weblogic.jndi.requestTimeout Attribute Improve EJB Performance when a Member of a Cluster is Hanging (Doc ID 1424255.1)

Last updated on OCTOBER 11, 2023

Applies to:

Symptoms

An EJB modules from an application is deployed to a cluster. And normally, the application has a high performance and responds very quickly. However, when a member server in the cluster hangs, the performance of the whole application goes bad and clients need to wait very long for server response.
The application would look up the EJB instance on the whole cluster, and the PROVIDER_URL include URLs of all the member servers.(e.g. "t3://localhost:7003,localhost:7005,localhost:7007,localhost:7009")

Changes

This issue can be reproduced by the following steps:

1.Create domain with a cluster of 4 managed servers.
2.Develop a stateless EJB and define a remote method which would print a message.
3.Develop a servlet simulate server hang.
4.Deploy the ejb and the servlet to entire cluster.
5.Develop a ejb client. Below is an example code:
6.Run ejb client.
7.Run the hang servlet and make a member server hang.
8.The client becomes slow. And seeing from the log, you can observe a significant increase on time spent("Spend time" value)

9.If you stop the hanging server, the client will become fast again.

Cause

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