UltraSpoof
Bridging System
Your backend player count goes up, but your proxy count does not. So the network total stays wrong, and it looks off. The bridge fixes that.
It is a separate plugin for the proxy. Velocity and its forks, BungeeCord and its forks.
How it works
Each backend publishes a full snapshot of its fake players every couple of seconds. The proxy reads those snapshots and injects matching connections, so the network list and counts include them.
Whole snapshots, not add and remove events. Whatever state the proxy is in, the next snapshot corrects it. Either side can restart, or miss a message, and the two still line up. That is the part that used to drift.
Setup
- Drop the bridge jar into your proxy's
plugins/folder. - Pick a transport and fill in the credentials in the bridge's
config.yml. - On each backend, set
bridge.enabled: trueinconfig.yml, with the same transport and the same credentials. - Set
bridge.server-nameto the exact name that server has in your proxy config.survival,hub1, whatever it is. That string is how the proxy knows where a bot belongs, so it has to match. - Restart everything.
MySQL or Redis
Both carry the same data.
MySQL is polled. One row per server in the ultraspoof_bridge table. The proxy reads it every poll-seconds, and after stale-after-seconds with no movement it drops that server's bots.
Redis is pushed. The snapshot goes to ultraspoof:bridge:snapshot:<server> with a TTL and gets announced on a channel. An expired key removes those bots by itself.
Use whichever you already run. Redis reacts a bit faster, MySQL is one less thing to install.
Proxy config
transport,mysql,redis. Match the backends.poll-seconds. How often MySQL is read, or how often Redis keys are re-scanned.stale-after-seconds. MySQL only. How long a silent backend keeps its bots.on-ping.real-count-for-ips. Addresses in this list see the real count in the server list, without the fake players. Put your own IP here.debug. Extra logging about what got read and injected.
Backend config
bridge:
enabled: false
server-name: "PUTSERVERNAMEHERE"
publish-interval-seconds: 2
transport: MYSQL
The mysql and redis blocks sit under it.
When a backend shuts down cleanly it publishes one last empty snapshot, so the proxy drops its bots straight away instead of showing ghosts until the row goes stale.
Lobby adoption (Pro)
Here is the thing. When your SMP restarts, its bots just vanish from the network. Real players do not vanish, they land in the lobby and come back after.
Adoption does the same for bots. The SMP goes down, its bots appear on the lobby. The SMP comes back, they leave the lobby and return. Same names, so it reads as the same players moving around.
Set it on the server doing the covering, usually the lobby:
bridge:
adopt:
enabled: false
servers: []
down-after-seconds: 20
max-bots: 40
serversis who to cover, by proxy name. A single*covers every other server that publishes.down-after-secondsis how long a server must be silent before it counts as down. A clean restart is noticed at once, because that server says goodbye on the way out. This setting is really for crashes.max-botscaps how many guests you hold, however many servers go down at the same time.
It needs the bridge enabled, since that shared database is how servers hear about each other.
Handover is per bot. As the SMP spawns each one back, the lobby lets that one go, so a bot is never in two places and never in neither.
Nothing changes on the proxy for any of this. The bridge just publishes whoever is on the lobby at the time.