Skip to main content

pocketPG documentation

pocketPG is an agentless PostgreSQL diagnostics tool. It connects to a PostgreSQL instance, reads system views and settings, and explains what the current state means for performance and operations.

This documentation is written as a technical reference. The goal is to explain:

  • what each part of the product does
  • which PostgreSQL views or settings it depends on
  • what the results mean
  • where the limits are

What pocketPG is good at​

pocketPG is strongest in these areas:

  • health and incident-oriented monitoring
  • query analysis from pg_stat_statements
  • WAL, checkpoints, and replication pressure
  • vacuum, bloat, and wraparound risk
  • configuration recommendations with evidence and rollback SQL

It is not a full monitoring platform, failover orchestrator, or compliance tool.

What pocketPG reads​

Most features are built from standard PostgreSQL catalog and statistics views, including:

  • pg_stat_database
  • pg_stat_activity
  • pg_stat_statements when installed
  • pg_stat_user_tables
  • pg_stat_user_indexes
  • pg_stat_wal
  • pg_settings
  • selected pg_catalog views and helper functions

Some pages become more useful when specific extensions or privileges are available. The documentation for each page calls that out explicitly.

How to use the docs​

If you are new to pocketPG, read in this order:

  1. Connecting to your database
  2. Overview tab
  3. Health dashboard
  4. Queries tab
  5. How recommendations work

If you already use the product and want a specific topic, go straight to the matching section in the sidebar.

Core documentation areas​

Getting started​

Use this section for connection methods, provider-specific setup, and shared-instance behavior.

Monitoring​

Use this section to understand the dashboard pages that surface operational risk.

Query analysis​

Use this section for pg_stat_statements, query interpretation, and plan analysis.

Tuning​

Use this section for how recommendations are formed and how to interpret parameter changes safely.

Vacuum and bloat​

Use this section for MVCC cleanup pressure, dead tuples, wraparound, and index waste.

Operations​

Use this section for maintenance actions, alerting, digests, and the printable report.

Documentation style​

These docs try to keep two rules:

  1. Prefer direct explanations over product language.
  2. State uncertainty clearly when pocketPG cannot prove something from SQL alone.

That means some pages say "this is a strong signal" and others say "this is only a heuristic." That distinction is deliberate.