logos-delivery-module/docs/run-node.md

Run a delivery node

Runs a delivery node (logoscore daemon + delivery_module). There is no GUI or HTTP API — interaction is via the logoscore CLI. You can run it three ways:

• With Docker — quickest; everything runs in a container.
• Prebuilt binaries — download release binaries, nothing to build (Linux and macOS).
• Build with Nix — build from source on any platform.

All three connect the node to the logos.test fleet by default.

With Docker

Prerequisites

• Docker with Compose

Start

git clone https://github.com/logos-co/logos-delivery-module.git
cd logos-delivery-module
docker compose up -d --build

First build runs Nix and downloads release packages — allow ~30–45 min. Later starts are fast.

Boot the node

The daemon is running; load the module and start the node:

docker exec logos-node logoscore load-module delivery_module --json
docker exec logos-node logoscore call delivery_module createNode @/conf/logos-test.json --json
docker exec logos-node logoscore call delivery_module start --json

Verify:

docker exec logos-node logoscore status --json

Stop

docker compose down

Without Docker: prebuilt binaries

Run a node from released binaries — nothing to build, no repository clone. You need three CLIs from the Logos releases:

• logoscore — the node daemon (logos-logoscore-cli)
• lgpd — package downloader, fetches modules from the Logos catalog (logos-package-downloader)
• lgpm — package manager, installs them locally (logos-package-manager)

All three are published for Linux (x86_64 / aarch64) and macOS (Apple Silicon / aarch64).

Install the tools

This downloads logoscore, lgpd, and lgpm for your OS/arch into ./bin (the script pins a known-good release of each — bump the *_TAG values in it to move to newer builds):

curl -fsSL https://raw.githubusercontent.com/logos-co/logos-delivery-module/master/scripts/install-node-tools.sh | sh
export PATH="$PWD/bin:$PATH"

Download the module and boot the node

# Fetch delivery_module from the Logos catalog, then install it into ./modules
mkdir -p packages modules
lgpd download delivery_module --output ./packages
lgpm install --dir ./packages --modules-dir ./modules

# logos.test node config
cat > logos-test.json <<'JSON'
{ "preset": "logos.test", "logLevel": "DEBUG" }
JSON

# Run the daemon (it binds capability_module automatically, so ./modules only
# needs delivery_module), then boot the node
logoscore -D -m ./modules > logs.txt &
logoscore load-module delivery_module
logoscore call delivery_module createNode @logos-test.json
logoscore call delivery_module start

Verify with logoscore status; stop with logoscore stop.

Without Docker: build with Nix

Build the runtime and this module with Nix, then run the logoscore daemon directly on the host.

Prerequisites

• Nix with flakes enabled
• Linux or macOS

Build the runtime and module

Build the logoscore CLI (the headless runtime) and the lgpm package manager from their flakes, then build and install this module's .lgx:

git clone https://github.com/logos-co/logos-delivery-module.git
cd logos-delivery-module

# Runtime + package manager
nix build 'github:logos-co/logos-logoscore-cli' --out-link ./logos
nix build 'github:logos-co/logos-package-manager#cli' -o lgpm

# This module, built from the current checkout
nix build '.#lgx' -o delivery-lgx

# Seed the modules dir with the bundled capability module, then install
mkdir -p modules
cp -RL ./logos/modules/. ./modules/
./lgpm/bin/lgpm --modules-dir ./modules --allow-unsigned install --file delivery-lgx/*.lgx

The first build compiles the whole runtime stack through Nix — allow ~30–45 min. Later builds are fast.

Start the daemon

Put logoscore on your PATH and start it in daemon mode pointed at ./modules:

export PATH="$PWD/logos/bin:$PATH"
logoscore -D -m ./modules > logs.txt &

Boot the node

logoscore load-module delivery_module
logoscore call delivery_module createNode @conf/logos-test.json
logoscore call delivery_module start

Verify:

logoscore status

Stop

logoscore stop

For a fully pinned, build-from-this-commit walkthrough — plus notes on the blocking Kademlia bootstrap in headless runs — see the runtime doc-test.

Configuration

The node config is just the logos.test network preset. The repo ships it as conf/logos-test.json: with Docker it is mounted into the container at /conf (@/conf/logos-test.json); with the Nix build, pass the path directly (@conf/logos-test.json). The prebuilt-binaries path above writes the same config inline as logos-test.json so no clone is needed. Edit it and re-run the boot steps to change settings.

To target the dev network instead, use conf/logos-dev.json (preset logos.dev). Available keys are documented in the README.

The node is now connected to the logos.test network. See query-node.md to read its peer ID, ENR, and metrics.

Metrics

The node already aggregates Prometheus metrics internally (the same set exposed on metricsServerPort, rendered behind the "Metrics" node-info attribute). collectOpenMetricsText() hands that exposition text back verbatim so the openmetrics module can scrape this module without standing up a separate HTTP server — no in-module parsing or reshaping. The openmetrics scraper parses the text, injects a module="delivery_module" label on every series, and merges it with other modules.

Point the openmetrics module at this one by name, selecting the text-source convention with "format": "text":

logoscore --config-dir /tmp/om call openmetrics start \
  '{"port":9090,"modules":[{"name":"delivery_module","format":"text"}]}'
curl http://localhost:9090/metrics   # every series carries module="delivery_module"

Before a node is created (or if the read fails) collectOpenMetricsText() returns an empty document so a scrape never errors out on this module.

For a one-off read without the openmetrics module, the raw exposition text is also available via getNodeInfo Metrics — see query-node.md.