Before you install StreamBase:
-
Read this Release Notes document, paying special attention to the Issues and Cautions section.
-
Confirm that your target system matches the hardware and software requirements discussed on Supported Configurations.
-
Read the Installation Guide.
-
For a summary of what's new in this release, see New and Noteworthy.
-
If you have used a previous StreamBase release, see the important information in Migrating StreamBase .
-
If you are new to StreamBase, read Getting Started with StreamBase Studio after installation.
See the documentation home page to begin exploring the StreamBase documentation.
If you are reading this document in printed form, note that the online version of the StreamBase documentation is available in several ways:
-
On the installation CD, open
doc/index.htmlin a browser. -
From the installed file system, open the documentation home page in a browser. The default locations are:
UNIX /opt/streambase/doc/index.html Windows C:\Program Files\StreamBase Systems\StreamBase
.n.m\doc\index.html -
In StreamBase Studio, select → from the top menu.
-
On Windows, open → → → →
-
On UNIX at a terminal window's command prompt, run
sbhelp &.
Take note of the following issues in the current StreamBase release.
Release of StreamBase Studio starting with 5.x use a new Eclipse workspace format that is incompatible with the workspace from previous releases. When Studio 5.x first starts and prompts for the location of its workspace, use the provided default location or specify a new, empty directory. Do NOT specify the location of any previous StreamBase Studio workspace. Studio 5.x issues a warning dialog if it detects an older format workspace.
Contact Information contains information about available support options. You can send us e-mail or get information from your StreamBase Systems representative about our support portal.
| Fixed in 5.1.10 | |
|---|---|
| Number | Resolution |
| SB-11908 |
The java-prepend-classpath parameter
in the StreamBase Server configuration file was
introduced in release 5.1.1 as a workaround for then-new
behavior. This parameter was found to be unnecessary and was
removed in later releases, but typechecking failed to flag the
parameter as incorrect. As of this release, using the
java-prepend-classpath parameter is
correctly flagged as a configuration file error.
|
| SB-11898 | In previous releases, when performing updates to a memory Query Table, if an error was thrown while evaluating an expression, the original row being updated was deleted from the table. This was corrected, and errors thrown during Query Table updates now leave the original row unchanged. |
| Fixed in 5.1.9 | |
|---|---|
| Number | Resolution |
| SB-11615 | The Wombat adapter was preventing users from connecting to a feed that used a separate transport for downloading the data dictionary. The cause was identified and corrected. |
| Fixed in 5.1.8 | |
|---|---|
| Number | Resolution |
| SB-11584 |
A problem with dequeuing tuples was caused by the following limitation in the Java Virtual Machine (JVM):
This release of StreamBase includes a workaround for this Java problem. |
| SB-11532 | If an input adapter was configured with two or more worker threads, and had a wide output schema such that overflow fields were used, a race condition could develop, resulting in an IndexOutOfBoundsException error. The cause was identified and fixed. |
| SB-11448 |
In releases prior to 5.1.8 StreamBase Studio did not examine the application module parameters when resolving parameters in Java adapters or Java operators. The application module parameters are defined in the Parameters tab of the EventFlow Editor. This problem was fixed. |
| SB-11392 |
StreamBase Server failed to start due to incorrectly
parsed JVM arguments if the jvm-args parameter in the java-vm section of the configuration
file contained any JVM path setting that contained spaces. It did
not matter whether the setting was explicit or specified by means
of an environment variable. This was fixed.
|
| SB-11369 |
If you generated a precompiled (.sbar) file, then changed any parameter settings
in the data-sources section
in the configuration (.sbconf) file
for the application, and then ran the precompiled file, it was
likely that the behavior of the application was not as expected.
This was fixed.
