Flux News

A maintenance release for Flux 7.9 is now available. This release includes 2 improvements and 6 bug fixes.

Flux customers can download Flux 7.9.13 from the Portal website at:

my.fluxcorp.com

Evaluation versions of the latest Flux release are also available at:

www.fluxcorp.com/download

Flux 7.9.13 is a drop-in replacement for Flux 7.9.x.

Upgrade Notes

• The CLUSTER_NETWORKING option from Flux 7.8.x (and lower) has been reintroduced in this release, and renamed to CLUSTER_NETWORKING_ENABLED. Take care to update this property name in your engine configuration if you were previously using the CLUSTER_NETWORKING option. CLUSTER_NETWORKING_ENABLED defaults to true. In most circumstances, leave the CLUSTER_NETWORKING_ENABLED property set to true.

Improvements

• Added a new configuration option, CLUSTER_NETWORKING_ENABLED, to control whether Flux engines in a cluster should attempt to communicate across the network using RMI. This option defaults to true. In most circumstances, leave the CLUSTER_NETWORKING_ENABLED property set to true. Only in special circumstances should it be set to false. By disabling cluster networking, the engine instance still participates in a cluster, but all its communication with other engine instances is performed through the database server. The flow chart real-time monitor, file transfer monitor, caching, and agents depend on cluster networking functionality. These features will be disabled when cluster networking is turned off.
• File triggers and actions that encounter an error while evaluating an include on a file criteria now log a greater level of detail, along with the specific cause of the error, in addition to the previous error message “Could not evaluate include <include>.”

Bug Fixes

• Flow charts are no longer failed over from the PAUSED or JOINING states if the engine where they were running is not disposed cleanly.
• The error message com.jscape.inet.ftp.FtpException: Could not list file is no longer unnecessarily logged when a file action or trigger runs successfully.
• The For Each Delimited Text File Action now correctly logs errors instead of causing a NullPointerException when an error occurred.
• Pausing and resuming a flow in the JOINING state no longer causes the flow to roll back to the last transaction break. Now, pausing a flow in the JOINING state correctly pauses the incoming flows that have not yet arrived.
• Flow charts now correctly run in priority order after priorities are updated in the runtime configuration on a running engine.
• Updating runtime configuration priorities on an engine now updates the priorities for flow charts that were already waiting in the engine as well as flow charts that are exported after the update.

A maintenance release for Flux 7.9 is now available. This release includes 1 improvement and 8 bug fixes.

Flux customers can download Flux 7.9.12 from the Portal website at:

my.fluxcorp.com

Evaluation versions of the latest Flux release are also available at:

www.fluxcorp.com/download

Flux 7.9.12 is a drop-in replacement for Flux 7.9.x.

Improvements

• The For Each Delimited Text File Action now provides a specific cause when an error occurs, improving the readability of error messages and making it easier to troubleshoot this action.

Bug Fixes

• The Flow Chart Action now correctly saves in the Web-based Designer when only one of the “Namespace” or “Repository Namespace” properties is set.
• The For Each Collection Element, For Each File, For Each Delimited Text File, and For Each Number actions no longer cause the error NotSerializableException when storing their results in the database.
• The default error handler no longer causes a NotSerializableException error when multiple engines are running together in a cluster.
• The For Each Delimited Text File Action now correctly supports the wild card character ‘*’ and no longer fails with the error message flux.file.FileActionException: Could not delimit the file “<pattern>”.
• The For Each Delimited Text File Action no longer fails with the error java.lang.ClassCastException when the action uses multiple file criteria.
• The For Each Delimited Text File Action can now correctly locate files that are not stored in the Java working directory.
• The engine now correctly considers the thread context class loader when deserializing custom classes, preventing a CloneNotSupportedException when a class is only available from that class loader.
• The engine no longer causes a CloneNotSupportedException when trying to persist an org.hibernate.collection.PersistentSet object inside JavaAction Key.

A maintenance release for Flux 7.7 is now available. This release includes 1 improvement and 8 bug fixes.

Flux customers can download Flux 7.7.20 from the Portal website at:

my.fluxcorp.com

Evaluation versions of the latest Flux release are also available at:

www.fluxcorp.com/download

Flux 7.7.20 is a drop-in replacement for Flux 7.7.x.

Improvements

• The For Each Delimited Text File Action now provides a specific cause when an error occurs, improving the readability of error messages and making it easier to troubleshoot this action.

