[JBoss JIRA] Created: (JBMICROCONT-232) VFSUtils.addManifestLocations does not work for nested archives
by Scott M Stark (JIRA)
VFSUtils.addManifestLocations does not work for nested archives
---------------------------------------------------------------
Key: JBMICROCONT-232
URL: http://jira.jboss.com/jira/browse/JBMICROCONT-232
Project: JBoss MicroContainer
Issue Type: Bug
Components: VFS
Affects Versions: JBossMC-2.0.0.Beta10
Reporter: Scott M Stark
Assigned To: Scott M Stark
Fix For: JBossMC-2.0.0.CR1
The logic for VFSUtils.addManifestLocations is not working for wars in ears, probably all nested archives with class path manifest entries. One example in the jbossas testsuite is:
org.jboss.test.web.test.ClasspathUnitTestCase
2008-01-24 16:09:03,284 DEBUG [org.jboss.virtual.VFSUtils] Manifest Class-Path entry jbosstest-web-util.jar ignored for manifest-web.war reason=java.io.IOException: Child not found il.jar for JarHandler@1842226989[path= context=file:/home/svn/JBossHead/jboss-head/testsuite/output/lib/manifest-web.ear real=file:/home/svn/JBossHead/jboss-head/testsuite/output/lib/manifest-web.ear]
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
18 years, 5 months
[JBoss JIRA] Created: (JBAS-5180) Make ClusterPartition mbean queryable
by Brian Stansberry (JIRA)
Make ClusterPartition mbean queryable
-------------------------------------
Key: JBAS-5180
URL: http://jira.jboss.com/jira/browse/JBAS-5180
Project: JBoss Application Server
Issue Type: Feature Request
Security Level: Public (Everyone can see)
Components: Clustering
Reporter: Brian Stansberry
Assigned To: Brian Stansberry
Fix For: JBossAS-5.0.0.Beta4
>From Greg Hinkle:
"One thing i will suggest for the future (i've been telling everyone this) is to make the mbeans queryable. "jboss:service={PartitionName}" is not queryable since i need to know what the partition name is to find the bean... and i'll probably be looking for the bean to determine the name. Also, what happens if I name my partition "Mail"? A better idea would be "jboss:type=ClusterPartition,name={PartitionName}" or something similar."
I'll register the mbean as "jboss:service=HAPartition,partition=${jboss.partition.name:DefaultPartition}"
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
18 years, 5 months
[JBoss JIRA] Closed: (JBREM-480) Insure reliable tests in socket invoker for unusable socket connections.
by Ron Sigal (JIRA)
[ http://jira.jboss.com/jira/browse/JBREM-480?page=all ]
Ron Sigal closed JBREM-480.
---------------------------
Resolution: Out of Date
Assignee: Ron Sigal
This issue is out of date.
1. The "open connection checking" facility introduced for JBREM-690 gives a connection checking facility that does not depend on Socket.isConnected().
2. There are now explicit tests in place, on both the client and server side, that treat a version of -1 as an end of file condition.
> Insure reliable tests in socket invoker for unusable socket connections.
> ------------------------------------------------------------------------
>
> Key: JBREM-480
> URL: http://jira.jboss.com/jira/browse/JBREM-480
> Project: JBoss Remoting
> Issue Type: Feature Request
> Security Level: Public(Everyone can see)
> Affects Versions: 2.0.0.CR1 (Boon)
> Reporter: Ron Sigal
> Assigned To: Ron Sigal
> Fix For: 2.4.0.Beta1 (Pinto)
>
>
> The socket invoker needs to test for unusable socket connections, on both the client and server side. There are currently two tests in place, only one of which is correct. On the client side, if the shouldCheckConnection flag is set, then MicroSocketClientInvoker.getPooledConnection() and SocketClientInvoker.getPooledConnection() will call ClientSocketWrapper.checkConnection(), which will call ObjectInputStream.readByte(), which throws an EOFException at end of file, i.e., when the remote endpoint closes. If the shouldCheckConnection flag is not set, then SocketClientInvoker.getPooledConnection() will call Socket.isConnected() on its socket, and MicroSocketClientInvoker.getPooledConnection() makes no test. In any case, Socket.isConnected() does not appear to be useful, since it seems that a Socket, once connected, never returns to a disconnected state (at least in linux and jdk 1.4). On the server side, ServerThread.acknowledge() will call ServerSocketWrapper.checkConnection(), which behaves like ClientSocketWrapper.checkConnection(), if the shouldCheckConnection flag is set.
> The problem on the client side is that if the performVersioning flag is set, a version byte will be read by calling InputStream.read(), which returns -1 at end of file instead of throwing EOFException. If an InputStream at end of file is not tested for end of file, -1 will be read and treated as an incorrect version. resulting in an Exception. There needs to be an explicit test for -1.
> On the server side, a version of -1 is set to 1, and the end of file condition is caught later. It would be simpler to test explicitly for -1.
> In any case, we need to verify that there are reliable tests in place for unusable connections. It is difficult to see how to write a unit test that creates an error condition on a socket, but there should be unit tests for client side and server side end of file conditions.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
18 years, 5 months
[JBoss JIRA] Created: (JBAS-4762) Make distributable webapps depend on the clustering cache
by Brian Stansberry (JIRA)
Make distributable webapps depend on the clustering cache
---------------------------------------------------------
Key: JBAS-4762
URL: http://jira.jboss.com/jira/browse/JBAS-4762
Project: JBoss Application Server
Issue Type: Task
Security Level: Public (Everyone can see)
Components: Clustering, Web (Tomcat) service
Reporter: Brian Stansberry
Assigned To: Brian Stansberry
Fix For: JBossAS-5.0.0.Beta3
In 4.x, in the all config the JBossWeb service itself depends on the jboss.cache:service=TomcatClusteringCache mbean. In AS 5 this is not viable, as the relevant bean is the TomcatDeployer, which can't/shouldn't depend on a deploy folder service like the cache. In any case, it's really the <distributable/> webapps that depend on the cache, not the deployer.
The TomcatDeployer should be adding a dependency on the cache to deployments where WebMetaData.getDistributable() is true (probably in superclass AbstractWarDeployer.deployWebModule()).
A problem with this dependency-based approach is it breaks the current "feature" whereby apps marked <distributable/> can be deployed in the default config. When this is done, the 4.x deployment code detects the absence of the cache and falls back on using the standard non-distributable session manager. That won't work if we introduce a deployment dependency.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
18 years, 5 months
[JBoss JIRA] Created: (JBREM-798) Implement quick Client.removeListener() for polled callbacks, part 2
by Ron Sigal (JIRA)
Implement quick Client.removeListener() for polled callbacks, part 2
--------------------------------------------------------------------
Key: JBREM-798
URL: http://jira.jboss.com/jira/browse/JBREM-798
Project: JBoss Remoting
Issue Type: Feature Request
Security Level: Public (Everyone can see)
Affects Versions: 2.4.0.Beta1 (Pinto)
Reporter: Ron Sigal
Assigned To: Ron Sigal
Fix For: 2.4.0.Beta1 (Pinto)
This issue is a continuation of JBREM-757. There are two additions:
1. The unit tests org.jboss.test.remoting.transport.http.timeout.HttpQuickDisconnectTestCase and org.jboss.test.remoting.transport.socket.timeout.SocketQuickDisconnectTestCase did not properly test the the quick version of org.jboss.remoting.Client.removeListener() in the case of polled callbacks.
2. The facility should be tested for the RMI transport when per invocation timeouts are implemented for that transport. (See JBREM-701.)
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
18 years, 5 months