Welcome!

Java Authors: Esmeralda Swartz, Pieter Van Heck, Elizabeth White, Pat Romanski, Gary Kaiser

Related Topics: Java, SOA & WOA, .NET, AJAX & REA, Cloud Expo, DevOps Journal

Java: Article

DevOps Role in Troubleshooting JVM Issues

Application developers can deploy their applications with success in a staging or QA environment

The JVM issues vary from Java OutOfMemory Error to JVM Crash. Application developers might be not completely equipped to determine the root cause of the issue, hence DevOps can play a vital role in narrowing down the issue and connecting the right people/team to rectify the problem.

Application developers can deploy their applications with success in a staging or QA environment and broadcast that the application is working fine. However, the staging or QA environment is not completely identical with production at - least with respect to load. Due to huge load, the application can behave differently. In order to prevent an occurrence of these kinds of scenarios, it is a good practice to engage a DevOps team [with right JVM skills] to proactively analyze the running application. Proactive engagement of a DevOps team will be beneficial for the organization and for the development team as well.

As a proactive engagement, the DevOps team needs to conduct an analysis on the collected memory snapshot.The memory snapshot can be gathered at regular intervals time and multiple samples of the same, so that a good picture of what is happening internally is known.

By employing appropriate JVM diagnostics tools, the DevOps team can provide recommendations to application developers with respect to memory consumption, memory reserved or memory wasted in Java collections.The topmost consumer of Java Heap memory, classes loaded, object promotion rate,object liveness, object references and many more details can be shared with application developers proactively. These in-depth details will provide application developers with insight into the application behavior and help identify any potential issues.

Apart from conducting memory-related analysis, the DevOps team can look into the performance of the application and determine the root cause of the degraded performance (if any). With regard to this, employing the right tool and conducting analysis on Garbage Collection activities will reveal whether the poor performance is due to GC or outside of the GC. The Memory Analyzer tool, which is available as a plugin to Eclipse, can be employed to conduct memory-related analysis and gather enough information. In the same pattern, the plugin " Garbage Collection Memory Visualizer" can be employed to conduct analysis on verbose GC data.

As a case study I would like to share one of the scenarios that we encountered in our organization and how we isolated the issue before engaging the right team to fix it. One of our customers upgraded to the latest application release from a previous release and noticed a significant performance degradation in the throughput. As a first step, we confirmed the customer's claim by reproducing the reported issue in our environment. Having a local reproduction environment provided us with various opportunities to try out various things when narrowing down the issue. After the first level of analysis, we confirmed the area or component that was contributing to the observed behavior and a command-line parameter was recommended to workaround the issue.The provided recommendation was tested in our local reproduction environment with success and this gave us the confidence that the  issue can be overcome with the current recommendation.

On further analysis, we engaged the appropriate tool to determine the percentage of time consumed and captured the snapshots with the latest release and the previous release where the issue was not noticed.This level of analysis revealed the exact area where more time was spent. Once these snapshots were shared with the development team, the development team's task was much easier as the area of code was already narrowed down for fixing.This way, we isolated the issue and helped the customer overcome the problem.

More Stories By ChandraShekar Dattatreya

Chandra Shekara Dattatreya is a DevOps guy working in a Fortune-500 company and has 10+ years' of experience debugging JVM-related issues.In the course of debugging, he has encountered multiple scenarios from various customers and provided solutions to all of them with success. In his current role, he is engaged in identifying and resolving JVM-related issues for an e-commerce company.