|
| SB-10884 |
This release adds a way to control the size of the internal
queue used between a Java StreamBaseClient's main thread and
the dequeuing thread. Previously, you could control the number
of batches of tuples in the queue using the Java property
To remedy this situation, there is a new Java property in the StreamBaseClient API:
The default value is 10000 tuples. This property limits the absolute number of tuples in the queue, promoting better client behavior. This same Java property also affects the StreamBase-to-StreamBase input adapter. There is also a new Java property for the server:
This property limits the number of packets the server includes in any one dequeue packet. The default value is 10000, which matches the default of the new client property described above. Administrators can set both server and client with matching values to make sure client dequeues are not overwhelmed. |
| SB-10564 | In StreamBase Studio you could not delete values from the enumerated list for a field in a schema in a feed simulation process definition. Also, if there were more than 35 fields in the schema, Studio failed to display the dialog for creating the enumerated list for a field. This was fixed. |
| #73006 | Very large feed simulations with more than about 160 fields could crash StreamBase Studio when you clicked the Customize Fields button. This has been fixed. |
| Fixed in 5.1.7 | |
|---|---|
| Number | Resolution |
| SB-11339 | The <jar> and <dir> elements in the StreamBase Server configuration file are used to add classes required by custom operator and embedded adapter classes. Prior to 5.1.7, there were situations where the added classes were not made visible to the custom classes, resulting in custom classes that would fail to load or operate correctly. As of release 5.1.7, these visibility problems were eliminated. |
| SB-10519 |
When using the EMS/JMS adapters, it is now possible to specify a
timestamp format for use with ToJMSMessageConverter.
|
| SB-10443 |
The shutdown(String
containerName) method of the StreamBaseAdminClient class was previously
documented to shut down the containing server if the specified
container is the last container. This method now performs the
same operation as removeContainer().
|
| SB-9106 | The sb-ems-enqueue command (part of the EMS/JMS adapter package) now emits log messages with connection information for every connection and disconnection from a JMS server. |
| SB-9031 | In previous releases, when using the EMS/JMS embedded reader adapter and the EMS/JMS external enqueuing adapter, incoming strings too large for the corresponding StreamBase field were silently set to null. It is now possible to configure these adapters to truncate incoming messages to fit. |
| SB-9006 | The EMS/JMS adapter now supports the TIBCO EMS RELIABLE_DELIVERY mode. |
| Fixed in 5.1.6 | |
|---|---|
| Number | Resolution |
| SB-10695 | UTF-8 characters stored as strings were being doubly interpreted when using the JDBC Adapter. This was corrected. |
| SB-10694 | Feed simulations with file data source and a specified data rate were not correctly interleaving in multiple stream simulations. This was corrected. |
| SB-10638 |
A large StreamBase application with many client
enqueue connections would cause the Statistics manager thread to
show increasing CPU usage over time, and sometimes result in a
connection error stating Too many open
files. The root cause was identified and fixed.
|
| SB-7569 | The Reuters RFA Publishing Adapter was experiencing crashes when it received a tuple with a Null item field. This was fixed. |
| Fixed in 5.1.5 | |
|---|---|
| Number | Resolution |
| SB-10669 | Using the control portion option in the properties for the CSV Reader Input adapter was creating an elusive port in the Studio canvas that was hard to select. The cause was identified and fixed. |
| SB-10649 | Performance of time-based materialized windows was improved with a change in the way they were indexed. |
| SB-10579 | Stream throughput was not optimal for a stream with a very wide schema. An internal change in the handling of wide tuples was implemented, resulting in significant performance improvement. |
| SB-10536 | The Reuters Schema Designer now has improved error reporting when a single invalid RIC is entered. |
| SB-10496 | Materialized windows created in StreamSQL were not appearing in the Outline View in StreamBase Studio. This was fixed. |
| SB-10447 | A problem loading a custom Java function from a directory whose name contains spaces was corrected. |
| SB-10357 | A problem running a feed simulation with sbfeedsim using a very large data file was causing an NPE error. The cause was identified and fixed. |
| SB-10339 | The New Java Operator wizard, given a default value entry of one double quote, would generate invalid code with non-standard formatting. The cause was identified and fixed. |
| Fixed in 5.1.4 | |
|---|---|
| Number | Resolution |
| SB-10508 | In previous releases of StreamBase 5.1.x, the StreamBase Java Client API would mis-report a null value in a Timestamp field as being a nonsensical value, rather than null. This issue was fixed. |
| SB-10397 | Under certain circumstances, the Pattern Operator time window was failing to close within its specified time. The cause was identified and fixed. |
| SB-10388 | Disk-based query tables, where the primary key was not the first column in the table schema, would deadlock during certain read queries. |
| SB-10386 | The StreamBase ActivFeed Adapter was recompiled to use version 1.8.3.1 of the Activ APIs, the latest available version. |
| SB-10340 | Java code generated by the StreamBase Java Operator wizard could have non-standard line spacing, depending on your Eclipse settings and operating system. The cause was identified and fixed. |
| SB-7561 | In previous releases, if you used the reconnect interval parameter with either the StreamBase-to-StreamBase Input or Output adapters in a remote container connection URI, an invalid parameter exception was thrown. This was fixed. |
| Fixed in 5.1.3 | |
|---|---|
| Number | Resolution |
| #73359 | Small usability issues with the Query Prefix functionality were resolved. |
| #73244 | StreamBase Studio could display a Cannot Save error dialog when changing indexes on a Query table. This was fixed. |
| #72075 | For certain adapters, adding an Error Output Port could inadvertently also add an extra non-error output port. The cause was identified and fixed. |
| #70675 | StreamBase Server, when started from the command line in debugging mode (sbd -d) sometimes failed to release TCP sockets when many containers were added and removed. The cause was found and fixed. |
| #70533 |
You can use →
(or the
toolbar button equivalent) to display a list of all
StreamBase-related files in all projects in the
current workspace. Formerly, the list of files displayed included
all sbapp, ssql, and sbfs
files. As of release 5.1.3, the list includes sbconf files as well.
