hotline-docker

General-purpose Hotline server / tracker container images. They're usable for real deployments, and are also what GtkHx's CI pulls instead of rebuilding on every run.

Images

Image (images/<dir>)	Published tag	Default ports	Role
mhxd	mhxd	5500 / 5501	Hotline server (mhxd)
janus	janus	5500 / 5501 (+TLS)	Hotline server (VesperNet Janus)
argus	argus	5498 tcp / 5499 udp	Tracker (v1/v2/v3, VesperNet Argus)
hxtrackd	hxtrackd	5498 tcp / 5499 udp	Tracker (pre-spec v1, mhxd's hxtrackd)

Each image runs a general default config (features off, conventional ports, no seeded content). Customise at runtime with env vars or by mounting your own config — see below and each image's README.md.

Configuration

All servers accept a mounted config file (over the image's default) and a few env knobs applied by the entrypoint:

• Janus — JANUS_PORT, JANUS_TLS_PORT, JANUS_ENABLE_HOPE, JANUS_ENABLE_CHAT_HISTORY, JANUS_ENABLE_VOICE, JANUS_TLS_CERT / JANUS_TLS_KEY, TRACKERS.
• mhxd — TRACKERS, BANNER_MODE / BANNER_URL / BANNER_FILE.
• Argus — ARGUS_TCP_PORT, ARGUS_UDP_PORT.
• hxtrackd — none (its ports are hardcoded); mount hxtrackd.conf to tune the drop interval etc.

TLS (e.g. Let's Encrypt / certbot)

Janus serves TLS when given a cert + key. Mount them and point Janus at them — no image rebuild, and certbot renewals are picked up on restart:

docker run -d --network host \
  -v /etc/letsencrypt/live/example.com:/certs:ro \
  -e JANUS_TLS_CERT=/certs/fullchain.pem \
  -e JANUS_TLS_KEY=/certs/privkey.pem \
  ghcr.io/mishan/janus

Where images are published

GitHub Container Registry (GHCR), under the repo owner's namespace:

ghcr.io/mishan/mhxd:latest
ghcr.io/mishan/janus:latest
ghcr.io/mishan/argus:latest
ghcr.io/mishan/hxtrackd:latest

Each is also tagged with the commit SHA; server images get the git tag name on v* releases. Two workflows publish (both push with the built-in GITHUB_TOKEN — enable packages: write, already declared):

• .github/workflows/publish-images.yml — the five server/proxy images, matrix-built with per-image GHA layer caching. Triggers on images/** changes, weekly, and manually.

After the first run, make the packages public (or grant readers) on GHCR so they can be pulled without auth — new packages are private by default.

Running the images

Each image is an independent server — run whichever one you need on its own. A typical deployment is a single server on its conventional port:

# mhxd Hotline server
docker run -d -p 5500:5500 -p 5501:5501 ghcr.io/mishan/mhxd

# Janus Hotline server (with HOPE + a Let's Encrypt cert)
docker run -d --network host \
  -v /etc/letsencrypt/live/example.com:/certs:ro \
  -e JANUS_ENABLE_HOPE=true \
  -e JANUS_TLS_CERT=/certs/fullchain.pem \
  -e JANUS_TLS_KEY=/certs/privkey.pem \
  ghcr.io/mishan/janus

# Argus tracker
docker run -d -p 5498:5498 -p 5499:5499/udp ghcr.io/mishan/argus

# hxtrackd tracker
docker run -d -p 5498:5498 -p 5499:5499/udp ghcr.io/mishan/hxtrackd

To have a server register with a tracker, pass TRACKERS (mhxd/Janus) — see each image's README.md. Use --network host for Janus if you're running its WebRTC voice path.