DBeaver with Arrow Flight SQL

Back to the client guide. DBeaver connects to this server through the Apache Arrow Flight SQL JDBC driver. Use Generic as the driver type and the tested 19.0.0 driver below.

The Voltron Data walkthrough has helpful screenshots of Driver Manager, library selection and connection creation. Use the settings on this page: that walkthrough shows JDBC 11.0.0, SQLite mode and TLS settings for a different endpoint. Do not copy those settings.

Start the test endpoint

Use a disposable Redis Enterprise cluster with Search, verified TLS, and username/password authentication. Build Flight through the qualified enterprise composition and export its authenticated rediss:// REDIS_URI from your secret manager. See integration setup. If the test endpoint is already running with this configuration, reuse it:

# Build using the qualified enterprise composition.
# Export REDIS_URI from your secret manager: authenticated rediss:// Enterprise endpoint.
./bin/redis-flight-server --listen localhost:18815

Keep the Flight process running while testing. These commands use the isolated Flight test port 18815; the project's normal Flight default is 8815. The DBeaver JDBC URL must use the Flight port, not the Redis port.

Download and configure the driver

Install DBeaver Community for your platform if it is not installed, then configure the driver below.

Download flight-sql-jdbc-driver-19.0.0.jar directly from Maven Central. The 19.0.0 directory also lists checksums and signatures. Choose the plain .jar, rather than a sources or tests JAR. The direct download was faster in the reported manual setup; this is an observation from that run, not a download speed guarantee.

  1. Open Database → Driver Manager → New.
  2. In Settings, enter the values below and select driver type Generic.
  3. In Libraries, choose Add File and select the downloaded JAR.
  4. Save the driver, then open Database → New Database Connection and select it.
Setting Value
Driver name Apache Arrow Flight SQL
Driver type Generic
Class name org.apache.arrow.driver.jdbc.ArrowFlightJdbcDriver
URL template (local test without TLS) jdbc:arrow-flight-sql://{host}:{port}/?useEncryption=false
Host localhost
Port 18815 for the endpoint above; 8815 for the normal default
Username/password Empty for this unauthenticated local test

The resulting URL for the test endpoint is:

jdbc:arrow-flight-sql://localhost:18815/?useEncryption=false

Use autocommit (Auto); this server does not support transactions. Click Test Connection, then Finish. For TLS and authenticated deployments, follow the JDBC connection options and build profile availability.

JVM options

The Arrow JDBC documentation requires --add-opens=java.base/java.nio=ALL-UNNAMED. For the tested JDBC 19.0.0 configuration, Java 23+ also needs --sun-misc-unsafe-memory-access=allow.

Check dbeaver.ini, after its existing -vmargs line. Both options were already present in the tested DBeaver Community 26.2.3 macOS installation, so no edit was needed. If missing, add the applicable options and restart DBeaver. The macOS file is /Applications/DBeaver.app/Contents/Eclipse/dbeaver.ini. See JDBC setup for the exact options.

Create a table and verify browsing

A fresh Redis test instance has no SQL tables. SQL tables appear in DBeaver; arbitrary Redis keys and named Flight datasets do not become SQL tables automatically. Open an SQL console on the connection and execute each statement separately. Use this fixture name only in the disposable test database; if it already exists, use a fresh name throughout.

CREATE TABLE dbeaver_sample (id BIGINT NOT NULL, name VARCHAR, qty INTEGER);
INSERT INTO dbeaver_sample (id, name, qty) VALUES (1, 'Apples', 10);
INSERT INTO dbeaver_sample (id, name, qty) VALUES (2, 'Oranges', 20);
INSERT INTO dbeaver_sample (id, name, qty) VALUES (3, 'Pears', 30);

Refresh the connection in the navigator, then expand redis → public → Tables → dbeaver_sample. Expand Columns and inspect id, name and qty. Run:

SELECT "id", "name", "qty"
FROM "redis"."public"."dbeaver_sample"
ORDER BY "id";

Expected result:

id name qty
1 Apples 10
2 Oranges 20
3 Pears 30

For additional manual coverage, open the table's Data tab, inspect the SQL DBeaver issues, and execute this filtered aggregate in the SQL console:

SELECT COUNT(*) AS n, SUM(qty) AS total
FROM dbeaver_sample
WHERE qty >= 20;

Expect n = 2 and total = 50. Record each check separately; the 2026-10-30 manual result confirms connection, navigator browsing and the three-row SELECT. It does not claim completion of every additional check above. Remove the fixture when done:

DROP TABLE dbeaver_sample;

Stop the foreground Flight process with Ctrl+C. To stop the Compose test Redis, run docker compose stop redis. Keep both services running until manual testing is complete.

Automated checks and known limits

With a full JDK installed, run this from another terminal against the endpoint:

make jdbc-smoke JDBC_URL='jdbc:arrow-flight-sql://localhost:18815/?useEncryption=false'

This downloads and verifies JDBC 19.0.0, selects the Java memory options, creates unique fixtures and cleans them up. It checks metadata, parameters, writes, sessions and a 100,000-row concurrent read. DBeaver's bundled runtime may lack the compiler needed to run the Java source file; use a full JDK via JAVA=/path/to/jdk/bin/java if necessary. A JDBC smoke pass does not establish that the DBeaver UI passed.

Earlier accessibility automation attempts crashed DBeaver on macOS; the later manual connection and query succeeded. Track the automation crash separately in DBeaver #42640. DBeaver PR #42645 proposes a built-in Flight SQL driver entry; the validated application setup used a custom driver. See the client matrix for remaining JDBC and application coverage limits.