|
| #56829 | StreamBase Studio now issues a warning if you attempt to paste text into a Properties View field if the pasted text contains any multibyte characters that are not supported in the EventFlow file format. |
| Fixed in 5.1.2 | |
|---|---|
| Number | Resolution |
| #72363 | In the previous release, reads on Query Tables did not set the security tag correctly. This was fixed. |
| Fixed in 5.1.1 | |
|---|---|
| Number | Resolution |
| #64234 | StreamBase Studio sometimes failed to detect field mismatches in query operators correctly. Such queries could fail at runtime, or the EventFlow could not be converted to StreamSQL. These errors are now typechecked correctly. |
| #60328 | Changing a stream used by a dynamic variable sometimes failed to trigger a typecheck. |
| #57066 |
The custom-java-function example formerly demonstrated a custom
function, which was called random(). The random() function is now a
built-in function provided by StreamBase. Therefore,
the custom-java-function sample has been updated to feature a
different custom function called hypotenuse().
|
| #52670 | Explicit schema typechecking was field-order sensitive. Now, schemas do not have to have the same field order to match. |
| #37350 | Studio sometimes failed to automatically typecheck in a module that was referenced by other applications currently open in other EventFlow Editors. |
| #32492 |
The API documentation for getOperatorProperty was misleading. It gave the
impression that this property could be used for non-Java
operators, when in fact the command is restricted for use with
Java operators and embedded adapters only.
|
| #14807 |
Very large applications that ran in standard mode sometimes
failed to start in debug mode, with errors such as code too large. This was fixed.
|
For the list of issues resolved in previous releases, see:
This section describes known limitations in the current release of StreamBase. Each item includes a tracking number, description, and whenever possible, one or more workarounds.
| SB-12016 | Studio HTML display issues with Linux and Firefox 3.x | |
| Description |
Recent Linux distributions use Firefox 3.x as the default browser. Firefox 3 splits the display engine into a separate piece called Xulrunner so that it can be called by applications other than Firefox. In a default Linux installation, this can cause some display problems with StreamBase Studio. Some of the symptoms are:
|
|
| Workaround |
Install the xulrunner compatibility package for your Linux distribution. For example, for Ubuntu 8.04, use the following command: sudo apt-get install xulrunner-gnome-support |
|
| SB-10697 | .NET cache files remain after uninstalling. | |
| Description |
After uninstalling StreamBase on Windows, the .NET
libraries are left behind in C:\Windows\assembly. There are no ill effects of
these files, other than taking up disk space.
|
|
| Workaround |
To reclaim the disk space, remove the files manually. Navigate to
C:\Windows\assembly in Windows
Explorer and remove the assembly named sbclient. Removing the assembly removes all files
associated with the assembly.
|
|
| SB-10468 | Some SQL queries can fail after reconnecting to the database server. | |
| Description | When using the JDBC operator and a JDBC data source, be aware that some SQL queries and operations that succeed after first connecting to the database server can fail after reconnecting following an error. In StreamBase releases before 5.0, StreamBase Server shut down when an error occurred. Starting with 5.0, the server does not shut down on error by default, but the connection between the JDBC operator and the database server is closed when a SQL error is thrown. The issue is that both the database server and StreamBase Server can continue to throw errors while the connection is closed, for as long as the connection is closed. Reconnection attempts encounter these as new error conditions, which can cause new queries to fail. | |
| Workaround |
Restore the connection by setting the <param name="jdbc-reconnect-regexp" value=".*" /> |
|
| SB-10363 | No StreamBase data type corresponding to Sybase timestamp. | |
| Description |
When used with Sybase data sources, StreamSQL fails to convert
timestamps, resulting in typecheck errors. This is because the
Sybase timestamp is based on the Sybase
varbinary data type.
