Skip to content

Databases

Plugins like LuckPerms and CoreProtect, and most bots and web apps, want a database. A server’s Databases tab gives it one on a database instance running on the same node, included in the node’s price. Each database has its own user, which reaches that database and nothing else.

  1. An admin makes a database instance on the node first, on the Databases page (see Database instances). One instance serves every server on its network.
  2. On the server’s Databases tab, click New database and give it a name, like luckperms. It’s made on the instance and linked to this server.
  3. Copy the password from the dialog that opens. You can show it again later with Password, which asks for your passkey.

The database belongs to its instance, not the server: deleting the server keeps it, on the Databases page.

Admins can give servers one database between them, like a network’s LuckPerms:

  • Link existing on a server’s Databases tab links it to a database already on an instance it can use. It gets the details at its next start.
  • Unlink takes it off the server; the database stays.
  • Delete on a server’s tab works only while no other server is linked to the database. Otherwise, unlink it, or delete it from the Databases page.

Each database shows its Host, Database, and User, with the password behind Password. Put those in the plugin’s config. For LuckPerms, in plugins/LuckPerms/config.yml:

storage-method: MariaDB
data:
address: community-a-mariadb.db.raptor:3306
database: s3e4f5a6b7c8d_luckperms
username: s3e4f5a6b7c8d_luckperms
password: the password

The host is a name, like community-a-mariadb.db.raptor, that works from the node’s servers. It stays the same even if the instance moves to another address.

Plugins on the node connect without TLS, and MySQL 8 then makes the client fetch the server’s key before it can sign in. Plugins that use MySQL’s Java driver fail with “Public Key Retrieval is not allowed” unless you allow it. In LuckPerms, add it under pool-settings → properties:

properties:
useUnicode: true
characterEncoding: utf8
allowPublicKeyRetrieval: true

Other plugins take the same setting as allowPublicKeyRetrieval=true at the end of their JDBC URL. For the mysql client, add --get-server-public-key. MariaDB and PostgreSQL don’t need this.

If an admin gave the database’s instance external access, its row also shows its address from outside, and a link to its certificate. Only the addresses the admin listed can connect there, with TLS.

A database on a Redis or Valkey instance has a user and a password, but no name: Redis’ databases are numbered, and yours is 0. Each Redis or Valkey instance is for one server, so pick one nothing else uses. Valkey takes everything Redis does, with the same REDIS_* variables. A MongoDB database’s user is stored in the database itself, so connect with authSource set to its name; the URL Raptor gives (MONGODB_URI) has it.

Open on a Redis or Valkey database shows its keys: search them with a pattern like player:*, see each one’s value and when it expires, and with Change databases edit them, add new ones, or run Redis commands from the Command tab. On a MongoDB database it shows its collections and their documents: find documents with a filter like {"name": "alex"}, sort them, and edit, insert, or delete them as JSON. MongoDB databases can be exported and imported as mongodump archives; a Redis database’s data is in the server’s backups, and Restore from a backup puts it back.

The server also gets its database’s details as environment variables at its next start: DB_HOST, DB_PORT, DB_NAME, DB_USER, DB_PASSWORD, and DATABASE_URL (like mysql://user:password@host:3306/name). A Redis database gives REDIS_HOST, REDIS_PORT, REDIS_USER, REDIS_PASSWORD, and REDIS_URL; a MongoDB one MONGODB_HOST, MONGODB_PORT, MONGODB_DATABASE, MONGODB_USER, MONGODB_PASSWORD, and MONGODB_URI. With more than one database, each also gets them prefixed with its name (LUCKPERMS_DB_HOST), and the first one’s are also the plain DB_HOST and the rest. Apps and bots that read their settings from the environment pick them up with nothing to edit.

Password shows a database’s password; it asks for your passkey first. The key button gives the database a new password: the old one stops working at once, and the server gets the new one at its next start. Change it in any config files too.

Open on a database shows its tables on the left, and:

  • Rows: a table’s rows, a page at a time. Click a column’s name to sort by it, and filter on any column (“contains”, “is”, “is NULL”, and the rest). Click a row to see all of it; with Change databases you can edit it there, Insert row, or tick rows and delete them. A table without a primary key can only be changed with SQL, since there’d be no safe way to say which row.
  • Structure: its columns and indexes. Add or drop columns and indexes, rename or drop the table, or make a New table; each shows the SQL it runs first.
  • SQL: write any SQL and run it with Run or Ctrl/⌘ + Enter. Table and column names are suggested as you type. With several statements, the last one’s result shows. Rows show as they arrive, 200 to a page, and Cancel stops the query on the database itself. A query stops after 30 seconds unless you set another limit (up to 600, ten minutes). A result shows its first 10,000 rows and says how many it had; Export the whole result writes every row to a file (below).
  • Processes: what’s connected to the database right now, like the server’s own plugins, and stopping a slow query.

Someone with only See databases can look at everything and run queries that don’t change anything: one statement at a time, as a user that can only read.

Changes are in the server’s activity, with how many rows they changed, never the SQL itself.

From an opened database (needs Change databases):

  • Export writes the whole database, or the table you’re looking at, to the server’s database-exports folder: as .sql.gz, or a table as .csv. Open the folder takes you there in the file manager to download it.
  • Import runs a .sql or .sql.gz file from the server’s files against the database. Upload it in the file manager first, then give its path. It stops at the first error and says what it was; what ran before stays.
  • On the SQL tab, CSV saves the rows on screen. When a result has more than fit on screen, Export the whole result runs the query again, read-only, and writes every row to database-exports as .csv.

A dump from another host may make or switch databases, set owners, or name a DEFINER; the database’s user can’t do those, so the import stops there. Remove those lines, or export without them (--no-owner for PostgreSQL).

For a tool that should only look (a stats page, a web map, a backup script), give the database a read-only user: Make one under the database. It reads every table, including ones made later, and changes nothing. Its Password asks for your passkey; the key button gives it a new one; the bin removes it. It reaches the database from the node’s servers on the same network, like the database’s own user.

The server’s backups hold its databases. To put one database back as it was, without touching the server’s files, click the clock button next to it (Restore from a backup, needs Change databases) and pick a backup. It asks for your passkey.

What the database has now is exported to the server’s database-exports folder first, so you can import it again if you picked the wrong one. The database is then emptied and filled from the backup; its user, password, and read-only user stay the same.

Two permissions on a server’s access cover the databases linked to it: See databases shows the tab and the connection details, and Change databases lets someone make and delete databases and see and change their passwords. Deleting a database deletes everything in it, and restoring one replaces everything in it, so both ask for a passkey. Linking and unlinking, and the Databases page, are for the org’s admins and owners.

© 2026 Xena Studios. Pterodactyl is a trademark of its owners; Raptor isn't affiliated with it.