Bug Fixes (4.3)

Description

When sending an HTTP/2 RequestCycle UDR to another workflow using Workflow Bridge with the Force Serialization check box selected in the Workflow Bridge profile used, some of the fields in the UDR will become blank.

Resolution

This problem has now been fixed and the values for all fields will be retained when using Force Serialization.

Description

When an {{HttpRequestCycleUdr}} was passed to a different workflow and back, it was impossible to send an HTTP response back due to a missing HTTP context.

Resolution

An HTTP context cache has been implemented per HTTP2 server agent, which can be configured with a maximum size and a maximum time a context will be kept in the cache.

Description

If the ‘Stop Immediately’ option is not enabled in the SAP RFC Processor, it will drop 1 UDR per thread when it receives a ‘Stop Workflow’ command while processing the UDRs. Additionally, an Interrupted Exception will be thrown and printed out in the debug log.

Resolution

This is fixed now and the SAP RFC Processor works as intended. Also, the Interrupted Exception is removed from logging to the debug log.

Description

When executing the SAP RFC function on UDRs, the JSON name of the UDR field will be used as field names in the JCORecords. Certain fields will contain the string ‘null’ for its JSON name which is an invalid field name in the JCORecord.

Resolution

This is fixed by adding a check for the JSON name containing the string ‘null’ and handling it accordingly before setting the value into the JCORecord object.

Description

Encrypted exports that contain external reference values or archive runtime data will fail to import.

Resolution

The encryption has been corrected.

Description

The option, File System Profile, was missing from the list of agent profiles where External References can be defined.

Resolution

The option, File System Profile, has been added to the list of agent profiles where external references can be defined.

Description

When multiple Picos are located in the same host and point to the same log directory it will show all files. This means that when filtering with ‘Platform’, files named EC will show up.

Resolution

A File Name filter has been added. When a user filters on ‘Platform’ a file name filter for ‘Platform’ is also created. The File Name filter can be deleted as needed and the Pico filter will continue to work as it did previously.

Description

Workflow properties are not displayed in the correct order. They are not grouped properly by node name, section ID, and title.

Resolution

This problem has now been fixed, and the workflow properties are displayed in alphabetical order by node name, section ID, and title.

Description

When using APL cancelBatch or hintEndBatch functions in Unit Test Runner, the error messages given are unclear.

Resolution

The error messages have been updated.

Description

When you drag agents to new positions in the workflow editor in the Legacy Desktop, NullPointerExceptions will be logged in the desktop.log.

Resolution

This problem has now been fixed and agents can be moved around without exceptions being logged.

Description

The caching logic in the {{validateJwt()}} function currently uses the JWKS URL and the full token as the cache key, rather than the key ID. As a result, the JWKS must be fetched from the server for every request, even when the key ID remains the same. This can lead to issues, as servers typically impose limits on the number of allowed requests.

Resolution

The cache key has been updated to use the JWKS URL and key ID, with the JWKS content stored as the cache value. This reduces server connections, ensuring that the same key ID triggers only a single request to the server.

Description

Workflow routes saved in Desktop with customized routes are rendered incorrectly.

Resolution

This problem has been fixed by adding a start and end point to the routes when the are opened in Desktop for the full route to be drawn properly.

Description

Using a primary index places is not best practice since it heavily burdens the Couchbase index server and affects the performance.

Resolution

This has now been fixed with a secondary index using only relevant fields instead of the primary index.

Description

The *TMO Full Collection Strategy* in SFTP and Disk agents exhibit differences between MediationZone 8 and 9. The default value for the *Keep* field in MediationZone 8 will appear empty, while in MediationZone 9 it is set to -1. The validation is skipped in the Agent Configuration dialog if the field is configured in the workflow table, while in MediationZone 9 validation is done for both. The workflow properties in MediationZone 9 do not display the *Keep* and *Destination* fields.