|
|
| Workaround |
In StreamSQL code that will be used for Sybase data sources, use
the Sybase datetime data type
instead of timestamp.
|
|
| #71055, SB-7509 | Missing C++ library errors on Red Hat Linux | |
| Description | On Red Hat Enterprise Linux systems, the StreamBase Server may fail to start with an error message about a missing libstdc++.so.5. | |
| Workaround |
Install the compat-libstdc++-33
package from your Red Hat distribution. If you use the package
provided on the StreamBase installation media, you may
need to create a symbolic link to your system's library directory.
For example:
or
|
|
| #69299, SB-7446 |
Starting a StreamBase application fails with a
Not enough space error.
|
|
| Description | On small systems, StreamBase may run out of virtual memory during the pre-compilation phase. This is reported as a CompilationException: I/O exception: Not enough space error message from StreamBase Server. | |
| Workaround | Increase the size of your system's paging file. | |
| #65757, SB-7363 | Failure-to-load message from the Wombat adapter is found in the Error Log view on Studio startup. | |
| Description | This error message is a known consequence of loading the Wombat adapter without also configuring the Wombat MAMA middleware to be visible to StreamBase. | |
| Workaround | Follow the instructions in Wombat MAMA Input Adapter in the Adapters Guide to eliminate this error message. | |
| #62441, SB-7266 | Operator configuration properties evaluated at compile time | |
| Description |
Operator properties are evaluated at compile time rather than at
runtime. This can cause confusion with precompiled applications
(for example, using sbargen) because any operator properties
in the runtime sbd.sbconf will be
silently ignored.
|
|
| Workaround | Don't use precompiled applications. | |
| #58027, SB-7170 | Module reference sometimes fails to typecheck. | |
| Description |
StreamBase Studio may report a typecheck error on a
module reference, with a message similar to:
Duplicate stream When this message appears, StreamBase Studio refuses to typecheck the module reference and may prevent you from continuing to develop operators downstream, even if the application has no other errors and is runnable by either StreamBase Server or StreamBase Studio. |
|
| Workaround | StreamBase is investigating this limitation. For now, you might be able to work around the problem by restructuring the flow of your application, moving the location of the module reference in the graph. | |
| #56829, SB-7148 | Invalid byte error when opening EventFlow application. | |
| Description |
Studio sometimes fails to open an EventFlow application with this
error:
org.xml.sax.SAXParseException: Invalid byte 1 of 1-byte UTF-8 sequence |
|
| Workaround |
Perform the following steps to recover:
|
|
| #55871, SB-7125 | Group labels can appear truncated. | |
| Description | Group labels longer than 12 characters are truncated and elided if the group is open, and truncated if longer than 5 characters if the group is closed. | |
| Workaround | Pad the end of the group label with space characters to prevent truncation. | |
| #52959, SB-7075 | Changing perspectives can rearrange the Studio toolbar. | |
| Description | The position of toolbar sections can change when you go from perspective to perspective, especially when going from the Authoring to the Demo perspective and back. | |
| Workaround | Eclipse toolbars do not have fixed locations, so this is expected behavior. | |
| #49739, SB-6984 | Actions unavailable when all folders are selected | |
| Description | After you collapse a Group in the EventFlow editor, selects all the items in the group. However, actions performed on the group do not succeed. For example, does not delete the items inside a group. | |
| Workaround | Uncollapse the group before performing the action | |
| #48285, SB-6862 | Not all databases have native blob support. | |
| Description | For example, Microsoft SQL Server and Sybase do not have blob support. For these databases, you can write blobs into varbinary and possibly other types. But when blobs are read from a select statement, they are returned as strings, not blobs. | |
| Workaround | Add a map operator and convert the string to a blob. | |
| #37350, SB-6469 | Submodule changes are not automatically reflected in a typecheck. | |
| Description | If you modify a module that is used by other applications currently open in other EventFlow Editors, and the relationship between the two applications is more than one level deep, Studio may fail to re-typecheck the application after the first is saved. | |
| Workaround | Invoke typechecking manually in such cases. | |
| #37149, SB-6454 |
shutdown() called with a
container-qualified URI shuts down the container, not the server.
