pgwatch alongside postgres_exporter

[yoav-katz]

Hi,

I'm evaluating pgwatch and trying to understand how it typically fits into a modern Prometheus/Grafana monitoring stack.

Currently, I'm running:

• postgres_exporter and node_exporter

• Grafana Alloy (that remote_write to the Mimir)

• Grafana Mimir (centralized prometheus)

• Grafana dashboards based on Prometheus metrics

For those running pgwatch in production:

1. Do you still run postgres_exporter, or does pgwatch fully replace it?

2. If you run both, how do you divide responsibilities — e.g., postgres_exporter for alerting and operational metrics, pgwatch for deeper diagnostics and historical query analysis?

3. How do you handle metric overlap between the two (duplicate collection, storage cost, dashboard sprawl)?

I’d love to hear what architectures have worked well in practice — and just as much, what didn’t.

Thanks!