apache/seatunnel · warning

Failed to close Fluss connection for {}

Error message

Failed to close Fluss connection for {}

What it means

FlussAdminClient.close() closes the underlying Fluss Connection after closing the Admin, both inside a TemporaryClassLoaderContext to ensure correct classloading. If connection.close() throws, the exception is swallowed and logged as this warning with the client's description; close() still completes without propagating an error to the caller.

Source

Thrown at seatunnel-connectors-v2/connector-fluss/src/main/java/org/apache/seatunnel/connectors/seatunnel/fluss/source/FlussAdminClient.java:171

    @Override
    public void close() {
        // Best-effort: this client only backs short-lived schema/offset discovery, so a failure to
        // release it must not fail a job (or a discovery) whose data was already fetched. Both the
        // admin and the connection are always attempted; failures are logged, not thrown.
        // Pin the connector classloader (see createConnection): teardown may run on a framework
        // thread whose context classloader is not the connector's, and Fluss can lazily load
        // classes while closing.
        try (TemporaryClassLoaderContext ignored =
                TemporaryClassLoaderContext.of(FlussAdminClient.class.getClassLoader())) {
            try {
                admin.close();
            } catch (Exception e) {
                log.warn("Failed to close Fluss admin for {}", description, e);
            }
            try {
                connection.close();
            } catch (Exception e) {
                log.warn("Failed to close Fluss connection for {}", description, e);
            }
        }
    }
}

View on GitHub (pinned to cf67b549a7)

Solutions

1. Inspect the logged stack trace attached to this warning for the underlying cause (unreachable server, timeout, already-closed).
2. Confirm Fluss server reachability and that the job teardown happens after the cluster is stable.
3. Avoid closing the client twice; ensure the source reader lifecycle closes the client exactly once.
4. Typically safe to ignore — the warning only means server-side resources will be reclaimed by timeout.

// before shutdown, drop stale connections so close() is likely clean
if (!connection.is reachableCheckEnabled()) pingBroker(); // replace with your health probe

try {
    adminClient.close();
} catch (Exception e) {
    // connection-close errors are logged internally and not rethrown;
    // handle only unexpected outer failures here
    log.error("Unexpected failure closing FlussAdminClient", e);
}

When it happens

Trigger: Calling close() on FlussAdminClient when the Fluss Connection's close() throws — e.g. connection already broken, network failure to the Fluss server during teardown, or a timeout closing remote resources.

Common situations: Shutdown after a network outage; killing tasks while a connection is mid-RPC; duplicate close() invocations on an already-closed connection; Fluss server restarting at the same time the job is cancelled.

Related errors

• Failed to close Fluss admin for {}
• CLOSE_CQL_SESSION_FAILED
• Failed to close Google Pub/Sub subscriber
• Error closing MQTT client
• Failed to close Pulsar consumer.

AI-assisted analysis of apache/seatunnel@cf67b549a7 (2026-09-10). Data as JSON: /api/errors/82c5f1e8a4019ca8. Report an issue: GitHub.