|
|
| Description |
The Javadoc documentation for StreamBaseAdminClient.shutdown() correctly
states that it shuts down the server. However, if you create a
StreamBaseAdmin client with a URI specifying a container,
shutdown() shuts down the
container, not the server.
|
|
| Workaround | Make sure the URI you use for the shutdown call does not say "default" container. | |
| #36261, SB-6838 | Studio 3.x workspaces not compatible with Studio 5.x | |
| Description | StreamBase Studio 5.0 introduced a new workspace format that is incompatible with the workspaces from previous releases. Studio 5.0 detects an attempt to use a previous version's workspace and prevents you from proceeding. | |
| Workaround | Start Studio 5.0 with a new, empty workspace. Then use File Import StreamBase Projects from StreamBase Studio 3.x. See Migrating StreamBase Studio 3.x Projects. | |
| #35859, SB-6356 | Large apps cannot be debugged in StreamBase Studio | |
| Description | When running an application in debug mode in StreamBase Studio that contains many operators (in the range of hundreds, including all modules), it is possible to cause Studio to display and log an error message about running out of handles. The error message may vary, but the symptom is that Studio is unable to display all the intermediate streams in the Application Output View or the Manual Input View. | |
| Workaround |
Limit the number of output streams the application will expose when
running in debug mode, by setting the sbd's JVM property to include
the following:
For example, setting the JVM arguments to include:
allows any intermediate stream that contains the pattern of a
|
|
| #35743, SB-6346 | Custom function C++ code typechecking is delayed until the application is generated | |
| Description | Most portions of a StreamBase application are typechecked in real time as you add features to an EventFlow or StreamSQL application. However, starting with release 5.0, typechecking for a custom function written in C++ is delayed until the application is generated and run. | |
| Workaround | None. Be aware that typechecking errors may not be reported as usual when using custom C++ functions. | |
| #35429, SB-6327 | sbd cannot run applications that generate class files with too long a name. | |
| Description | If you use modules, StreamBase can generate internal class names that are too long for the native filesystem (for example, 257 characters on Windows), causing problems running the StreamBase Server (sbd). | |
| Workaround | To avoid this problem, use small module reference names and minimize module nesting as much as possible. | |
| #33330, SB-6241 | When an application is running in StreamBase Studio on Windows, the Manual Input view and other views can lose details. | |
| Description | This is a known Eclipse bug. The problem occurs when you use the tab key. Borders around fields, other decorations, and some controls can disappear. | |
| Workaround | Minimize and then maximize the window, or resize the affected view. | |
| #27349, SB-5894 | Evaluation exceptions in Data Parallelism key and Subscribe predicate should not be handled by upstream operator | |
| Description |
If an evaluation exception occurs for a Data Parallelism key, or in
a filtered subscribe predicate, the exception is thrown and caught
by the upstream operator, and reported as an error in the upstream
operator.
One incorrect side effect is that if one predicate in filtered subscribe fails, then predicates that come after it in the list are not executed and tuples are not delivered. Similarly, if other operators are downstream and only one is set for data parallelism, the other operators may not see tuples. Which operators see tuples is effectively non-deterministic. |
|
| Workaround |
Either ensure that you do not have errors in the expressions, or
use catchexception() as a wrapper.
|
|
| #27060, SB-5882 | Tuple dimensions do not reset counter when another dimension closes a window early | |
| Description | A tuple-based dimension that closes due to other dimensions (for example, after another dimension times out) does not reset its counter. When this happens, the next window will close based on the total number of tuples received, resulting in a partially full dimension. | |
| Workaround | None. | |
| #26888, SB-5848 | Issue with Parameterized Modules and Custom Java Operators or Embedded Adapters | |
| Description | In EventFlows, you cannot enter references to parameterized modules in custom Java operators or embedded adapters. This is due to a GUI limitation, currently under investigation, that prevents you from entering the required values. | |
| Workaround |
Embed your custom Java operator or custom embedded adapter in a
StreamSQL-defined module (see the APPLY
MODULE statement in the StreamSQL Guide), and use an EventFlow module
reference, through which you optionally share parameter values.
