6.2. Known Issues
Apache Server (httpd) and Connectors
- 900620 - APR natives are not being loaded if UAC is in place (Windows Server 2008 R2)
- On Windows servers, Apache portable runtime (APR) libraries will fail to load unless a user with administration rights is being used with User Account Control (UAC) disabled.
Clustering
- 901164 - Stale session data received after failover, sessions are lost
- 900707 - Property substition breaks when the string "localhost" is used as part of the default value
- Property substitution does not work correctly when the string
localhost
was used as part of a default value in the JGroups subsystem configuration. Property substitution is a feature provided for JBoss Enterprise Application Platform 6 subsystems. However not all subsystems have it enabled, and some subsystems (like JGroups) actually do some of their own substitution. However this behaviour is not standardised and differs from the supported subsystem properties. Red Hat recommends not to use property substitution with the JGroups subsystem until this issue is resolved. - 900378 - CDI beans with SET replication trigger are not replicating
- Due to a bug in the Weld component, the
setAttribute
method is not called correctly. This causes CDI beans with theSET
replication trigger to fail to be replicated. The workaround is to use theSET_AND_NON_PRIMITIVE_GET
trigger for these beans. This will be fixed in a future release. - 900549 - transient "WeldListener: java.lang.NullPointerException" with clustering and session replication
- A
WeldListener: java.lang.NullPointerException
is thrown occasionally. The client receives a HTTP response503
, along with the exception text. The cause of this bug is under investigation. - 900483 - CLONE - Stale session data received when using DIST SYNC on node shutdown
- During testing, a few cases showed that stale session data was received when a node shut down and
DIST SYNC
orDIST ASYNC
cache mode was used. This issue is still under investigation. - 959753 - DIST SYNC: Service jboss.infinispan.web.default-host/XY won't start after node restart with "Initial state transfer timed out"
HornetQ
- 959216 - Cannot send larger messages using Stomp protocol
- 959616 - Deadlock during clean shutdown of backup during activation
- An JBoss Enterprise Application Platform 6 server that is configured as a HornetQ backup server will fail to shutdown in the following situation:
- Message-Driven Beans are locally deployed on the server, and
- the shutdown command is performed during the server’s activation.
In this scenario the shutdown will result in a deadlock that prevents the shutdown process from completing. Once deadlocked, the server must be forcibly terminated. On Red Hat Enterprise Linux 6 this can be done using thekill -9
command.To avoid this situation, only attempt to shutdown the server before or after activation, and not during the journal loading process. - 959789 - HQ core bridge does not failover
- 928400 - pageSize >= maxSize causing inappropriate server state
- 912653 - Clone: Remote JNDI is not working with a HornetQ-only JMS bridge
- 922184 - Replicated journal - NullPointerException on backup side when saving add/commit message to non-existing journal
- 928288 - Replicated Journal - Illegal backup state if network failures present while initial journal replication from live
- 899966 - Problem with Netty on IPv6 link-local addresses
- Due to a JDK bug, if you use link local addresses, zone IDs are used to distinguish which interface is chosen. This problem does not affect global addresses. A workaround will be included in a future version of the Netty component.
Scripts and Commands
- 965315 - jboss-ec2-eap timeout of user script does not work properly
- The timeout feature implemented in the user script on JBoss Enterprise Application Platform's AMI images does not work. If a command in the user script hangs, this will prevent the virtual machine booting successfully. The timeout feature was intended to allow the virtual machine to continue to boot, regardless of a hung user script commmand, allowing diagnostics to be carried out on the user script. The cause of this problem is that the Bash shell does not support job control in non-interactive shells and this feature is required by the timeout implemented in the user script. No workaround is currently available but it is expected to be resolved in the next release.
- 956281 - Starting EAP 6.1 on 32bit jvm can result in failure to start JVM
- On Windows 8 32-bit JVMs, the default max perm gen space setting causes the JVM to not be created, which then prevents the server from starting. To work around this issue, remove the -XX:MaxPermSize=256m parameter from the startup parameters of the JVM. Once this is done the JVM will be created, and the server will start.
- 916960 - Standalone.bat does not work if parentheses are on path to EAP.
