Quip Node Manager

Desktop application for running and monitoring Quip network nodes. Supports Docker and native execution modes on macOS, Linux, and Windows.

Quick Install

macOS / Linux:

curl -fsSL https://gitlab.com/quip.network/quip-node-manager/-/raw/v0.2.0/scripts/install.sh | sh

Windows (PowerShell):

irm https://gitlab.com/quip.network/quip-node-manager/-/raw/v0.2.0/scripts/install.ps1 | iex

Manual Download

Download the latest release from the Releases page.

macOS

Download the .dmg, open it, and drag the app to /Applications.

Because the app is not yet notarized, macOS will quarantine it. Open Terminal (Applications > Utilities > Terminal) and paste:

xattr -dr com.apple.quarantine /Applications/Quip\ Node\ Manager.app

Then launch the app from /Applications, not from the .dmg or Downloads folder.

Linux

The recommended format is AppImage (works on any distro):

chmod +x quip-node-manager-linux-x86_64.AppImage
./quip-node-manager-linux-x86_64.AppImage

A .deb package is also available for Debian/Ubuntu:

sudo dpkg -i quip-node-manager-linux-x86_64.deb

Windows

Download the .exe and run it. Windows SmartScreen may show a warning because the binary is not yet code-signed.

Click More info, then Run anyway.

Features

• Full compose stack -- runs miner + validator + dashboard + postgres behind a Caddy front door via Docker Compose; dashboard UI is embedded in the app's Dashboard tab
• Two run modes -- Docker (default on Windows/Linux) drives the full container stack; Native (macOS) runs a standalone binary on the host and still runs the validator/dashboard/postgres/caddy containers, wired to the host via host.docker.internal
• CPU or CUDA image -- chosen automatically from the GPU device toggles (CUDA whenever an NVIDIA device is enabled); D-Wave mining runs on the CPU image via the [dwave] config section
• TLS toggle -- the dashboard (Postgres-backed telemetry UI) and Caddy front door are always part of the stack; TLS is optional, with automatic Let's Encrypt or ZeroSSL certificates
• Pre-flight checklist -- verifies Docker + Compose v2 availability (plus WSL on Windows, the miner binary in Native mode, and the D-Wave token when QPU mining is configured), node secret, public IP, hostname, and external port reachability before starting (images aren't pre-checked -- Start always pulls them)
• Live log streaming -- tails docker compose logs -f <node> in a collapsible drawer; switches to data/node.log once the node writes to it
• GPU configuration -- detects CUDA and Metal devices, per-device enable/disable, utilization slider, yielding mode
• D-Wave QPU support -- optional quantum processing unit configuration with daily budget controls
• Background update monitor -- checks for new node + dashboard image digests and manager app releases every 30 minutes; optional auto-restart on digest change
• TLS certificate guidance -- Caddy's ACME (Let's Encrypt or ZeroSSL) is wired up out of the box; set a DNS name + email and TLS "just works"

Development

Prerequisites

• Rust (stable)
• Bun (or Node.js)
• Docker Compose v2 (bundled with Docker Desktop; standalone docker-compose v1 is not supported)
• Platform-specific Tauri v2 dependencies (see Tauri docs)

Setup

This repo vendors the compose stack via a git submodule. After cloning, fetch the submodule at the commit locked by this repo:

make fetch-submodules

(Equivalent to git submodule update --init --recursive, or clone with git clone --recurse-submodules to do this in one step.)

Commands

bun install          # Install JS dependencies
bun run dev          # Launch development build
bun run build        # Production build for current platform

cd src-tauri
cargo check          # Type-check Rust code
cargo clippy         # Lint

CLI Mode

The app also supports a terminal UI mode:

quip-node-manager --cli

Architecture

• Frontend: src/ -- vanilla HTML/CSS/JS with Tauri IPC (withGlobalTauri: true). Dashboard tab embeds the running dashboard container in an iframe.
• Backend: src-tauri/src/ -- Rust + Tauri v2 commands. compose.rs drives the stack via docker compose; stack_assets.rs stages the bundled compose files into ~/quip-data/.
• Stack definition: vendor/nodes.quip.network/ -- git submodule tracking the upstream Docker Compose setup (miner + validator + dashboard + postgres + caddy).
• Config: TOML generation matching quip-protocol format; .env generated from settings on every Start.
• Data: ~/quip-data/ (configurable at first boot) holds app settings, runtime config, secrets, native binaries, and the staged compose files.

Native mode note (macOS): the node's REST API binds to 127.0.0.1:20100. Docker Desktop's vpnkit forwards container traffic from host.docker.internal through to the host's loopback, so the REST port is not exposed to the LAN.

Overriding the bundled stack

Node Manager rewrites ~/quip-data/docker-compose.yml from its embedded copy on every Start and Apply, discarding whatever you edited there. Put changes in ~/quip-data/docker-compose.override.yml instead. Staging never touches that file, and compose merges it over the base.

services:
  quip-validator:
    environment:
      RUST_LOG: debug

Compose loads an override file automatically only when it discovers the compose file itself. Node Manager always passes -f, so it adds the override explicitly when the file exists, after the base file so the override wins.

Two things to watch:

• command: replaces the whole list rather than merging it. To change one validator flag, copy the entire command: block and edit the line you want.
• Copy that block from the staged ~/quip-data/docker-compose.yml, not from vendor/nodes.quip.network/. Staging adds flags the vendored file doesn't contain, such as --public-addr.

The compose file caps the miner container at 16 GiB, which stops a runaway round from triggering a host-wide out-of-memory kill. On a dedicated miner with more RAM, raise it by setting miner_mem_limit_gb in ~/quip-data/app-settings.json:

{ "miner_mem_limit_gb": 48 }

Set it there rather than in .env, which Node Manager regenerates on every Start. There's no UI control for this yet.

Validator pruning isn't recommended. The validator runs --state-pruning=archive --blocks-pruning=archive, keeping every state trie and block body from genesis. Size the disk for that. Pruning reclaims disk, but it breaks the dashboard: the descriptor worker scans from genesis and fails with State already discarded once it reads past the pruning window. Pruning also likely reduces your point awards on the SNAG platform. Only prune if you accept losing the dashboard.

See AGENTS.md for detailed architecture documentation.

License

AGPL-3.0-or-later

Copyright (c) Postquant Labs