This release note contains information specific to the Borland Deployment Platform 6.0 Release Pack 1. This release of the Borland Deployment Platform 6.0 service pack 1 addresses most of the customer-reported problems and includes the value-added features set.
This release note contains the following sections:
The Borland Deployment Platform includes the following products:
For Borland Janeva Release Notes, see http://www.borland.com/devsupport/janeva/relnotes/6_0.html
Borland Deployment Op-Center 6.0 Release Pack 1 is an update release containing bug fixes to the original 6.0 release. It is available as a new install or as an upgrade install to an existing 6.0 product.
Borland Enterprise Server 6.0 Release Pack 1 is an update release that includes bug fixes and enhancements to the original 6.0 release and provides certification for JBuilder 2005. Certification for JBuilder 2005 means that Release Pack 1 is not certified for JBuilder X.
The BES 6.0 Release Pack 1 is available as a new install or as an upgrade install to an existing BES 6.0 product.
For installation instructions for Windows, UNIX and Linux, go to: http://techpubs.borland.com/am/bes/v6/
and chooseBES Installation Guide from the Select a title
dropdown.
For specific installation instructions on other platforms, check the "Documentation Updates" on the following BES Technical Publications pages:
http://techpubs.borland.com/am/bes/v6/
BES and Op-Center must be licensed and registered before they will run; there is no grace period.
For instructions on how to license your product, see: http://www.borland.com/devsupport/bdpcommon/faq/6.0/licensing/index.html
43159: License file destroyed when accessing license from multiple C++ processes simultaneously
42699: License file destroyed when accessing license from multiple threads in a C++ process simultaneously
1252: C++ servers fail to start with license violation when using FlexLM licensing on RedHat Linux
1226: C++ servers cannot import license from .txt or .eml file
1176: Memory leak in C++ license manager
43298: NullPointerException when traversing CMR fields
43240: MalformedURLException when deploying .war based on Java Server Faces (JSF)
43159: License file destroyed when accessing license from multiple C++ processes simultaneously
43038: OTS fails to start with license violation when using FlexLM licensing on HP Itanium
42927: JCA connections are returned to free pool prematurely
42873: iisredir2 IiopRPCTimeout attribute added
42757: Transaction rollback when number of beans participating in txn exceeds 128
42699: License file destroyed when accessing license from multiple threads in a C++ process simultaneously
42570: JSP compilation from JBuilder X fails when path contains underscore
42505: iisredir2 session IOR erroneously picked up from referrer header instead of cookie header
40537: Server hangs following number of busyTimeout expirations
1724: ServletException running webapp with signed .jar file
1420: CMR problems when cache contains > 100 items
1365: 129th edit fails of one of first 128 edits was on same bean
1252: C++ servers fail to start with license violation when using FlexLM licensing on RedHat Linux
1226: C++ servers cannot import license from .txt or .eml file
1714: Apache 2.0.50 not supplied with BES 6.0
1188: IIOP connector does not function properly using https when chunking is enabled
1176: Memory leak in C++ license manager
1083: IIOP connector does not handle request parameters sent as post data when chunking is enabled
This version includes all fixes contained in the VBE_06.00.00.P1 patch.
43160: The cache typecode implementation neglects that a typecode for GIOP 1.0/1.1 could have an empty repository id
41782: VBC should allow application to handle SIGUSR2. Note: Made VISThread::terminate() a no-op for all platforms.
42170: crash in calling create_recursive_sequence_tc() resolved.
42887: Server throwsBAD_PARAMin AIX platform when running in IPV6 address Note: To provide property "vbroker.connection.ifname" in specifying network interface name for the scope id if IPv6 address type is configured as Scope:Link on Linux platform. Default is "eth0".
43077:VISThreadPoolDispatcher::do_work()may return null, and will need to check for null.
41080: idl2cpp fix. Generated AnyHelper for a boxed value type has a value() method returning an invalid address
43127: Check "servant" for null before calling "_remove_ref()".
42887: Server throws BAD_PARAM on AIX platform when using ipv6 address.
41247: Fixed version display error for HP and AIX.
37208: PolicyManager.set_policy_overrides() does not comply to the CORBA 2.5 spec
40214: Random port opened even when specifying:vbroker.se.iiop_tp.scm.iiop_tp.listener.portproperty.
Contains updated version of Certicom libraries (3.1.8) used with VBC++ security
Contains ITS Support for RedHat IA-64 Linux platform
Includes fixes in VBE_05.02.01.p11
dar file (jndi-definitions) file.
For example, if the localhost name is myhost and you specify
localhost:7222 in the jndi-definitions file, you must also
specify localhost:7222 for the Tibco Managed Object in the
Configuration file. Similarly if you specify myhost:7222 in the
jndi-definitions file you must also specify myhost:7222 in the
Configuration file. (Case ID: 41633)
If there are as many concurrent transactions in the Partition as the VBJ thread max of the Partition, then synchronization callbacks that need to happen as part of the transaction commit process, can not happen.
In other words, a situation can happen where all the threads in the
Partition are trying to call commit(), and hence wait on the
transaction service to finish the commit process, and transaction service is
waiting on the Partition process to allow synchronization callback threads to
enter the VM. Due to this cyclic dependency, the Partition process will appear
to hang.