|
|
| #26477, SB-5779 | SBAPP-to-SSQL Conversion Assistant Resets Module Parameters to Default Values | |
| Description | If you run the SBAPP-to-SSQL Conversion Assistant on an EventFlow application that makes use of module parameter references, the resulting StreamSQL application does not have any module parameter references. They are replaced by their default values. | |
| Workaround | After the conversion, edit the StreamSQL file to add the non-default module parameter values. | |
| #26255, SB-5749 | Feed Simulation Editor's Data Rate Edit and Save Issue on Linux | |
| Description | In StreamBase Studio, the Feed Simulation Editor lets you set the rate (number of tuples per second) at which data will be generated for a stream. On Linux only, if you use the keyboard to enter an integer in the Data Rate textbox, the new value might not be saved. | |
| Workaround | After clicking into the text box and entering the value, press Enter or change the cursor's focus (for example, press Tab). Then press the Save icon or enter Ctrl+S. Note that if you press Ctrl+S while your cursor is within the Data Rate text box, the new value is not saved. You must press Ctrl+S again. | |
| #25825, SB-5722 | Input Adapter output and the Application Output View | |
| Description | Input adapters that read data from files, such as the CSV Input Adapter, when run under StreamBase Studio, may start so quickly that their output is not caught by Studio, if they have completed reading the file before Studio connects to the running server. | |
| Workaround |
The workaround is to mark the adapter to start paused, and manually start it with sbadmin after the application is running.
|
|
| #25228, SB-5628 | Include Package with Custom Java Functions | |
| Description | If you use a custom Java function (simple or aggregate) with no package, javac returns a compile error. | |
| Workaround | Put the function in a package. | |
| #25009, SB-5602, #25004, SB-5599 | Embedded Adapters and StreamSQL | |
| Description |
In StreamSQL applications, you cannot specify that an embedded
adapter should start in a paused
state, and you cannot manage the state of a running adapter in a
StreamSQL application (although you can with EventFlow
applications).
|
|
| Workaround | Use an EventFlow application with embedded adapters. | |
| #15898, SB-5078 | C++ Custom Aggregate functions cannot return a string | |
| Description |
Custom aggregate functions that are implemented in C++ cannot
return a string.
|
|
| Workaround | Create the custom aggregate in Java. | |
| #14807, SB-4951 | Problem running large applications in debug mode | |
| Description |
Very large applications that run in standard mode may not start in
debug mode, with errors such as code too
large.
|
|
| Workaround | Organize the application into smaller modules. | |
| #14763, SB-4937 | Problem when tuple in top-level module consumes large amounts of CPU time | |
| Description | If the processing of a single tuple in the top-level module consumes a large amount of CPU time, contained parallel modules that are connected directly to output streams will see their output blocked until the tuple/operator releases the CPU and the tuples can be passed to the network. | |
| Workaround | Put the time-consuming computation into its own parallel module. This workaround may not solve the issue in all cases, because inter-module queues can grow very large, but may help in many cases. | |
| #14745, SB-4931 | SBAPP-to-SSQL Conversion Assistant Does Not Support Parallel Operators | |
| Description | If your EventFlow application file has individual operators that are marked for parallel operation, the SBAPP-to-SSQL Conversion Assistant cannot generate equivalent functionality in the generated StreamSQL application because there is no direct analogy for parallel operators in StreamSQL. | |
| Workaround |
After the conversion, if appropriate for your application (no
dependencies on data in other operations), you can use an
APPLY PARALLEL MODULE statement or
APPLY PARALLEL JDBC statement in the
StreamSQL file. For details on the APPLY statement, see the StreamSQL Guide.
|
|
| #14238, SB-4868 | Memory problem with external JDBC data source queries | |
| Description | StreamBase Server can fail with an OutOfMemoryError exception while running an application with a JDBC query that generates a large result set using MySQL. | |
| Workaround |
Note: Make sure you are
running MySQL Version 5 or greater. StreamBase access to JDBC data
sources does not support earlier versions of MySQL.
If you experience the problem described, consider limiting the
number of rows that MySQL can read at one time. You can do this
by enabling the new |
|
| #14207, SB-4854 | Typecheck problem with DB2 JDBC datasources | |
| Description |
SQL INSERT statements submitted to a JDBC data source in a Query
operator or in an APPLY JDBC statement may pass typechecking even
when they contain errors. This problem was detected only with an
IBM DB2 data source.