Resolution

We have minimized the disparity between MediationZone 8 and 9 by. Updating the default value for the *Keep* field to be set to -1. Validation will now be performed in the Agent Configuration dialog and the workflow table when the configuration exists in the workflow table. The *Keep* and *Destination* fields will be visible in workflow properties.

Description

Invalid HTTP POST requests only generate generic stacktraces in the log files.

Resolution

The exceptions are now caught and written to the log with added information about the request, source, and destination host/ports.

Description

The Ultra XML Schema validation is broken for xs:all elements that contains child elements that are of complex type. The standard limitation that maxOccurs may only be a maximum of 1 is validated recursively for all child elements of xs:all. Furthermore, if the value “unbounded” has been used, an exception is thrown instead of a good validation message.

Resolution

The validation has now been corrected.

Description

During certain circumstances, the ECDs stay in pending or disabling state for a very long time.

Resolution

The ECD state handling has now been improved to mitigate this issue.

Description

If a single batch workflow is stuck on Interworkflow-related operation due to I/O issues, it will block operations on all other IWFs on the affected EC, effectively preventing all workflows relying on IWFs from starting.

Resolution

The synchronization of IWF data was modified so that a single operation on an IWF only holds a lock on that particular IWF instead of the whole list.

Description

Audit Profile's Add/Edit Table field list is not shown due to the window being too small.

Resolution

Corrected the window size to take up more screen space.

Description

When sessions are written to file storage and then read back, this can change the object graph. This has the effect that if the session contains multiple references to the same object, the effect of updates to that object may vary unpredictably depending on whether the session has been swapped down to storage during processing.

Resolution

The serialization strategy for the aggregation session storage has been changed in a backward-compatible way. Any writes to storage will now preserve the object graph.

Description

When using the real-time Disk agent and the SAP RFC agent within a workflow, any running loops created in the following Analysis agent will stop running once the Disk agent has finished collecting the files. A stop signal is sent from the Disk agent to the RFC agent which stops all currently processing loops.

Resolution

A timer task was added, which constantly checks if the RFC Agent is still in processing state. It will block the stop signal until the RFC agent completes processing the UDRs received from the disk agent.

Description

When UDR is generated for the SAP RFC Function, certain data elements have fields which do not have a name. The UDR generation process will rename these fields to “_” which is not a valid identifier in Java 9 and above as it is considered a keyword.

Resolution

Add handling to skip UDR generation for fields that do not have a name.

Description

The default behaviour of APL function tableCreate and other database functions having the default parameter for disableCommit as True, this may lead to deadlock.

Resolution

Fixed by executing a commit on every execution, and removing the disableCommit parameter for configuration.

Description

When creating an ECD and selecting a PVC, the ECD will get stuck in ContainerCreating and never become ready when you try to start it.

Resolution

This problem has now been fixed and the created ECD pod will start and become ready as expected.

Description

When an ECD pod mounts a persistent volume, the volume directory is mounted with root permissions instead of mzadmin user. This issue occurs when the directory’s permissions depend on the fsGroupPolicy setting in the CSIDriver.

Resolution

This issue has been resolved. The ECD pod is now configured with the PodSecurityContext, and applied the correct FSGroup value.

Description

There is no warning message when an untrusted/self-signed certificate is used in the Desktop Launcher.

Resolution

There is now a warning icon with a hovering message visible when untrusted/self-signed certificate is used in Desktop Launcher.

Description

An error message saying "Failed to write to the [file: Strea|file: Stre]m closed" is displayed after exporting SQL statement successfully from the SQL Agent.

Resolution

This problem has now been fixed and the error message will not be displayed upon successful export.

Description

When the RFC Processor Agent is running an RFC Function with a long execution time, the agent’s worker thread will remain in a suspended state until the function completes. During this time, any attempts to stop the workflow through user input or error aborts will be ineffective until the function execution is completed, and a response is returned.