Bug Fixes

• FlowChartElement objects returned from a FlowChartElementIterator no longer return only null values for the getStartOfRun() and getAverageRuntime() methods.
• Setting a concurrency throttle of 1 on the root node in a runtime configuration file no longer causes the throttle to be ignored.
• Message triggers awaiting assignment of a message no longer wait unnecessarily when messages are available to be assigned.
• The Search Logs page in the Operations Console now correctly displays results when a logger type is selected from the drop down menu.
• The forecast page of the Operations Console now correctly shows all forecasted firings in the given range, rather than only the next firing for a given flow chart.
• The For Each Delimited Text File Action now correctly supports the wild card character ‘*’ and no longer fails with the error message flux.file.FileActionException: Could not delimit the file “<pattern>”.
• The For Each Delimited Text File Action no longer fails with the error java.lang.ClassCastException when the action uses multiple file criteria.
• The For Each Delimited Text File Action can now correctly locate files that are not stored in the Java working directory.

The Flux offices will be closed Monday, 5 July 2010 in observance of Independence Day (US).

We will reopen Tuesday, 6 July 2010.

A maintenance release for Flux 7.8 is now available. This release includes 3 improvements and 17 bug fixes.

Flux customers can download Flux 7.8.4 from the Portal website at:

my.fluxcorp.com

Evaluation versions of the latest Flux release are also available at:

www.fluxcorp.com/download

Flux 7.8.4 is a drop-in replacement for Flux 7.8.x.

Improvements

• The Mail Action now supports Gmail as an SMTP provider.
• Automated emails for Business Process Triggers can now be disabled by setting the DASHBOARD_MAIL_SERVER to a null or empty value.
• Business Process forms now support file uploads up to 10 megabytes in size.

Bug Fixes

• FlowChartElement objects returned from a FlowChartElementIterator no longer return only null values for the getStartOfRun() and getAverageRuntime() methods.
• Setting a concurrency throttle of 1 on the root node in a runtime configuration file no longer causes the throttle to be ignored.
• Message triggers awaiting assignment of a message no longer wait unnecessarily when messages are available to be assigned.
• Variable substitution is now performed correctly in a File Exist Trigger postscript when the trigger times out.
• Flux now correctly uses the “%2F” notation in URLs for specifying files from the root directory on FTP, SFTP, and FTPS hosts.
• Using triggers in a flow chart no longer causes duplicate values to be added to the HAS_RUN_PRESCRIPT variable.
• UNC hosts that use the “Port” property no longer cause an IllegalArgumentException when upgrading from Flux 7.7 (or lower) to Flux 7.8 (or higher).
• Flow charts with a “Sort Order” file criteria property no longer cause an error when upgrading from Flux 7.7 (or lower) to Flux 7.8 (or higher).
• The API method Action.removeFlow() now correctly returns “true” if the flow is successfully removed.
• Claiming or executing a Business Process Trigger no longer causes an error for engines running with the PostgreSQL database.
• Business process forms with a “choice” element no longer lose all of the ID and description values in the choice element when reading the flow chart from XML.
• POJO variables that do not implement the Serializable interface are now written correctly to the database as long as they follow the standard Flux rules for persistent variables.
• Engines running with a DB2 database no longer encounter an error during job selection when there is no room in the concurrency throttle.
• Users with Message Administrator permissions can now successfully publish messages on secured engines with no addition permissions required.
• The forecast page of the Operations Console now correctly shows all forecasted firings in the given range, rather than only the next firing for a given flow chart.
• The Desktop Designer no longer displays an IllegalArgumentException intermittently.

Flux offices are closed today in observance of a US holiday.

A maintenance release for Flux 7.9 is now available. This release includes 1 improvement and 7 bug fixes.

Flux customers can download Flux 7.9.11 from the Portal website at:

my.fluxcorp.com

Evaluation versions of the latest Flux release are also available at:

www.fluxcorp.com/download

Flux 7.9.11 is a drop-in replacement for Flux 7.9.x.
Improvements

• Supplying a null value for the “relativeTo” parameter in the EngineHelper.applyTimeExpression() API method now produces a more easily understandable error message, in place of the previous error “flux.EngineException: scheduledItem is null”.

Bug Fixes

• The Search Logs page in the Operations Console now correctly displays results when a logger type is selected from the drop down menu.
• Flow charts now execute as expected when there are two or more engines in a cluster with the same SERVER_NAME configuration property.
• The number of threads in an engine no longer grows unbounded during flow chart execution.
• Engines configured to use Microsoft SQL Server no longer display an exception when attempting to fail over flow charts.
• Engines no longer attempt to set timeouts on the database connections they use, preventing the error “flux.EngineException: org.postgresql.util.PSQLException: Method org.postgresql.jdbc4.Jdbc4Statement.setQueryTimeout(int)” that could occur when running PostgreSQL.
• The engine now starts correctly when the RMI_PORT is set to a value other than the default.
• The engine now correctly displays an SQLException when it fails to persist an exception object to the database because the exception object contains references to custom classes that do not implement the Serializable interface.

A maintenance release for Flux 7.9 is now available. This release includes 2 improvements and 16 bug fixes.