For example, the following statement passes typechecking even if the named table or fields do not exist in the JDBC data source:
APPLY JDBC DB2 "INSERT INTO mytable (id, name) VALUES({id},
{name})" FROM InStream INTO OutStream;
|
|
| Workaround | Run the StreamBase application. If an error occurs against an INSERT statement, verify that the table name is correct. You can validate the INSERT statement by running it against a test database. | |
| SB-4625 | Spurious typecheck error when saving changes to a data construct associated with an operator | |
| Description | In some circumstances StreamBase Studio generates a typecheck error with a message to remove a row that you cannot remove because the row is not displayed in the Properties view. This can occur when you have edited a data construct associated with an operator and then attempt to save the changes. | |
| Workaround | You can work around the typecheck error by forcing StreamBase Studio to reread your changes. For instance, cut and paste an existing row, or change one character and change it back. | |
| #13017, SB-4559 | Join timeout is not asynchronous | |
| Description | Join timeout in the Join operator is not asynchronous. That is, the Join operator does not remove a tuple from its buffer until the timeout has passed and another tuple arrives on the same input stream. Tuples received on one input stream do not affect the buffer of the other input stream. | |
| Workaround | None. | |
| #10674, SB-4225 | Some disk drive configurations can corrupt disk-based query tables. | |
| Description | Some caching disk drives do not guarantee the sequencing of disk writes, nor do they guarantee that all the writes to the drive have actually happened. This behavior improves disk performance, but at the risk of data corruption in the event of a power outage or equipment failure. StreamBase is vulnerable to this kind of data corruption if your StreamBase application uses disk-based query tables. | |
| Workaround | Disk drives with this behavior usually have a driver-level setting to turn this capability on or off. For example, the driver may present you with a checkbox labeled Enable write caching on the disk. Unchecking this box would prevent the problem from occurring. | |
| #10291, SB-4160 | JVM crash during StreamBase Application compilation | |
| Description | When StreamBase Server first loads an application, it compiles it to Java code using the Sun Java compiler. Some StreamBase applications can cause the Sun Java compiler to crash during this phase. If this happens, StreamBase Server displays a stack trace that mentions OopFlow::build_oop_map. StreamBase is pursuing this issue with Sun, but in the meantime if you encounter this issue, contact StreamBase Systems for assistance in modifying your application to avoid the error. | |
| Workaround | Make slight changes to your application until the error no longer occurs. We are unfortunately unable to provide guidance about what specific changes to make. | |
| #9620, SB-3955 | Finding sbd process failure details | |
| Description |
When StreamBase Studio runs an application, it launches the server
process (sbd) in the background. If
sbd fails for any reason, Studio tries
to report the reason to you. However, the actual message that
StreamBase Studio displays may not contain enough
information,depending on the failure.
|
|
| Workaround |
To get more information about the reason for the sbd process failure, open the Eclipse error log as
described in Error Log View in the StreamBase Studio Reference.
|
|
| #8174, SB-3585 | Inconsistent support for undo in StreamBase Studio | |
| Description | Many editing operations in Studio cannot be undone. This includes operations such as renaming operators, editing query tables, copying schemas, adding operators from the palette, editing feed simulations, and so on. | |
| Workaround | The Revert command in the File menu can be used to undo all changes since the last time a file was saved. | |
| #7361, SB-3308, #2585, SB-1601 | From Windows, custom-simple-function and custom-aggregate-function samples do not work with Solaris server or 64-bit Linux server | |
| Description |
On Windows, the installed custom-simple-function and custom-aggregate-function samples contain library
files (log.so and stdev.so, respectively) that are built for Linux.
These custom functions will work properly only when
StreamBase Studio is connected to a 32-bit Linux
server, not a Solaris server, or a 64-bit Linux server.
|
|
| Workaround |
To use the custom-simple-function and
custom-aggregate-function samples
with a Solaris server:
|
|
| #7346, SB-3301 | Heartbeat and Metronome do not respect pause in the debugger | |
| Description | The pause command in the debugger will not prevent Heartbeat and Metronome operators from outputting tuples due to the passage of time. This could cause unexpected results when debugging applications that make use of these operators. | |
| Workaround | None. For more information about these operators see the Heartbeat Operator and Metronome Operator documentation. | |
| #7191, SB-3260 | Using non-member Schema.Field to access a Tuple Does Not Throw Useful Exception | |
| Description |
The StreamBase client API does not throw a useful exception if you
inadvertently reference the wrong Schema while using Schema.Field to access a Tuple. Under C++, no exception is thrown and memory
corruption may occur. Under Java, an ArrayIndexOutOfBoundException
is thrown.
