Chapter 3. Known Issues

The JBoss A-MQ GA distribution is missing bin/contrib scripts (for installing the broker as a service). The missing scripts have been made available in the rollup patch JBoss A-MQ 6.3.0R1 and all subsequent rollup patches.

When a master/slave broker with shared storage fails (unclean shutdown) it can cause a missing kahadDB journal. The issue was originally reported in a Fabric8 environment, but it also affects standalone Fuse instances. You can work around this issue by setting ignoreMissingJournalfiles=true in the kahaDB configuration. Without this setting the broker would refuse to start after failure, if it detects a missing journal.

When upgrading to 6.3, the message broker might report a missing journal files error in the KahaDB store. If you encounter this error, stop and restart the broker (thus forcing the index to rebuild). This issue occurs because of some additional checks we put in place for the KahaDB store.

When the subscriber to a topic is a Qpid C++ client, the client receives only the first message sent. This issue is a regression in 6.3 and will be fixed in an upcoming patch.

Do not set the option disableAsyncDispatch="true", as this can lead to a deadlock condition.

The new qpid-jms-client is supported in FUSE 6.3 on only JDK 1.8.

When using the profile-import Karaf console command with Windows paths (using backslashes), the command fails, because the backslashes are left out. The workaround is to use forward slashes instead. For example:

JBossFuse:karaf@root> profile-import file:C:/jboss-fuse-6.3.0.redhat-069/out.zip

In the Fuse Management Console (Hawtio console) on the Apache Karaf container, when you navigate to the Jetty menu tab, the URL links for the Jetty applications shown on this page are incorrect.

By default, child containers created using the admin:create Karaf console command do not support the new JBoss A-MQ patching mechanism, because the requisite patch feature is not available. You can work around this limitation by creating a child container using the following command:

admin:create --featureURL mvn:io.fabric8.patch/patch-features/1.2.0.redhat-630187/xml/features --feature patch test

Where 1.2.0.redhat-630187 might need to be replaced by whatever version of fabric8 you are using in your container.

If you click on the Jetty tab of the Fuse Management Console (Hawtio), the links appearing in the Url column of this page are incorrect.

The Camel JMS component is affected by a regression error that can cause problems integrating with Oracle AQ. This issue has already been fixed in the Apache Camel community project and the fix will be made available in a patch for JBoss A-MQ 6.3.

The patching functionality in the Fuse Management Console (Hawtio) has been disabled since JBoss A-MQ 6.2.1. This is because the Hawtio UI has not been updated to use the new patching mechanism. To install Fabric patches, use the Karaf console instead. This issue will be fixed in a future release.

When a JBoss A-MQ 6.2.1 container is deployed on RHEL Openstack Platform, it can take a long time (several minutes) for the container to start up. A workaround is to add the following option to the JVM that starts the container (for example, by setting the JAVA_OPTS environment variable):

-Djava.security.egd=file:/dev/./urandom

The extra dot in the pathname, as in /./, is required.

To connect to the ZooKeeper server, you must change the connectivity type from IPv6 to IPv4. For the detailed solution, see the following Customer Portal article: Cannot create fabric in JBoss Fuse 6.1 Beta on Windows 7

The Client session is dropped after fabric:create is run. The workaround is to reconnect to the karaf console after fabric is created.

If the zooKeeperPath property in the io.fabric8.gateway.http.mapping-apis.properties is changed, a gateway node restart will be required.

Insight Elasticsearch page is not supported in IE. The error occurs when clicking rectangles with numbers at Insight - Elasticsearch page.

The camel-leveldb feature is not supported on IBM AIX, HP-UX, and Oracle Solaris operating systems.

RHEL customers need to upgrade openjdk to java-1.7.0-openjdk-1.7.0.79-2.5.5.3.el6_6 or later which will fix the SIGABRT.

After restarting OpenShift, containers created from Hawtio cannot be started. This happens because the containers started before the ZooKeeper starts running will fail to start. The workaround is to wait for the start and keep restarting until we get connected to ZooKeeper.

The fabric:mq-create command requires that the Java keytool command-line utility is provided on your PATH. If necessary, you can work around this requirement by specifying the --no-ssl option (which disables support for the SSL/TLS protocol).

The Qpid C++ client cannot be compiled from sources through the Client or ComputeNode channels on Red Hat Enterprise Linux (RHEL) 6 (due to a missing dependency on the package, nss-devel). In order to compile from sources, it is recommended that you install the Qpid C++ client through the Server channel instead.

When you install and use the jmx.restart feature, you might encounter errors.