Flux customers can download Flux 7.9.10 from the Portal website at:

my.fluxcorp.com

Evaluation versions of the latest Flux release are also available at:

www.fluxcorp.com/download

Flux 7.9.10 is a maintenance release. It is a drop-in replacement for Flux 7.9.x.
Improvements

• The Operations Console now regularly reloads the .fluxwebapp.properties file, rather than only loading it after a restart.
• Removed support for MySQL 4 and lower, which prevents the engine from inappropriately using the FLUX_FLOW.CONDITION column instead of the correct CONDIZION column when an older version of the JDBC driver is used.

Bug Fixes

• An engine no longer becomes deadlocked if database connectivity goes down for a significant amount of time while an action is executing.
• Running multiple engines in the same JVM will no longer cause some engines to stop executing jobs when caching is enabled.
• Clustered engines will no longer hang if a network outage prevents communication to other engines in the cluster.
• Engines using the Postgres or SQL Server databases no longer throw the exception “The index 15 is out of range” during job selection.
• In Internet Explorer 8, scheduling a flow chart from the repository or flow charts page now displays the dialog box correctly.
• The Repository and Designer menu links in the Operations Console are now displayed correctly when the user does not have Administrator permissions.
• Custom audit trail events are now visible in the Operations Console’s audit trail page.
• The Web-based Designer now correctly loads flow charts that contain a file criteria with an empty include pattern.
• Accessing the result set from a Database Stored Procedure Action no longer produces the error: “The result set is closed”.
• Database triggers and actions no longer throw an exception when using an external database connection from a data source.
• POJO variables no longer cause a CloneNotSupportedException when attempting to commit to the database.
• The REST Action no longer throws the exception “javax.ws.rs.WebApplicationException: java.lang.IllegalArgumentException: No MessageBodyWriter for body part of type ‘java.lang.String’ and media type ‘text/plain’” when executing.
• The Operations Console now displays an engine’s status as “Unknown” instead of “Unreachable” if there is a problem looking up the status but the engine is otherwise available.
• Host names are now correctly displayed for unreachable engines in the Operations Console.
• Entering a single quote in a message filter on the audit trail page of the Operations Console now works correctly for engines running the Oracle database.
• Exceptions are no longer displayed on standard error when a database connection cannot be obtained.

Workarounds

The following list describes workarounds for known bugs in Flux, for cases where the problematic behavior can be corrected or avoided by applying existing functionality. If a bug has been corrected in a maintenance release, that will be noted along with its workaround below.

• If a timer trigger throws an exception because its time expression cannot be satisfied, it will always invoke the default error handler even if an error flow is available to follow.

  The workaround for this problem is to add the following line to the timer trigger’s prescript:

  Factory.makeInstance().makeEngineHelper().applyTimeExpression(new Date(), action.getTimeExpression(), action.getBusinessInterval(), TimeZone.getDefault());

  If the timer trigger’s time expression has at least one future firing date, this call will return silently. Otherwise, it will throw an exception and force the timer trigger to follow its error flow instead of using the error handler.

A maintenance release for Flux 7.9 is now available and includes twenty-two bug fixes and four improvements.

Flux 7.9.9 download are available here:

my.fluxcorp.com (for Flux customers)

The latest version of Flux is available from:

www.fluxcorp.com/download (for Flux evaluators)

The following maintenance changes were made in this Flux release, version 7.9.9.

Improvements:

1. For long-running actions, database connections are now queried every three minutes to prevent database timeouts while the actions are running.
2. Added the ability to specify an admin username for the cluster (in addition to the existing cluster password) to specify which user account the web app should use when looking up information for the connected engines.
3. If a flow chart is interrupted, it will now continue to run on the same engine when it resumes execution (previously, it could run on any available engine in the cluster).
4. Reduced the volume of log messages and audit trail events that occurred when the Operation Console detected engines.

Bug Fixes:

