Alauda support for PostgreSQL

Monitoring

TOC

Overview

The monitoring data embedded in the panel can be used for PostgreSQL monitoring and alerting from the perspectives of resources, performance, and capacity, and it supports the configuration of notification policies.

The intuitively presented monitoring data can provide decision support for operational inspections or performance tuning.

Monitoring Metrics

The platform collects commonly used monitoring metrics related to the resource, performance, and capacity of PostgreSQL by default. In the instance's Monitoring tab, real-time monitoring data of these metrics can be viewed.

MetricDescription
CPU UtilizationThe percentage of CPU time used by PostgreSQL processes
Memory UsageThe amount of resident memory and virtual memory used by PostgreSQL processes
File DescriptorsThe number of file descriptors opened by PostgreSQL
Connection CountThe number of current active connections and the maximum connection limit
Transaction StatisticsThe number of committed and rolled back transactions for each database
Query StatisticsThe number of SELECT, INSERT, UPDATE, and DELETE operations
Cache Hit RateThe cache hit rate percentage for each database
Lock StatisticsThe number and waiting time of table-level locks
Temporary FilesThe size of temporary files created by each database
CheckpointsThe frequency of checkpoints and the volume of data written
Conflicts/DeadlocksThe number of conflicts and deadlocks that occur in each database
Background WritesStatistics on the buffer of the background writer

Main Monitoring Item Descriptions

Resource Usage

  • CPU Utilization: Reflects the CPU load of PostgreSQL processes
  • Memory Usage: Includes resident memory and virtual memory, monitoring for memory leaks and overuse
  • File Descriptors: Ensure that the system limits are not reached

Performance Metrics

  • Transaction Statistics: Tracks the commit/rollback rate for each database, assessing transaction success rates
  • Query Statistics: Analyzes the ratio of various types of queries to identify performance bottlenecks
  • Cache Hit Rate: Evaluates the efficiency of the shared buffer, which should maintain a hit rate above 95%
  • Lock Statistics: Detects contention for table-level locks

Capacity Planning

  • Connection Count: Monitors connection pool usage and adjusts max_connections in a timely manner
  • Temporary Files: Identifies large queries that need optimization
  • Checkpoints: Evaluates the impact of checkpoint frequency on performance
NOTE
  1. Some metrics require the statistics collector to be enabled
  2. It is recommended to set alert thresholds for critical metrics