This applies both to using fields from other schemas, or to using the Field objects that were used to create the Schema. Fields that are used to create a schema are not fully initialized, and so you must get the actual Field from the new Schema object in order to access the fields of a Tuple. |
|
| Workaround | None via the API; you must examine your client code and verify that the correct Schema is referenced. | |
| #6860, SB-8698 | Aggregate dimensions do not support advance on infinite dimensions | |
| Description |
In an Aggregate operator's Properties View, the Dimensions tab
includes a setting for Window Size. One of the size options is
Do not close window based on this
dimension. If selected, this dimension cannot cause the
closing of a new window for the Aggregate. If a new window is never
opened, this creates an infinite sized window that never closes
(for the life of the StreamBase Server). Application designers may
set up this feature, for example, to ensure that an Aggregate
processes an entire day's volume of streaming data. However, a
limitation is that Aggregates with infinite dimensions do not
support an Advance value that you set in the Opening Policy.
|
|
| Workaround | None. | |
| #6539, SB-3121 | Mozilla issue with StreamBase Studio Welcome page and Help | |
| Description | On Linux, the StreamBase Studio Welcome page and Help require that Mozilla or Firefox is installed and can be located. If you removed Mozilla, or installed it in a non-standard location, or moved it, the Welcome page or Help may not appear. | |
| Workaround |
Before using StreamBase Studio, set the environment variable
MOZILLA_FIVE_HOME to the path of your
Mozilla installation. For example, on Linux, set it to:
MOZILLA_FIVE_HOME=/usr/lib/mozilla-1.7.3 export MOZILLA_FIVE_HOME |
|
| #5400, SB-8640 | Java Operator with zero inputs does nothing | |
| Description |
While it is not an error to create a Java Operator with zero input
ports, the processTuple method will
never be called, thus the Java Operator can never emit any tuples.
|
|
| Workaround | None. | |
| #5389, SB-2931 |
Relocatable RPMs install incorrect symlinks in /usr
|
|
| Description |
The Linux RPM packages are relocatable. This means that the default
installation location of /opt/streambase can be changed to a different
location. However, if this is done, symbolic links are still
installed in /usr/bin and
/usr/lib that point to /opt/streambase.
|
|
| Workaround | Either edit the symbolic link by hand, or use the tar file instead of the RPMs to install StreamBase on Linux. | |
| #5327, SB-2891 | Error messages while creating Java operators lack detail | |
| Description | Any exceptions during the creation of a Java operator from within StreamBase Studio will result in a "Can't make new object: (operator name)" message, with no additional information. | |
| Workaround | Run the application in sbd directly for additional diagnostic information. Verify the correct JAR files are specified in your project's Custom Libraries. Verify the classes are in the JAR files. Verify that the manifests of the JAR files are correct. For more information, see Creating Java Clients. The error log may have additional diagnostics. | |
| #4824, SB-2617 | For Java operators and modules, typechecking may occur regardless of connections | |
| Description |
Java operators and modules are typechecked independently of their
connections to other components. For modules, this may result in
errors that do not clearly describe the problem. For example, if
you have an application containing a module with its inputs
connected, then delete an input stream that the module references,
and then return to the referencing application, the arc
(connection) will be dropped. However the following typecheck
message will appear: "Error in referenced
application: <input> lacks required attribute
'stream'".
|
|
| Workaround | None. | |
| #4451, SB-2507 | Java API IOException when client program is suspended | |
| Description |
A program using the StreamBase Java API may receive an unexpected
IOException if it is suspended (using either Ctrl-Z, or kill
-SIGSTOP). The IOException will be thrown when the program
is resumed, or placed in the background. The exception may be
thrown even if there is no network traffic at the time. This issue
does not apply to Windows systems.
|
|
| Workaround | Do not suspend StreamBase clients written in Java. | |
| #4289, SB-2482 | Data sent to multiple streams may get out of order | |
| Description | StreamBase Server does not preserve the relative ordering of input tuples received on different input streams. This affects tools such as sbfeedsim, which replays recorded data. It also affects the client API, if tuples are enqueued to different streams. | |
| Workaround | In some cases you may be able to use a Merge operator to reorder your input data. However, note | |
