Prometheus openmetrics for Nostream

A devops component for directly monitoring your nostream relay, which can be integrated with your self-hosted SRE stack. If you don't have a monitoring system set up already, this won't be useful to you. It's basically a python script that runs Postregsql queries to compile basic metrics.

Installation and Usage

pip install nostream-exporter

Define the connection parameters to your Nostream via environment variables. There are multiple ways you can set it up. One example is to create an .env file with the following contents:

export DB_HOST=/var/run/postgresql
export DB_PORT=5432
export DB_NAME=nostream
export DB_USER=nostream
export DB_PASSWORD='hunter2'
export METRICS_PORT=9101
export METRICS_BIND=127.0.0.1

Both the METRICS_PORT and METRICS_BIND options are optional, the defaults are shown.

I would recommend running this as an unprivileged user via systemd (or anything comparable):

useradd -mb /opt -k /dev/null -s $(which nologin) nostream_exporter

Create /etc/systemd/system/nostream_exporter.service:

[Unit]
Description=Basic openmetrics for Nostream nostr relay

[Service]
User=nostream_exporter
Group=nostream_exporter
EnvironmentFile=/opt/nostream_exporter/.env
ExecStart=nostream_exporter
Restart=on-failure
RestartSec=2m

[Install]
WantedBy=multi-user.target

Confirm that it is serving metrics:

curl http://127.0.0.1:9101/metrics
...
# HELP events Count of events by kind
# TYPE events gauge
events{kind="7"} 48184.0
events{kind="1"} 24659.0
events{kind="6"} 3737.0
events{kind="1984"} 3236.0
events{kind="4"} 2265.0
events{kind="other"} 762.0

Development

Clone the repo and install dependencies into a local virtual environment:

git clone git@github.com/bleetube/nostream_exporter
cd nostream_exporter
python -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install --editable .
source .env
nostream_exporter

Note that there is a difference between source .env and systemd EnvironmentFile which is that you'll need to have the export keyword. i.e. export DB_NAME=nostream

Roadmap

• Include tests
• Change print() messages into sensible logging with configuration debug levels.
• Review events table to possibly discover more metrics.
• Dockerfile