Troubleshooting Java applications on OpenShift

By Frédéric Giloux August 16, 2017August 14, 2017

What is it about?

OpenShift has seen a lot of traction with the release of its third version based on Kubernetes a couple of years ago. More and more companies after a thorough evaluation of OpenShift Container Platform (OCP) have built an on-premise or in the cloud PaaS. With the next step, they have started to run their applications on OCP. One of the important aspects of running applications in production is the capacity of quickly restoring services to the normal service level after an incident followed by the identification and the resolution of the underlying problem. In this respect, I want to present in this blog a few approaches for troubleshooting Java applications running on OpenShift. Similar approaches can be taken with other languages.

Debugging applications during development phase can be done thanks to features like:

• Debug mode for resolving issues during startup.
• Port forwarding for connecting an IDE like JBDS to an application running in a remote container and debugging it with breakpoints and object inspection.

This has been presented in blogs like here and here.

In this blog, on the contrary, I want to focus on troubleshooting applications in production and to cover things like capturing heap and thread dumps, resource consumption per thread. These are techniques that have more than once been helpful in the past for resolving deadlocks, memory leaks or performance degradation due to excessive garbage collection for instance.

Let’s get into the heart of the matter!