Heroku PostgreSQL Monitoring

The Heroku PostgreSQL Visibility Gap

Heroku provides essential database metrics: row count, table size, cache hit ratio, and connection count. When an application slows down, the operational questions get sharper: Which queries are consuming the most time? Which tables need indexes? Is autovacuum keeping up? Are you approaching connection limits?

MonPG bridges this gap. By deploying a lightweight collector as a Heroku worker dyno, you get deep PostgreSQL monitoring that Heroku's built-in tools don't provide — without changing your database configuration.

What You Can Monitor on Heroku Postgres

• Query performance with pg_stat_statements: Heroku Standard and Premium plans have pg_stat_statements enabled. MonPG tracks every query — calls, execution time, rows processed, block I/O — with historical trends so you can see when performance started degrading.
• Connection tracking: Heroku plans have strict connection limits (20 for Hobby, 120 for Standard-0, 500 for Premium-0). MonPG tracks active, idle, and idle-in-transaction connections by application, helping you identify connection leaks before you hit the limit.
• Index recommendations: MonPG's index advisor analyzes your Heroku Postgres query patterns and recommends indexes with estimated impact. Stop guessing which indexes to create.
• Table bloat and vacuum: On Heroku's managed PostgreSQL, autovacuum runs automatically but may not be optimally tuned for your workload. MonPG tracks dead tuple accumulation and autovacuum effectiveness per table.
• Lock detection: Identify blocked queries and lock contention in real-time. See the lock tree to understand which query is blocking others.
• Heroku log collection: MonPG collects and analyzes Heroku Postgres logs via log drains, detecting slow queries, connection errors, and PostgreSQL-specific warnings that Heroku's logging doesn't surface.

Deploying on Heroku

The MonPG collector runs as a worker dyno in your Heroku app. It connects to your Heroku Postgres using the DATABASE_URL config var and pushes metrics to MonPG over HTTPS. Setup takes under 5 minutes:

1. Add the MonPG collector buildpack or use the Docker deployment
2. Set your MONPG_API_KEY config var
3. Scale the worker dyno: heroku ps:scale collector=1

The collector uses minimal resources — a single Hobby dyno is sufficient for monitoring. For production workloads, a Standard-1X dyno provides more headroom.

Heroku Plan Considerations

Not all Heroku Postgres plans support pg_stat_statements. Here's what you can monitor on each tier:

Outgrowing Heroku? Take MonPG With You

Many teams start on Heroku and eventually migrate to AWS RDS, Google Cloud SQL, or self-hosted PostgreSQL. With MonPG, your monitoring setup migrates with you. Same dashboards, same alerts, same query history — just point the collector at your new database.

Related Resources

• Connection Pooling with PgBouncer — Essential for Heroku's strict connection limits.
• PostgreSQL Index Optimization — Optimize queries before upgrading to a larger Heroku plan.
• How to Find and Fix Slow Queries — Diagnose performance issues on Heroku Postgres.
• PostgreSQL Slow Query Monitoring — Track slow query trends before upgrading to a larger Heroku plan.