reclaimed. | |
max_sent_delay | The data send latency on the secondary RDS instance. | |
max_replay_delay | The data replay latency on the secondary RDS instance. | |
long_query_5s | The number of SQL queries that run for 5 seconds or longer. | |
long_query_1s | The number of SQL queries that run for 1 second or longer. | |
long_query_3s | The number of SQL queries that run for 3 seconds or longer. | |
long_idle_in_transaction_3s | The number of transactions that are idle for 3 seconds or longer. | |
long_idle_in_transaction_5s | The number of transactions that are idle for 5 seconds or longer. | |
long_idle_in_transaction_1s | The number of transactions that are idle for 1 second or longer. | |
long_2pc_1s | The number of two-phase commit (2PC) transactions running for 1 second or longer. | |
long_2pc_5s | The number of two-phase commit (2PC) transactions running for 5 seconds or longer. | |
long_2pc_3s | The number of two-phase commit (2PC) transactions running for 3 seconds or longer. |
You can also view a metric's description by clicking the 
icon next to its name on the monitoring dashboard.