Storage and Databases

Storage and Databases

VelocityNavigator does not require MySQL, MariaDB, PostgreSQL, or MongoDB. A normal single-proxy setup keeps its configuration and small pieces of persistent state in the plugin folder.

Files on the Velocity proxy

File	Purpose
navigator.toml	Routing, commands, health, metrics, and optional systems
messages.toml	Language and player-facing text
gui.toml	Java inventory and Bedrock form layout
servers.toml	Lobby information added through server-management commands
affinity-store.json	Unexpired sticky-lobby records when affinity is enabled
drained-servers.txt	Servers kept out of new routing during maintenance

Back up the whole plugins/velocitynavigator/ folder before a major update or a large configuration change.

What stays in memory

Live health results, current player counts, circuit-breaker state, parties, and queue positions are runtime information. Most of it is rebuilt naturally while the proxy runs.

Party membership and queue positions do not survive a proxy restart and are not stored in a SQL database.

When Redis helps

Redis is useful when:

• you run more than one Velocity proxy;
• proxies should share health, circuit, backend-state, or affinity information;
• autoscaled backends need to register and unregister themselves.

Redis is not a replacement for the config files, and enabling it does not make parties or queues global. See Redis and Multi-Proxy Networks for the full setup.

Do I need a database plugin?

No. Do not install a JDBC driver or create SQL tables for VelocityNavigator. If another plugin handles global parties, punishments, permissions, or player profiles, it can continue using its own database independently.

Moving to another proxy

1. Stop the old proxy cleanly.
2. Copy the complete VelocityNavigator plugin folder.
3. Copy or update the related server entries in velocity.toml.
4. Start the new proxy and run /vn config validate.
5. Check /vn servers and try /lobby before opening it to everyone.

If Redis is enabled, give the new proxy its own node_id rather than copying the old node identity unchanged.