Connections are leaking

[MTRNord]

What did you do?

I am running the prometheus-community helmchart. No autodiscovery, no config, one DSN setup. Nothing unusual afaik.

What did you expect to see?

I expected it to go up in connections and then hit a ceiling.

What did you see instead? Under which circumstances?

The connections went up to 50 until pg ran out of connections (total at pg is 240 connections + 3 superuser connections):

ts=2023-08-31T18:39:22.792Z caller=collector.go:173 level=error msg="Error opening connection to database" err="error querying postgresql version: pq: sorry, too many clients already"

Environment

• System information:

  Linux 6.1.31-gentoo-x86_64 x86_64 (inside kubernetes pod using the docker image provided in this repo)

• postgres_exporter version:

  postgres_exporter, version 0.13.1 (branch: master, revision: 5890879)
  build user: root@43776d249d3b
  build date: 20230825-09:22:12
  go version: go1.20.7
  platform: linux/amd64
  tags: netgo static_build

• postgres_exporter flags:

--config.file=/etc/postgres_exporter.yml --web.listen-address=:9187

• PostgreSQL version:

  matrix-postgres-cluster-0:/home/postgres# postgres --version
  postgres (PostgreSQL) 15.2 (Ubuntu 15.2-1.pgdg22.04+1)

• Logs:

ts=2023-08-31T18:39:22.792Z caller=collector.go:173 level=error msg="Error opening connection to database" err="error querying postgresql version: pq: sorry, too many clients already

(this is repeated a lot. But otherwise no logs after startup)

[sysadmind]

This was fixed in #902