Hogging And Stuck Threads With ADS Long-polling Occurring in an ADF Application (Doc ID 2930544.1)

Last updated on MARCH 06, 2023

Applies to:

Symptoms

As part of a migration from Weblogic 12.1.3 to Weblogic 12.2.1.4, an issue with hogging and stuck threads (HT and ST) was encountered, which is causing concern for application performance and stability. The issue occurs when users login to the ADF application via a browser; this immediately causes one hogging thread( HT) and one stuck thread (ST); so there is 1 HT and 1 ST per session.
The issue was traced to the use of long-polling connections in the Active Data Service (ADS) for instant messaging functionality.
To confirm the cause, ADS was tested with a "polling" transport and did not observe the issue, which confirms that long-polling is the root cause of the issue. The Stuck thread (ST) by setting the keep-alive-interval to 9 minutes, just under the 10-minute stuck thread timeout. However, the hogging thread (HT) threads are still present, and there is a concern about their impact on application performance.
All ADF patches and Weblogic patches until October 2022 were installed. Also, a test with the January 2023 Weblogic patch was done, but the issue persists. The following is the setting being used:
  
 

Cause

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