Docs

RDS for PostgreSQL

Monitoring and alerts

Monitoring metrics

Click the Monitoring tab to access detailed information on the Resource Monitor, Engine Monitor, Deployment Monitor, and Agent Monitor for the instance. The monitoring metrics for each type of monitoring are shown in the table below.

tip

For the convenience of display, the monitoring alarm chart will automatically adjust the unit level according to the amount of data, and the specific display shall prevail. For example, when the replication delay is high, the unit is adjusted from μs to ms; when the replication delay is high, the unit is adjusted from ms to s.

You can call the GetMetricData interface to query each monitoring indicator data and alarm information. When calling the query interface, you need to obtain the Namespace of the product, the SubNamespace to which the index to be queried belongs, and the MetricName of the index to be queried. The RDS for PostgreSQL has VCM_RDS_PostgreSQL Namespace, and the MetricName and SubNamespace of each indicator are shown in the following table.

You can also get the detailed information of each monitoring metric in the Cloud Monitor Metrics.

Name	MetricName	SubNamespace	Unit	Description of indicators
CPU utilization	CpuUtil	resource_monitor	Percent	With idle-time overcommitment allowed, CPU utilization may exceed 100%.
Memory utilization	MemUtil	resource_monitor	Percent	Overuse during idle time is allowed, and memory utilization may be greater than 100%.
Disk utilization	DiskUtil	resource_monitor	Percent	The ratio of the used disk space to the total disk space of the instance.
User CPU utilization	UserCPUUtil	resource_monitor	Percent	User CPU utilization is one of the components of CPU usage.
System CPU utilization	SysCPUUtil	resource_monitor	Percent	System CPU utilization is one of the components of CPU utilization.
RSS memory usage	RSSMemUsage	resource_monitor	Bytes(SI)	The physical memory size actually occupied by the instance.
Cache memory usage	CacheMemUsage	resource_monitor	Bytes(SI)	Amount of memory used by the file system cache.
network input traffic	NetworkReceiveThroughput	resource_monitor	Bytes/Second(SI)	Input traffic counted by the PostgreSQL kernel.
network output traffic	NetworkTransmitThroughput	resource_monitor	Bytes/Second(SI)	Output traffic counted by the PostgreSQL kernel.
Disk usage	DiskUsage	resource_monitor	Bytes(SI)	The total amount of disk space used.
Data file disk usage	BaseDiskUsage	resource_monitor	Bytes(SI)	The disk size occupied by the data resource in the PostgreSQL instance.
WAL log disk usage	WalDiskUsage	resource_monitor	Bytes(SI)	The disk space occupied by the WAL logs in the PostgreSQL instance.
run log disk usage	LogDiskUsage	resource_monitor	Bytes(SI)	The disk size occupied by the running logs in the PostgreSQL instance, including error logs and slow logs, is used as the source of audit data.
Temporary file disk usage	TempFileDiskUsage	resource_monitor	Bytes(SI)	The size of the disk occupied by the pgsql_tmp directory in the PostgreSQL instance. The pgsql_tmp directory is used to store the temporary data generated during the execution of SQL.
Maximum WAL log accumulation in replication slots	SlotMaxWALAccumulation	resource_monitor	Bytes(SI)	The maximum amount of WAL logs accumulated in all replication slots.
Disk read-write state	DiskReadWriteStatus	resource_monitor	Count	Current disk read-write status, 0 means read-write, otherwise it is not 0.
Database size tip Only used to configure alarm strategies and does not provide dashboard display.	DBSize	resource_monitor	Bytes(SI)	The disk space occupied by the data resources in the database.
Database bloat rate tip Only used to configure alarm strategies and does not provide dashboard display.	DBExpansionRate	resource_monitor	Bytes/Second(SI)	The rate at which the disk space used by data resources in the database bloats.
Inactive transactions tip Only used to configure alarm strategies and does not provide dashboard display.	InactiveSlotCount	resource_monitor	None	The number of replication slots that are inactive.
Size of the largest table in the database tip Only used to configure alarm strategies and does not provide dashboard display.	DBMaxTableSize	resource_monitor	Bytes(SI)	The size of the largest table in the database.
TPS	TPS	engine_monitor	Count/Second	The number of transactions processed per second in a PostgreSQL instance.
Connection utilization	ConnectionUsedUtil	engine_monitor	Percent	The percentage of the maximum number of connections used by all connections for the current instance.
Total connections	TotalConnection	engine_monitor	Count	The number of all connections to the current instance, both active and inactive.
Slow queries per second	SlowQueries	engine_monitor	Count/Second	The number of queries per second that take longer than the duration set by the log_min_duration_statement parameter.
Slow queries per 30 seconds	SlowQueriesIncrement	engine_monitor	Count	The number of queries per 30 seconds that exceeded the duration set by the log_min_duration_statement parameter.
Slow queries in the last n seconds	SlowQueriesCustomIncrement	engine_monitor	Count	The number of slow queries that occurred in the past n seconds. The value of n can be adjusted by modifying the running value of the parameter slow_query_custom_count_range.
Instantaneous slow queries	InstantaneousSlowQueries	engine_monitor	Count	The number of slow queries currently queried from the kernel that take longer than the query duration set by the log_min_duration_statement parameter.