Resolution

The pending RFC Function cannot be terminated from the client side. If an abort signal is received while the function is still processing, notification events will be logged in the System Log. Additionally, an optional log event will be generated to alert the user if the function execution exceeds the user-defined time threshold.

Description

Each time an RFC Processor Agent is consuming UDRs, the UDRs are being added to a queue to be processed and removed once it is completed. The queue has a maximum capacity configured by the user in the agent, and when this limit is reached, additional UDRs cannot be added, leading to the “RFCErrorUDR” error. When the queue remains at full capacity, the error is being generated and duplicated multiple times.

Resolution

This issue is solved by fixing the logic in RFC Processor Agent to prevent duplicated “RFCErrorUDR” error entries.

Description

Workflow groups with certain naming patterns are incorrectly removed, and the operator will try to re-recreate the workflow group indefinitely.

Resolution

This problem has now been fixed and workflow groups will not be removed incorrectly.

Description

An OutOfMemoryError will occur when attempting to decode a JSON string with the incorrect datatype using the JsonDecodeUdr APL function.

Resolution

This is resolved by fixing the decoding logic to generate the appropriate error message with the relevant information.

Description

When decoding an empty json string with the JsonDecodeUDR function, an incorrect exception is thrown.

Resolution

This problem has now been fixed and decoding an empty json string does not generate an exception.

Description

The way that invalid dates are handled when decoding JSON to UDRs in APL or through the Ultra decoder has changed since MZ8, causing backward compatibility issues. In MZ8 invalid dates were decoded as null, but now this results in an exception.

Resolution

Empty fields are now treated as null and do not cause an error. Any other string that is not recognized as a valid date will also be decoded as a null value. Null values will cause a warning about the invalid date string.

Description

The workflow editor does not read the default values for checkboxes and cannot reset to the default values in Desktop Online.

Resolution

This problem has now been fixed, and the checkboxes will pick up the default values. It is also possible to reset to the default values.

Description

When decoding JSON to a UDR, you will get decoding errors if you try to decode a null JSON value into a UDR field that is Java primitive. This happens either using jsonDecodeUdr APL function or the JSON decoder in the Decoder agent.

Resolution

Now, this will be decoded to the default value for the corresponding type. Ie {{false}} for boolean fields and 0 for numeric fields.

Description

When changing and resetting logging settings, the logs entries are written to the wrong file.

Resolution

This problem has now been fixed and the log entries are written to correct file even after changing and resetting logging settings.

Description

When sending large blocks of data over RCP, the buffers used for marshaling data grow, but are never cleaned up until the writing thread stops. This can cause a limited memory leak.

Resolution

These buffers are now kept on WeakReferences so that the garbage collector can clean them up and they will no longer cause memory issues for the process.

Description

The PrometheusUDR UDR type does not support the MZTagged format. This means that if you try to encode and decode this UDR type to for example send it to another workflow, the data will be lost.

Resolution

The issue has now been fixed and the PrometheusUDR supports MZTagged.

Description

The workflow properties popup throws an exception when new Kafka agents are not licensed.

Resolution

This issue has been resolved so that when new Kafka agents are not included in the license, the new Kafka profile will be unavailable in the Execution Settings.

Description

When having Prometheus setup and configuring autoscaling for an ECD, the autocomplete drop-down list does not show any metrics when you start to type and you have to enter the metrics manually.

Resolution

This problem has now been fixed and the drop-down list will display matching metrics when you start typing.

Description

: The mzsh system --local command does not work when credentials are provided.

Resolution

: A fix has been provided that allows mzsh system command to be executed with credentials.

Description

When using the Diameter Stack agent in workflows in a Kubernetes environment, the pod IP is sent out as Host_IP_Address in Capabilities-Exchange-Request/Answer which causes a mismatch between sender IP and Host_IP_Address.

Resolution

A new field has been added to Diameter Stack to enter the Load balancer Service name that is used for external communication. Diameter then adds the external IP of the Service to the list Host_IP_Address in Capabilities-Exchange-Request/Answer.