Running on docker

Running on Docker.

General recommendations:

• use environment variables instead of YAML configuration.
• disable systems collectors - it is unnecessary to collect system metrics being inside the container.
• disable collectors which require access to PostgreSQL data directories - postgres/settings, postgres/storage, postgres/logs.
• setting up listening to allow network connection from other containers or hosts.

An example below demonstrates a typical use case when a database and pgSCV are running in separate containers.

1. Create example network.

docker network create example

2. Run postgres container.
   • Enable pg_stat_statements, it is preferred (but not strict).
   • Use POSTGRES_PASSWORD=secret to define password.

docker run -d --rm --name db -ti \
-e POSTGRES_PASSWORD=secret \
--network example \
postgres:14 -c 'shared_preload_libraries=pg_stat_statements'
docker exec -ti db psql -U postgres -c 'CREATE EXTENSION pg_stat_statements'

3. Run pgscv container.
   • Specify Postgres connection settings using DATABASE_DSN.
   • Disable some collectors with PGSCV_DISABLE_COLLECTORS.
   • Setup listen address with PGSCV_LISTEN_ADDRESS.
   • Expose port 9890.
   • Don't detach, it's better to see logs if something goes wrong.

docker run -d --rm --name pgscv -ti \
-e PGSCV_LISTEN_ADDRESS=0.0.0.0:9890 \
-e PGSCV_DISABLE_COLLECTORS=system \
-e DATABASE_DSN="postgresql://postgres:secret@db/postgres" \
-p 9890:9890 --network example \
cherts/pgscv:latest

4. Open the second terminal and fetch metrics using curl.

curl -s http://127.0.0.1:9890/metrics

5. In the end, stop containers and remove example network.

docker stop db pgscv
docker rm db pgscv
docker network rm example