This Release
pgtelemetry 1.0.0
Other Releases
Useful monitoring and metrics views for PostgreSQL
pgtelemetry is a suite of views, functions and tables which gather and produce a variety of useful metrics for monitoring health of clusters including locks, state of queries, query plans, object size, replication health, and WAL telemetry.
Released By
Special Files


pgtelemetry 1.0.0
Useful monitoring and metrics for PostgreSQL


pg_telemetry, an extension for getting usage states from PostgreSQL

Welcome to pg_telemetry, an extension for monitoring usage of PostgreSQL.

This module provides functions for deriving usage stats from system stats and other system administration functions. It aims to provide a generally useful and reusable series of metrics functions for monitoring purposes.

Most of the data is set to export numbers and can be used to supply SNMP agents, Zabbix agents, and other monitoring programs with data.

Intended Usage

The module is intended to be used with monitoring and visualizaiton programs such as pganalyze, zabbix, and munin. These tend to use an agent which collects performance data and sends it to a server for graphics generation. Some of the functions here are wrapped in views.

A second class of functions are in place for administrators to use in troubleshooting and debugging performance issues. However these are to be used as needed while the general stats functions are assumed to be run every few min at least.

We expect this extension usually to be installed in its own schema.
However, it can be safely installed anywhere its names don't conflict with anything else.

Areas of focus

In the initial phase, there are several areas of focus for this project:

  • Disk usage
  • Sources of concurrent queries
  • WAL throughput
  • Replication Monitoring


Currently this is expected to require PostgreSQL 9.6 and PostgreSQL 10 will be added in the near future. However support for PostgreSQL instances before 9.6 is an area of interest and may be added anyway.

This module also requires that pg_stat_statements is installed.

Documentation of monitoring views, relations, and functions

See the pgtelemetry.html in the doc folder.


pg_stats_statements must be installed and preloaded. Once that is the case, you can use the standard make/make install process to install this extension:

make install

That copies the files into the PostgreSQL extensions directory. In some cases (Windows) you may need to install appropriate development tools such as MinGW and on Linux in some cases you may need the development libraries.

After that, in the database you want to base your monitoring on:

create extension pgtelemetry;

This will create the pgtelemetry schema and place all the objects there.


In the extras directory there are a number of important integration examples.


  • queries.yaml includes a basic yaml for pulling stats into Prometheus
  • queries_wal.yaml is a basic wal telemetry yaml that can be run on master databasess


  • raises an alarm if any tablespace is too big
  • raises an alarm if queries have been running too long.
  • raises an alarm if too many queries are waiting on locks and latches.
  • raises an alarm if replication lag is too high (checks downstream).

What is too big? To high? Too long? These are set in the beginning of the scripts making this easy to integrate with Zabbix.