Known Issues

This section lists any issues for the current release that were known when this readme was published. For known issues found later, go to the Knowledge Center on the Empower website.

When you use Software AG Installer to create a Universal Messaging Docker image, the image fails to start.

The issue occurs because the logs directory is not present in the installation by default. Instead, the directory gets created when you run the container. This sets the directory's user to root, which makes the logs directory inaccessible to the Universal Messaging server in the container, because the server runs as non-root user. As a result, the Universal Messaging server fails to start and the container exits immediately.

There is currently no workaround for this issue.

The Universal Messaging server might fail to start when run as a console process on Linux due to an issue with the Java Service Wrapper version 3.5.50.

There is a workaround for this issue that requires closing the STDIN stream of the server process. To do so, edit the SoftwareAG_directory\UniversalMessaging\server*server_instance*bin\Server_Common.conf file and add the following line anywhere in the file: wrapper.disable_console_input=TRUE

Restart the server to apply the new configuration. This workaround is not applicable to deployments where access to the server's console input is required.

A Universal Messaging client may fail to receive events sent to a Universal Messaging server over AMQP via the AMQPNetLite library. The client returns an ArrayIndexOutOfBoundsException.

For a workaround to this issue, contact Software AG Global Support.

NUM-14533 (UM log messages using LOG level are logged with ERROR level when using Logback.)

When Universal Messaging is configured to use Logback as the logging framework, many of the server start-up messages in the server's nirvana.log file will be written with status ERROR. This happens due to a limitation in Logback that does not provide usage of custom log levels. Therefore, Universal Messaging messages logged with LOG level are translated to ERROR level when Logback is used.

There is no workaround for this issue.

The Universal Messaging instance manager tool may crash when run on Windows with an instance name longer than 42 characters, although the instance may still be created.

The only workaround is to use a shorter instance name.

NUM-7839 (If an exclusive asynchronous durable is created, closed, and then re-subscribed immediately, it is possible for the client to receive an exception that the subscriber is already bound.)

The issue occurs because the session disconnected operation does not wait for all resources to be cleaned. If a client waits for a while before the second subscription, all resources on the server are cleaned so the subscription is successful, but there is no way currently to ensure the exact time at which everything is cleaned.