Gradle FAT jar and path problem


I have strange problem while starting embedded MQTT broker by my application, when all dependencies are packed inside a fat jar created by gradle.

2024-06-11 10:16:06,707 INFO  - Starting EmbeddedHiveMQ.
2024-06-11 10:16:06,713 INFO  - HiveMQ version: Development Snapshot
[2024-06-11 10:16:09] [SEVERE] | (log.LogManager debugLog) | java.util.concurrent.CompletionException -> java.nio.file.InvalidPathException: Illegal char <:> at index 4: file:\C:\Program Files\Server\Server.jar!\BOOT-INF\lib\hivemq-community-edition-embedded-2024.5.jar!\extensions
Stack Trace:
        at java.util.concurrent.CompletableFuture.reportJoin(null:-1)
        at java.util.concurrent.CompletableFuture.join(null:-1)
        at com.client_server.server.launcher.MqttBroker.start(
        at com.client_server.server.launcher.Launcher.startServer(
        at com.client_server.server.launcher.Launcher$51$

my code to configure and start broker looks like this:

EmbeddedExtension mqttMessageInterceptorExtension = EmbeddedExtension.builder().withId("MqttMessageInterceptor").withName("Mqtt Message Interceptor").withVersion("1.0.0").withPriority(0).withStartPriority(1000)
				.withAuthor("author").withExtensionMain(new MqttMessageInterceptorExtension()).build();

EmbeddedHiveMQBuilder embeddedHiveMQBuilder = EmbeddedHiveMQ.builder().withConfigurationFolder(Path.of(conf)).withDataFolder(Path.of(data))

EmbeddedHiveMQ broker =;

The same works without issues, when hivemq libraries are outside the fat jar and they are listed with Class-Path at the jar MANIFEST.

I will appreciate any tip, how to solve the problem.

Thank you in advance!

Hello @mkokoszka

To help us investigate the issue, could you please provide the following details?

  1. The steps to reproduce the error.
  2. The repository link to your Java project.
  3. The commands you are using to build the project.
  4. The commands you are using to run the project and reproduce the error.

This information will allow us to better understand the problem and assist you more effectively.

Kind regards,
Diego from HiveMQ Team