pano-node & Pano Agent

pano-node is the daemon that runs managed servers on a machine. It always runs as its own process, so restarting or updating Pano never takes the game servers down. On Pano's own machine Pano runs it for you (the Local node); this page is for running it yourself.

Installing

Download it from your own Pano, so the version always matches — it is also attached to every Pano release. It needs Java 11 or newer.

curl -fsSL https://panel.example.com/api/node/pano-node.jar -o pano-node.jar

# First start: pair with the code from Nodes → Add node → Manual
java -jar pano-node.jar --pano https://panel.example.com --code 123456 --data ./node-data

# Every start after that
java -jar pano-node.jar --data ./node-data

Flag	Variable
--pano <url>	PANO_URL	Your Pano's address
--code <code>	PANO_PAIR_CODE	The pairing code
--data <dir>	PANO_NODE_DATA	The data folder (default ./node-data)
--name <name>	PANO_NODE_NAME	The node's name in the panel
--runtime DOCKER	PANO_NODE_RUNTIME	Run each server in its own Docker container
--port-range <a-b>	PANO_NODE_PORT_RANGE	Ports for its servers (default 25565-25600)
--service install		Write a systemd / launchd / Windows service file and exit

In a container: the image is ghcr.io/panomc/pano-node. Mount a volume at /data (or everything is lost with the container), publish the same ports as PANO_NODE_PORT_RANGE, and use a restart policy such as --restart unless-stopped.

Data and configuration

Everything lives in the data folder: config.conf, servers/, backups/, java/ and a few caches. Only one daemon can use a folder at a time.

platform {
  url = "https://panel.example.com"
  token = "<secret>"
  encryption-key = "<secret>"
}
node {
  stop-servers-on-exit = false
  java-auto-download = true
  tool-auto-download = true
}

• The token and keys are secrets. Never copy or share the file. If it leaks, delete the node in the panel and pair again.
• stop-servers-on-exit = false keeps servers running while the daemon restarts.
• Missing Java versions are downloaded into java/, and a small git for compiling Spigot into tools/. Set the two auto-download keys to false to prevent that.
• Edit the file only while the daemon is stopped.

Updating and removing

• Updating: the Nodes page shows Update available. The daemon downloads the new jar, checks it and restarts in a few seconds; its servers keep running. On Windows, copy the new jar over by hand.
• Removing: Nodes → Delete deletes every server, backup and downloaded Java on the node, after asking for its name and your password. An offline node can only be forgotten; its files stay.

Exit code	Meaning
75	Update staged — start it again
76	Another daemon already uses this folder
78	Removed from Pano — do not restart

Pano Agent

The Pano Agent hands a server you already run to Pano where it is: put pano-agent.jar in the server's folder and start it instead of the server jar. It starts the server, keeps its console in your terminal, and lets Pano manage the rest. Nothing is moved.

Add server → Link with the Pano Agent shows the download and a ready-made command:

cd /home/mc/survival
curl -fLo pano-agent.jar 'https://panel.example.com/api/node/pano-agent.jar'

# Stop the server, then run the dialog's command once (the code works once, for 1 minute)
java -jar pano-agent.jar --pano 'https://panel.example.com' --code k7m2x9qa4tj3n8wp

# Every start after that
java -jar pano-agent.jar

• The first run asks for the server jar, memory and Java arguments; Enter keeps the defaults it read from your start script. On a hosting panel such as Pterodactyl, set the server jar to pano-agent.jar and answer in the web console.
• Typing stop stops only the server; the agent keeps running so Pano can start it again. Ctrl+C stops both.
• The agent keeps its data in .pano-agent/ — leave it out when copying the folder for a new server.
• It updates itself without stopping the server. To remove it, remove the server in Pano, then start the server jar directly again.

Security

• The node only connects out to Pano; nothing on it has to be opened.
• Every message is AES-256-GCM encrypted — still use HTTPS for a node on another machine.
• A node paired with a code must be approved in the panel.
• File access stays inside each server's folder, and servers are started without a shell.