Welcome!

@DevOpsSummit Authors: Jason Bloomberg, Pat Romanski, Yeshim Deniz, Elizabeth White, Liz McMillan

Related Topics: @DevOpsSummit, Microservices Expo, Linux Containers, Containers Expo Blog, Server Monitoring

@DevOpsSummit: Blog Feed Post

Configuration with #Jenkins | @DevOpsSummit #DevOps #APM #Monitoring

Configuration settings that are enabled during the application development are very useful

When Configuration Settings from Development Wreak Havoc in Production
By Asad Ali

As applications are promoted from the development environment to the CI or the QA environment and then into the production environment, it is very common for the configuration settings to be changed as the code is promoted. For example, the settings for the database connection pools are typically lower in development environment than the QA/Load Testing environment. The primary reason for the existence of the configuration setting differences is to enhance application performance. However, occasionally there are instances where the application code is mistakenly promoted into production without changing these settings. In such cases, such promotion of code can cause performance havoc in the production environment. This blog describes one such scenario.

During their proof of concept, a prospect requested we identify and resolve an issue that they were observing in the production environment. They had just promoted a newer version of their critical application to production and soon after the promotion of the code they stared to see significant increase in the response time when the end users tried to login to their web application. To diagnose the issue the prospect injected our agents into their production JVMs that were exhibiting the issue to the end user. With Dynatrace agents injected, we observed both high response time for login and that most of the time for the login request was spent in class loading.

Additionally, breakdown of the response time showed that most of the time for login web request was spent in synchronization (92%).

The response time breakdown of the login request clearly shows that 92% of the time is spent in synchronization.

The response time breakdown of the login request clearly shows that 92% of the time is spent in synchronization.

With the response time breakdown showing the highest amount of time in synchronization we took a thread dump on the JVM where the login request was being processed to get insight into thread locking issue. The thread dump showed 67 threads that were blocked in the JVM.

67 Threads were blocked on the JVM.

67 Threads were blocked on the JVM.

Locking Hotspots view shows the breakdown of the blocked threads

Locking Hotspots view shows the breakdown of the blocked threads

Further analysis of the content of the thread dump showed that most of the blocked threads were waiting for the resource (CompoundClassLoader) held by one running thread (Thread Id 5570678).

Thread Id 5570678 owned the monitor on which the other threads were blocked.

Thread Id 5570678 owned the monitor on which the other threads were blocked.

CompoundClassLoader is held by Thread 5570678.

CompoundClassLoader is held by Thread 5570678.

All blocked threads are waiting on CompoundClassLoader.

All blocked threads are waiting on CompoundClassLoader.

The thread stack trace for the running thread showed that it is trying to load the class from the file system. Examining the full trace of the thread dump shows that the threads on this JVM are creating a new facelet every time a web request is received.

getFacelet being called every time a request is processed by the thread.

getFacelet being called every time a request is processed by the thread.

While it is very normal for a JVM to spend some time to load classes when the JVM is initially started, it is NOT normal (or good for performance) for the class loading to continue even after the JVM has warmed up and most of the classes are already loaded.

The thread stack showed call to getFacelet(java.util.URL) method in the com.sun.facelets.impl.DefaultFaceletFactory class. Review of the source of this class showed that the method tries to load the class if the method needsToBeRefreshed() returns true.

Code snippet of com.sun.facelets.impl.DefaultFaceletFactory class.

Code snippet of com.sun.facelets.impl.DefaultFaceletFactory class.

And finally the code for needsToBeRefreshed() clearly shows that it returns true  if the refreshPeriod is set to 0.

needsToBeRefreshed() method shows that if the refresh period is set to 0, the class is refreshed every time.

needsToBeRefreshed() method shows that if the refresh period is set to 0, the class is refreshed every time.

The Facelets are capable of precompiling if you set javax.faces.FACELETS_REFRESH_PERIOD to -1. However, once set to -1, the JSF never re-compile/re-parses the Facelets files and holds the entire SAX-compiled/parsed XML tree.

Insert11

In the development process the REFRESH_PERIOD is typically set to 0 because it allows the developers to keep editing the Facelets file without the need to restart the server. What happened at this prospect is that the application code was promoted into production with the REFRESH_PERIOD is to 0 value and hence every time the user tried to login, the Facelets were forced to recompile and that in turn resulted in high response time.

Conclusion
Configuration settings that are enabled during the application development are very useful as they reduce the number of times the application server has to be started to test code changes. However, as this example shows, it is very important to disable development-level settings as the code moves out of the development environment because they can cause performance havoc in the production environment. One of the best practices to eliminate occurrences of such scenario is to make the configuration changes a part of the continuous integration tools like Jenkins.

The post When configuration settings from development wreak havoc in production appeared first on about:performance.

More Stories By APM Blog

APM: It’s all about application performance, scalability, and architecture: best practices, lifecycle and DevOps, mobile and web, enterprise, user experience

@DevOpsSummit Stories
"Our strategy is to focus on the hyperscale providers - AWS, Azure, and Google. Over the last year we saw that a lot of developers need to learn how to do their job in the cloud and we see this DevOps movement that we are catering to with our content," stated Alessandro Fasan, Head of Global Sales at Cloud Academy, in this SYS-CON.tv interview at 20th Cloud Expo, held June 6-8, 2017, at the Javits Center in New York City, NY.
Andi Mann, Chief Technology Advocate at Splunk, is an accomplished digital business executive with extensive global expertise as a strategist, technologist, innovator, marketer, and communicator. For over 30 years across five continents, he has built success with Fortune 500 corporations, vendors, governments, and as a leading research analyst and consultant.
Hardware virtualization and cloud computing allowed us to increase resource utilization and increase our flexibility to respond to business demand. Docker Containers are the next quantum leap - Are they?! Databases always represented an additional set of challenges unique to running workloads requiring a maximum of I/O, network, CPU resources combined with data locality.
The current age of digital transformation means that IT organizations must adapt their toolset to cover all digital experiences, beyond just the end users’. Today’s businesses can no longer focus solely on the digital interactions they manage with employees or customers; they must now contend with non-traditional factors. Whether it's the power of brand to make or break a company, the need to monitor across all locations 24/7, or the ability to proactively resolve issues, companies must adapt to the new world.
In his keynote at 18th Cloud Expo, Andrew Keys, Co-Founder of ConsenSys Enterprise, provided an overview of the evolution of the Internet and the Database and the future of their combination – the Blockchain. Andrew Keys is Co-Founder of ConsenSys Enterprise. He comes to ConsenSys Enterprise with capital markets, technology and entrepreneurial experience. Previously, he worked for UBS investment bank in equities analysis. Later, he was responsible for the creation and distribution of life settlement products to hedge funds and investment banks. After, he co-founded a revenue cycle management company where he learned about Bitcoin and eventually Ethereum.