Error queries per second	ErrorQueries	engine_monitor	Count/s	The number of queries with errors per second.
Incremental trend of error queries every 30 seconds	ErrorQueriesIncrement	engine_monitor	Count	The number of erroneous queries that occur within every 30 seconds.
Rows inserted per second	RowsInserted	engine_monitor	Count/Second	Number of rows inserted per second.
Rows deleted per second	RowsDeleted	engine_monitor	Count/Second	The number of rows deleted per second.
RowsUpdated	RowsUpdated	engine_monitor	Count/Second	The number of rows updated per second.
Current active connections	ActiveConnection	engine_monitor	Count	Total number of client processes currently executing in the database.
Free connections	IdleConnection	engine_monitor	Count	Total number of client side processes requested but not used in the current database.
Inactive transactions	IdleInTxnConnection	engine_monitor	Count	The total number of unused client threads in the current database transaction.
Transaction commits per second	CommitsPerSecond	engine_monitor	Count/Second	The number of times transactions are committed per second in a PostgreSQL instance.
Transaction rollbacks per second	RollbacksPerSecond	engine_monitor	Count/Second	The number of times transactions are rolled back per second in a PostgreSQL instance.
Longest transaction execution time	LongestTransactionExecTime	engine_monitor	Second	The longest statement execution time among the currently executing queries.
Queries with lock wait times exceeding 1 hour	LocksUngrantedCountOver1h	engine_monitor	Count	The number of queries in the PostgreSQL instance that have been waiting for locks for more than one hour.
2pc transactions	PreparedXactsCount	engine_monitor	Count	The number of open two-phase commit transactions in the database.
Maximum duration of a 2pc transaction	OldestPreparedXactsDuration	engine_monitor	Second	The maximum duration of a two-phase transaction.
Database maximum age XIDs	DbMaxAgeXids	engine_monitor	None	The maximum age of the database in the current instance. If the value is too large, it will cause the transaction to be rolled back, and the instance cannot be written during the transaction rollback.
Global read-only state tip Only used to configure alarm strategies and does not provide dashboard display.	ReadOnlyStatus	engine_monitor	None	Whether the current instance has enabled global read-only parameters and is in a non-writable state.
Locks waiting to be released	LocksUngrantedCount	engine_monitor	Count	The number of locks waiting to be released in the database.
Lock holdings	LocksGrantedCount	engine_monitor	Count	The number of locks that have been granted for the database.
Database deadlocks	DeadLocksCount	engine_monitor	Count	The current number of deadlocks in the database.
Cache hits per second	SharedBuffersBlksHit	engine_monitor	Count/Second	The number of times a disk block was found in the buffer without a disk read, in database.
Cache hit rate	SharedBuffersBlksHitRate	engine_monitor	Percent	The percentage of times that disk blocks were found in the buffer.
Ratio of data return to scan	DBFetchedReturnedRatio	engine_monitor	Percent	The database returns the ratio of the number of rows (tup_fetched) to the number of rows scanned (tup_returned). The larger the value, the more valid data is returned; the smaller the value, the more likely there is SQL that is not using an index.
Watches per second	RowsFetched	engine_monitor	Count/Second	The number of times the index matches the item.
Rows scanned per second	RowsReturned	engine_monitor	Count/Second	The number of rows scanned per second of the Heap table/Index table.
Transaction success rate	TransactionSuccessRate	engine_monitor	Percent	Transaction success rate per second.
Replication delay	ReplicationDelay	deploy_monitor	Second	The delay time of primary-secondary replication.
The number of transaction conflicts per second.	DBConflicts	deploy_monitor	Count/Second	The number of transaction conflicts per second.
Primary-Secondary writeLag	ReplicationWriteLag	deploy_monitor	Second	The delay between when the primary server's WAL records have been flushed to disk and when the secondary server receives the WAL data but has not yet written it to its own operating system cache.
Primary-Secondary flushLag	ReplicationFlushLag	deploy_monitor	Second	The delay during which WAL on the primary has been sent to the secondary, but the standby has not yet flushed (fsynced) it to its own disk.
WAL files to be archived	UnarchivedWALSegmentsCount	deploy_monitor	Count	The number of WAL files that are in the ready state but have not yet been archived.
Proxy QPS	ProxyQPS	proxy_monitor	Count/Second	The number of SQLs executed per second by the database (including INSERT, SELECT, UPDATE, and DELETE).
Proxy CPU utilization	ProxyCPUUtil	proxy_monitor	Percent	The CPU utilization of the proxy node.
Proxy memory utilization	ProxyMemUtil	proxy_monitor	Percent	The memory utilization of the proxy node.
Proxy connections	ProxyConnection	proxy_monitor	Count	The number of open connections.

tip

For setting log_min_duration_statement parameters, see Modifying parameters.

If there are no monitoring indicators of User mode CPU usage and System state CPU usage in the monitoring board of your instance, submit a ticket to apply for an upgrade of the instance version.

Last updated: 2025.11.17 16:15:08