JOLT 8.1/TUX 8.1 - Unnecessary Synchronized HashTable Usage in JoltDefinition Class Cause Transaction Throughput Bottleneck Under Load (Doc ID 777156.1)

Last updated on SEPTEMBER 14, 2023

Applies to:

Goal

DESCRIPTION:
Unnecessary Synchronized HashTable Usage in JoltDefinition Class Cause Transaction Throughput Bottleneck Under Load.

The problem was found by customer's transaction performance tests using WLS 6.1 SP5 with JoltSession Pool to Tux 7.1.
It was observed that the transaction throughput was bottlenecked by Jolt clients at about 2k TPM (transaction per minutes).
The Java profiling tool and consecutive JVM thread dumps reveal that most of the threads look like the following, and
are waiting to acquire the monitor for a Hashtable.

This is worked in internal defect CR123917, Bug 7933916 for the complete description. 

CONFIGURATION:
Tuxedo 8.1
Jolt 8.1
RedHat ES 3.0 32-bit

Solution

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