- This release of JBoss Enterprise Application Platform 6 has a known issue with the Windows installer and installations to a default directory containing parentheses. For example;
C: Program Files (x86)\jboss-eap-6.1\
.In these circumstances thestandalone.bat
file may fail with a message\jboss-eap-6.1 was unexpected at this time
.This is because the DOS processor interprets the ending parenthesis in the "echoed" JBoss path name as the end of the precedingif
statement.To resolve this issue:- Open the
JBOSS_HOME\bin\standalone.bat
file in an editor. - Find the following line:
echo JBOSS_HOME: %JBOSS_HOME%
- Add double-quotes around the
%JBOSS_HOME%
variable:echo JBOSS_HOME: "%JBOSS_HOME%"
This will ensure the parenthesis are interpreted correctly and thestandalone.bat
will execute successfully.
Security
- 920160 - Unauthorized access to a web application protected with a custom authorization module results in HTTP 200 (OK) instead of HTTP 403 (Forbidden)
- 963277 - Debugging Java Security Managers results in a StackOverflow on boot
- Running with a security manager enabled and
-Djava.security.debug=access:failure
results in a StackOverflow error message and an unbootable JBoss Enterprise Application Platform instance. This problem is caused because the AccessControllercontext’s debug output to the System streams causes an infinite loop while checking permissions. The root cause of this problem has been identified and it is expected to be fixed in a future release.
Apache Server (httpd) and Connectors,Documentation
- 900580 - httpd: libaprutil-1.so.0: file too short
- Apache portable runtime (APR) binaries are not natively packaged as part of JBoss Enterprise Application Platform, or JBoss Enterprise Web Server, for Red Hat Enterprise Linux. Symlinks are used to reference the base Red Hat Enterprise Linux libraries. This can result in an issue with unresolvable symlinks when trying to use APR binaries if the following packages are not installed:
apr
,apr-util
,apr-devel
,apr-util-devel
.The fix for this issue will be shipped in a future Red Hat Enterprise Linux APR update.
CDI/Weld
- 927895 - META-INF "Dependencies: org.foo.module services" doesn't work for CDI Portable extensions
- A Known Issue exists in this release of JBoss Enterprise Application Platform 6 wherein deployments with dependencies on CDI Portable Extensions delivered within modules will fail to deploy with the following exception:
org.jboss.weld.exceptions.DeploymentException: WELD-001408 Unsatisfied dependencies for type [<Extension Name>]
This is because deployments do not load CDI Portable Extensions from modules declared as dependencies.In order for an application to load a CDI Portable Extension defined in a extension JBoss Module, it is not sufficient for the application to declare a dependency on the extension JBoss Module. The application must also declare a dependency on the Portable Extension class in itsMETA-INF/services/javax.enterprise.inject.spi.Extension
file.
CLI
- 928444 - Unable to start CLI on windows if parent directory of aesh.log doesn't exist
- The CLI tool fails to launch if the parent directory of its log file does not exist. The CLI throws an exception (
IOException
) with with following error message.java.io.IOException: Couldn't get lock for C:\Users\hudson\AppData\Local\Temp\2\\aesh.log
To avoid this error, ensure that the specified directory exists before launching the CLI tool.
Clustering,EJB
- 900480 - CLONE - Extremely slow response times in Remote EJB invocation on cluster membership change
- On server shutdown or crash, membership changes can take up to one minute to complete. This issue is currently under investigation.
EJB,Naming,Remoting
- 923836 - remote:// protocol does not work over SSL with IBM JDK
- There is a Known Issue in this release of JBoss Enterprise Application Platform 6 wherein the remote naming lookup fails intermittently on an IBM JDK with the following exception:
org.jboss.remoting3.NotOpenException: Endpoint is not open.
This issue appears to present when the client uses the IBM JDK. In instances where the server is on the IBM JDK and the and the client uses another JDK, the issue does not present.The issue is being investigated for a resolution.
EJB,Remoting
- 962734 - IOException during EJB invocation causes a hang
- If a server-side EJB protocol receiver received invalid message headers from a client, a message similar to the following would be logged, and the client would hang.
