1. Release Notes for GoldenGate Stream Analytics
  2. GoldenGate Stream Analytics Release Notes
  3. Known Issues and Workarounds
  4. Release 19.1.0.0.0 - Known Issues

1.1.2.11 Release 19.1.0.0.0 - Known Issues

This section lists the known issues and workarounds for Oracle Stream Analytics release 19.1.0.0.0.

1.1.2.11.1 Batch Delay Increases When HA is Enabled

Bug Number: 27754912

Impacted Platforms: All

While publishing a pipeline, if you enable HA you can see that the batch scheduling is delayed. This is a known issue.

Workaround

No workaround available

1.1.2.11.2 Values of Drop-down List not seen when the Live Output Table has a single row

Bug Number: 27473663

Impacted Platforms: All

Full values within the drop-down list are not seen when the live output table has a single row. This is a known issue.

Workaround

Pull the horizontal splitter up so that more of the table is visible.

1.1.2.11.3 SDO_GEOMETRY Not Supported in Coherence Cache

Issue

Bug Number: 27480822

Impacted Platforms: All

In coherence cache, data type conversion to SDO_GEOMETRY is not supported. This is a limitation.

Workaround

No workaround available

1.1.2.11.4 Pattern Cannot Detect Events Correctly

Issue

Bug Number: 25132943

Impacted Platforms: All

When you have configured the batch duration as 1 second for underlying Spark Streaming, the pattern match clause cannot detect events sequence correctly happening within an interval of less than or equal to 1 second. This is a limitation.

Workaround

No workaround available

1.1.2.11.5 Default Character Limit for Text Fields

Bug Number: 27701535

Impacted Platforms: All

There is a limitation on the size of text fields in pipelines: events with text fields exceeding 1024 characters will not be processed.

1.1.2.11.6 Limited Support for Predictive Models

The import and scoring of Predictive Models might contain undocumented limitations and you should use them as is.

1.1.2.11.7 Possible loss of events in HA scenario after Executor Crash

Bug Number: 29697956

Impacted Platforms: All

Events that arrive when the executor is up and running but the CQL engine is still not bootstrapped, will result in failed tasks. By default each task is attempted 3 times by Spark and subsequent to that it reports the task as failed. This is a known issue.

Workaround

No workaround available.

1.1.2.11.8 Correlation pattern does not support number data type

Bug Number: 29801715

Impacted Platforms: All

Pattern Correlation does not support number data type. So when choosing the parameter of type Number for correlation pattern, it throws exception in the spark log file. The workaround is to use data type either of int, float, double, bigint or apply toDouble function for the number data type and use that in correlation pattern.

Workaround

The workaround is to use data type either of int, float, double, bigint or apply toDouble function for the number data type and use that in correlation pattern.

1.1.2.11.9 Rest Submission Server for Spark disabled

Oracle Stream Analytics when configured on Spark standalone cluster, uses its Rest Submission server to deploy the pipeline. Spark 2.4.0 and higher versions have Rest Submission servers disabled by default.

Ensure or enable the Spark Rest Submission server of the standalone cluster. To enable the Spark Rest Submission server, set the property spark.master.rest.enabled to True, when starting the cluster. One way is to add it in Spark default config file, ${SPARK_HOME}/conf/spark-defaults.conf, before starting the cluster. See Spark documentationfor details.

1.1.2.11.10 Loading Class Fails when using JMS Streams

Bug: 29022250

Impacted Platforms: All

When using JMS stream with durable subscription, user is possible to run into the error: FAILED TO LOAD CLASS WEBLOGIC.JMS.COMMON.INVALIDCLIENTIDEXCEPTION. This is because the JMS durable subscription is exclusive and an existing pipeline has created the subscription.

Workaround

Change the stream's setting to use another client ID or subscription name, or kill the unwanted pipeline that runs JMS subscriber.

1.1.2.11.11 No Output Event For Some Special Rule Configuration

Bug: 29015898

Impacted Platforms: All

In rule stage, modifying a value by a rule, already set by another rule, leads to no events to be emitted in live output table.

Workaround

Create a separate rule stage to handle this case.

1.1.2.11.12 Disorder Of Output Events With Pipeline Contains Obr For Executor Failover

Bug: 29112584

Impacted Platforms: All

Ordering of events across different partitions is not guaranteed. However events are ordered within same partition.

Workaround

No workaround available.

1.1.2.11.13 Query Group ->Table Stage does not work on certain Window Types

Bug: 29795496

Impacted Platforms: All

Query group->Table stage does not work on the 'constant value', 'row without slide' and 'now' window types. Adding a group based on above window types on the query group/table stage, renders the pipeline broken, with the following UI error: Failed to start pipeline. : Uncaught exception. : Syntax Error. Invalid usage of Reserved CQL Keyword: "on".

Workaround

Create another stage from query group and configure the 'Group' and 'Constant Value' window in the new stage.