As mentioned, this situation may only happen when the transaction service
is configured to run as an out-of-Partition process, and the Partition has a
small value for VBJ threadMax. Usually it is rare to have an out
of process transaction service, but should that be a requirement, the
situation can easily be avoided by keeping the Partition's VBJ threadMax at a
high value or by keeping the threadMax un-bounded.
Note that VBJ creates only as many threads as are necessary to meet the
concurrency. So keeping the threadMax un-limited does not mean
there will be too many threads. (Case ID: 40300)
When OTS is used as the external transaction manager, it must also have its thread dispatch pool size configured with a high value for applications involving high transaction concurrency. This can be achieved using the VisiBroker property:
vbroker.se.iiop_tp.scm.iiop_tp.dispatcher.threadMax
Set this property to 50 or 100, depending on the
expected concurrency level, in the OTS related vbroker.properties file
specified through -DORBStorage parameter. (Case ID 41605)
This problem seems to be due to an incompatibility between the jar files produced by the jar command on Solaris and Windows. This appears to be a JDK bug (Case ID 39631).
javax.jms.JMSException: java.net.ConnectException: Connection
refused: connect:localhost:2506
This problem can be reproduced in a standalone client outside of BES. It seems to occur when the number of connections required by the Sonic client exceeds the current number of available connections.
In our setup, if we run the client with 21 threads specified, the exception occurs. However, if you quit that client and wait for all messages to be consumed from the queue and then rerun the same client with 21 threads, it works without any reported exception. The same is true if you specify 30 threads.
The first run of the client produces the exception, but subsequent runs generate no exception (Both BES Partition and SonicMQ should not be shutdown between runs). (Case ID : 41376). A case (W311114101) has been filed with Sonic support.
Please contact Sonic support for more details, current workaround is to have SonicMQ with large number or unlimited connections.
Note: The default commit mode in BES is "Shared" (option B).
The symptom of this problem is the unexpected occurrence of
javax.ejb.ObjectNotFoundException in situations where a
transaction involves large number of entity beans.
ejb.maxBeansInTransactions, or disable
transaction cache compaction by setting this property to 0
(zero). (Case ID 41269).
CORBA.TRANSIENT error sometimes occurs, and the DAR is not
loaded.
In this case, the DAR does get deployed, but to load it you must restart the Partition.
If your Partition requires secure transport (EJB container sometimes does),
then change the security profile for your Partition to the
ssl_enabled profile:
out of memory
exception will be thrown in this case.
For example, compiling JSPs in a WAR with 370 JSPs may use as much as 1.5GB of memory. There are two workarounds:
ejb-borland.xml)
will create the proprietary file, but will include some extraneous information
in that file. This jar will deploy and function correctly.
This function is accessible on the right-click menu of the Partition. Executing this function will not cause a stack trace to be dumped, and will prevent the Partition from being stopped normally.
Web clustering is fully functional on all other platforms.
-ui none option.
See the Petstore cluster for an example.
Startup Type = Manual. Also be sure all the Recovery actions are
set to: Take No Action
If a licensed component is configured to access a license file on another system, for example, by way of a network mounted drive, the following error will be reported:
Cannot start: Borland Enterprise Server License error[1023]:
Sanctuary Initiation Error :License storage {0} does not apply to current
user. Please register your Borland product(s).
If this occurs, the licence files will be marked as unusable, and you must re-register that license to activate it again.
As a precaution, after registering your product, you should save a copy of
the <install_dir>/var/borland.lic and
borland.pkg files to another location.
Should the license file corruption occur again, licensing can be restored
by simply copying these files to the <install dir>\var
folder.
In the License Manager GUI (in the Borland Management Console by way of the
Tools/License Manager menu command, you can right-click on a license and
delete it. Or on command line: <install_dir>/bin/lmadm -r
info). However, this does not fully delete the license. To delete a
network-based license, you must manually delete the regxxx.txt
file and/or the concurrent_xxxx.slip file in the
<install_dir>/license folder.
Deletion of node-locked licenses does not have this problem.
Disk Space Information (for Installation Target):
Required: 605,149,820 bytes
Available: Error!
You should check your available disk space manually at this point. If you have sufficient disk space, installation will proceed normally.
In this case, the resulting install log will contain a warning:
<action name="Check Disk Space"
status="warning">
This warning can be ignored.
hosts file.
When running the BES Server on Linux, the Borland Management Console and
iastool are sometimes unable to discover that server when run from a different
system. This is because the default Linux hosts file has only the loopback
entry 127.0.0.1
To correct this, you must add an entry for the local host. For example, on
a machine named mylinux that is called
mylinux.work.mydomain.com at 172.1.1.1, the
following must be added to the hosts file and must be the
first entry in that file:
172.1.1.1 mylinux.work.mydomain.com mylinux
VisiBroker known issues:
Click here for Windows and Solaris.
Click here for HP-UX on PA-RISC and IA, AIX, SuSE on x86, Red Hat on x86 and IA, Windows on IA and Windows-BCB.
Click here for SLES 8 on zSeries.