15:06:46,092 WARN [org.jboss.as.ejb3] (Remoting "dhcp-4-126" task-4) JBAS014254: Received unsupported message header 0xffffffff on channel Channel ID 0c51fad8 (inbound) of Remoting connection 4589adff to /127.0.0.1:33329
When the invalid message header is received and logged, the EJB protocol receiver did not re-enrol itself on the channel, so it would not respond to any further messages. The client would be left waiting for a response which would never be sent because the original message was not processed by the server. No workaround is known for this issue but it is expected to be fixed in a future release.
EJB,Transaction Manager
- 952746 - Fix transaction recovery failures involving remote EJB resource
Hibernate
- 900321 - HHH-7287 Problem in caching proper natural-id-values when obtaining result by naturalIdQuery using NaturalIdLoadAccess
NaturalIdLoadAccess
behaves inconsistently when trying to load entities from the Level Two(L2) Cache after an update of their natural-id values from a non-transactional scenario.Hibernate currently does not support proper L2 caching of entities with natural IDs in scenarios where there is a mix of transactional and non-transactional access to the database.
IPv6 support
- 900564 - Wrong format of IPv6 addresses in log entries
- IPv6 addresses should be formatted with square brackets ([ and ]) around them, such as in
http://[2620:52:0:105f::ffff:2]:9990
. The brackets are missing in the log files for JBoss Enterprise Application Platform 6. This issue is scheduled to be fixed in a future release. - 900609 - Opened IPv4 sockets on Windows when server is bound to IPv6
- If JBoss Enterprise Application 6 is started in Microsoft Windows server with the IPv4 stack disabled and the IPv6 stack enabled, IPv4 sockets are still opened. This issue is still under investigation.
JCA
- 952277 - disable xa-data-source causes that data-source loses its attribute enabled set to false and after reload it becomes enabled
RESTEasy
- 958896 - RESTEasy: Invalid Content-Type in response - Bad behavior when invalid accept header is specified in request header
Remoting,Naming
- 913189 - Backward compatibility for JMS clients with JNDI lookups
Server
- 911496 - Change in module structure prevents JBDS to create an EAP 6.1 server
Web Services
- 900634 - JBossWS-CXF doesn't send fault message to a FaultTo endpoint when request-response message.
- If the
FaultTo
element of WS-Addressing is set to a WS client, the WS server does not send fault messages to theFaultTo
destination. However, if theReplyTo
element is set, the WS server does send the responses to theResponseTo
destination. The cause of this issue is still under investigation.
mod_cluster
- 901227 - node-timeout, worker-timeout, flush-wait and ttl mod_cluster attributes in web console
- 960243 - Regression in ProxyPass integration
- 958991 - mod_cluster core when use ProxyPass / balancer://bal and CreateBalancers 1
- Users of JBoss Enterprise Application Platform 6, who have
CreateBalancers 1
set in /conf.d/mod_cluster.conf, and who have not set thestickysession=JSESSIONID|jsessionid
configuration will encounter the following segmentation fault:[notice] child pid 7787 exit signal Segmentation fault (11)
Adding thestickysession
element to the CreateBalancers configuration will resolve the issue.For example:CreateBalancers 1 ProxyPass / balancer://qacluster stickysession=JSESSIONID|jsessionid
- 900801 - Adding modcluster via the CLI fails.
- 963720 - mod_cluster: proxy DNS lookup failure with IPv6 on Solaris
unspecified
- 900047 - SystemMemoryUsageLoadMetric is not correct on Linux/Unix
- The
SystemMemoryUsageLoadMetric
does not show useful information on Linux or UNIX operating systems. For these systems,HeapMemoryUsageLoadMetric
provides more useful information. The solution to this problem will be to change the algorithm ofSystemMemoryUsageLoadMetric
to subtract thebuffers/cache
value from theused
number. The best method for doing this is under investigation. - 901231 - Usage of finalize() needs extra guards on IBM JDK
- JBoss Enterprise Application Platform 6 can experience intermittent failures when running on the IBM JDK 6 and 7. This is because the IBM garbage collector is much more aggressive than that of other JVMs. This can sometimes result in memory being reclaimed in situations that were unanticipated and did not manifest when testing on other JVMs.Until this issue is resolved Red Hat recommends using the OpenJDK or Oracle JDK to run JBoss Enterprise Application Platform 6 on Red Hat Enterprise Linux.