Issue in EDQ OWS Realtime Screening with ActiveMQ (Doc ID 2139335.1)

Last updated on FEBRUARY 04, 2022

Applies to:

Symptoms

In the OWS real-time screening process, messages are not being cleared on the hub server message-queue after being passed from the filter server.  This only applies in limited circumstances, such as when messages are sent by the filter server while the web service process is not running on the hub server.  The queue size becomes greater than zero when the process starts on the hub and remains at this size, or grows if more requests are received, and is not cleared.  The effect of this is to cause this information to be re-screened each time the job is restarted.  This causes unnecessary review flags where Watchlist data changes for any of these records.

The content of the messages on the queue appears to be irrelevant as long as they are well-formed within the defined data types, hence the supplied process to generate simple messages to send across the queue is sufficient for recreating the issue.  A ‘real’ message format can be supplied, if required.

When messages are passed between the filter and hub servers when the receiving hub process is running and the initial queue size is zero, the behavior is correct: the queue size remains zero as new messages are received, processed, and removed automatically.  Note, however, that this is not an acceptable workaround as outputs are required for real-time reporting.

Modifying the real-time process to remove the group spigot connections and deleting the two Group And Merge processors restores the correct behavior regarding the queue.  That will cause messages to be cleared appropriately.  However, if the Group And Merge processors are merely disconnected but not deleted, the incorrect behavior persists.

Running the real time process with one line of data per input data feed, via the supplied reference data mappings, causes the correct behavior.

One exemplary scenario of this behavior when running a real-time process with full volume (~45k+ records) is with the Dow Jones input files, which causes the incorrect behavior. (You should be able to generate these input files by running the OWS Watchlist Management project to prepare sanctions records and pointing the DJW Entities SAN and DJW Individuals SAN snapshots at the produced .jmp files.)

Changes

 The issue appears to be present whether or not patch 21312966 is installed.  Even after attempting to rollback using OPatch and redeploy, the problem persists.

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.