Storia Worker

Set up Storia Worker on a helper machine to generate terrain for your main server.

Storia Worker is the Storia server started in a special mode (-Dstoria.worker=true) that only computes terrain for other Storia servers. It opens no player port, no query and no RCON, and it never changes your world: it only reads the world settings needed to produce the same terrain.

Requirements#

• Java 25.
• CPU cores: the more, the better. By default the worker uses all of them (offload.threads: -1).
• Memory: 2–4 GB of heap is plenty for most worlds. Set it with WORKER_MEMORY.
• A TCP connection between the worker and the main server (or the relay). The port is 25590 by default and can be any port you like (see Offloading terrain).

Install#

1. Download storia-worker-26.2.zip from the downloads page and unzip it:

   text storia-worker-26.2/ storia.jar storia.yml start-worker.sh start-worker.bat README.md

2. Copy the world settings. Copy the main server's world folder into a world/ folder next to storia.jar, without the chunk folders (region, entities, poi). What is left is small: level.dat, datapacks/ and the data/ folders, which in Minecraft 26.2 hold the seed and world generation settings.

   bash rsync -a --exclude region --exclude entities --exclude poi \ main-server:/srv/storia/world/ world/

   On Windows, copy the whole world folder and delete the region, entities and poi folders inside world/dimensions/*/*/.

   Copy the data folders too

   Copying only level.dat is not enough in Minecraft 26.2: the seed lives in dimensions/minecraft/overworld/data/minecraft/world_gen_settings.dat. Without it the worker cannot start.

3. Accept the EULA. Read the Minecraft EULA and, if you agree:

   bash echo "eula=true" > eula.txt

4. Set the secret in storia.yml, identical to the main server's:

   yaml offload: mode: worker secret: "a long random secret" bind: 0.0.0.0 port: 25590 relay: "" threads: -1 compress: true

5. Start it:

   bash ./start-worker.sh # Linux / macOS WORKER_MEMORY=6G ./start-worker.sh # with a 6 GB heap

   On Windows, run start-worker.bat.

6. On the main server, add the worker to offload.workers (see Offloading terrain) and restart it.

When the main server connects, the worker log shows which dimensions were accepted. Run /storia offload in the worker console to see its state.

Listening or connecting out#

A worker works in one of two ways:

Use a relay when workers are behind NAT, change often, or when several servers share workers. See Storia Relay.

Keeping the worker in sync#

The worker must produce exactly the same terrain as the main server. After you:

• change the seed or world generation settings,
• add, remove or update a datapack that affects world generation,
• update Storia on the main server,

copy the world settings again (step 2) and update the worker's storia.jar to the same version. If they differ, the worker refuses the affected dimensions (the log says terrain differs: check seed, datapacks and Storia build), and the main server generates them locally. Nothing breaks, it is just not offloaded.

Running as a service (Linux)#

# /etc/systemd/system/storia-worker.service
[Unit]
Description=Storia Worker
After=network-online.target

[Service]
User=minecraft
WorkingDirectory=/srv/storia-worker
Environment=WORKER_MEMORY=4G
ExecStart=/srv/storia-worker/start-worker.sh
Restart=on-failure

[Install]
WantedBy=multi-user.target

sudo systemctl enable --now storia-worker
journalctl -u storia-worker -f

Stopping#

Type stop in the console or stop the service. Requests in progress are handed back and the main server generates those chunks itself.