1. High memory usages could occur when executing actions and triggers that had timeouts set a long while in the future. Fixed.
2. If the Operations Console attempted to connect to an engine using external security (such as LDAP), a NotSerializableException could occur that would make the engine inacessible from the Console. Fixed.
3. If an action was monitoring for a signal, executed in a loop in a flow chart, and the action name was mixed-case, the signal would not be cleared when the flow chart looped back to continue execution. Fixed.
4. If three engines were running in a cluster and one of the engines was disposed using the Java API, the flow charts running on that engine were not failed over correctly. Fixed.
5. In Web-based Designer, FileAction containing an UNCHost in a FileCriteria would fail to save to the repository causing an EngineException for null port value. Fixed.
6. If variable substitution was used in the include of a File Action with a file criteria host and target, the substitution would only be evaluated on the first execution of a looping flow context. Fixed.
7. On FlowChartElement objects returned from a FlowChartElementIterator, calls to getStartOfRun() or getAverageRuntime() would always return null. Fixed.
8. After searching the audit trail on a remote engine, the engine could become deadlocked and stop executing flow charts. Fixed.
9. When running a process on an agent with multiple Flux engines in a clustered environment, the agent’s work would be wrongly failed over and re-executed. Fixed.
10. If a Process Action ran on an agent, it would log multiple Entering Action and Executing Action events to the audit trail. Fixed.
11. In the Mail Action, when the setToAddresses(), setCcAddresses(), or setBccAddresses() methods were called, email address arguments were appended to the existing list of email addresses instead of replacing them. Fixed.
12. If a flow chart contained a split, the branching flow contexts could be prematurely failed over if they ran longer than the failover time window. Fixed.
13. When a large number of flow charts were executing on an engine, triggers would not time out quickly even when their timeout expression was satisfied. Fixed.
14. When executing a process on an agent that runs longer than three minutes, the process would be failed over and re-executed. Fixed.
15. Flow charts would not execute on any engine in a cluster if one engine’s concurrency throttles were at capacity. Fixed.
16. When using a Database Condition Action or Trigger, if flow of control returned to that action or trigger, or the condition was not met on the first execution, the action would fail with a connection closed exception. Fixed.
17. The engine column of the flow charts page in the Operations Console always displayed the name of the engine that the user was connected to in the Operations Console. Fixed.
18. Editing a FlowChart containing a RestAction from the repository in Operations Console and saving it back to the repository would fail with EngineException. Fixed.
19. A ClassCastException could occur when calling Engine.getFlowChartElements() on a secure engine. Fixed.
20. If a secured engine was configured to use a JAAS login module, attempting to run a Flow Chart Action would fail with the message “user must be logged in to call this method”. Fixed.
21. On a secured engine, users could not publish messages from the Operations Console, even when Message Administrator permissions were enabled. Fixed.
22. Process Actions that run on agents are displayed multiple times in run reports. Fixed.

A maintenance release for Flux 7.9 is now available and includes seventeen bug fixes and six improvements.

Flux 7.9.8 download are available here:

my.fluxcorp.com (for Flux customers)

The latest version of Flux is available from:

www.fluxcorp.com/download (for Flux evaluators)

The following maintenance changes were made in this Flux release, version 7.9.8.

Improvements:

1. Enhanced Mail Action to support Gmail as an SMTP provider.

2. Added support for multipart form processing, HTTP headers, and SSL certificates in Rest Action.

3. The Database Stored Procedure Action now reports the out parameters from its result to the audit trail, allowing them to be viewed in the Operations Console.

4. Improved the error message for attempting to create a runtime configuration node when a node already exists with the same name.

5. The log message: “No matches were found for the path” was logged at the WARNING level instead of the correct FINE level. Fixed.

6. Added support for persisting variables larger than 10,000 bytes when using the in-memory database.

Bug Fixes:

1. When running secured engines in a cluster, only flow charts running on the same engine the user was connected to could be interrupted. Fixed.

2. If the File Delete Action could not find a specified include, it would issue a bad delete command to the server, incorrectly causing the action to fail. Fixed.

3. Restarting engines in a cluster could prevent flow charts from executing. Fixed.

4. An out of memory exception would occur do to a memory leak in long running Flux environments. Fixed.

5. When using a runtime configuration file, if a concurrency throttle of 1 was specified on the root of the runtime configuration tree, the throttle would be ignored. Fixed.

6. The SQLException: “Closed Statement” occurred when using the Database Stored Procedure Action to call a stored procedure that returned output parameters. Fixed.

7. If date substitution was used in the base directory property of a file criteria on a file action, the date substitution would fail to reset after the first loop in a looping flow context. Fixed.

8. Message Triggers awaiting assignment of a message would sometimes have to wait up to the system delay before message assignment occurred. Fixed.

9. If an agent was restarted while connected to an engine, the engine would display the agent’s status as “STOPPED” and would not submit any new work to that agent. Fixed.

10. The Flux Desktop Designer failed to open a flow chart that included a timer trigger with an invalid time expression. Fixed.

11. A SQLException could be thrown when using the Oracle 11gR2 JDBC driver. Fixed.

12. When two RMI engines are created in the same JVM which uses the same RMI port, the second engine would fail to create with port in use exception. Fixed.

13. When an engine’s flow chart selection became stuck, no other engines in the cluster would continue selection. Fixed.

14. If a file action that used a UNC host encountered an exception, the default error handler was not invoked. Fixed.

15. The Flux Web Application start-up script was not working properly in a Linux/Unix environment. Fixed.

16. Factory.lookupCluster() APIs threw a NullPointerException when invoked. Fixed.

17. The forecast page of the Operations Console only displayed the next scheduled trigger date, and not all forecast firings for the specified time frame. Fixed.