A Glance At The Java Performance Toolbox.pdf
0 likes60 views
The document discusses various tools in the Java performance toolbox for building optimized container images, monitoring and profiling Java applications. It describes tools like Jlink, Dockerfile, JFR and JMX for creating minimal custom runtimes, capturing events, tuning JVM flags and correlating data across monitoring systems. Specific techniques covered include using Jlink to build minimal custom Java runtimes, capturing native memory usage with Jcmd, enabling JMX access and collecting statistics with Jstat and Jmap. The document also provides links to resources for further reading.
A Glance At The Java Performance Toolbox.pdf
- 1. A Glance at The Java Performance Toolbox
- 2. Java Champion Alumni, Certified Architect Senior Developer Advocate at Oracle Passionate about solving complex scenarios involving Java and Kubernetes. ammbra1508 ammbra1508.mastondon.social
- 3. Run on Clouds by Walking The Oceans of Tools
- 4. Developer Questions Using Java Applications in Containers • Which tools help you create container images with only what is needed at runtime? • How to run the JDK tools in containers and proxy their output? • Why and how to capture performance relevant JVM and application events? • Which tools help you fine tune the JVM flags and keep application overhead at minimum? • How to correlate data from JVM monitoring tools with the one from tools like Prometheus, Grafana?
- 5. Tools to build an OCI Compliant Image • Docker build • Buildpacks • Jib • kaniko • buildah • s2i • And many more J
- 6. The Good Old Dockerfile
- 7. The Good Old Dockerfile Use a small base image
- 8. The Good Old Dockerfile Use a small base image Create images with common layers
- 9. The Good Old Dockerfile Use a small base image Install only what is strictly needed Create images with common layers
- 10. Creating Custom Java Runtimes for Container Images
- 11. Jlink JEP 282 “Link time is an opportunity to do whole-world optimizations that are otherwise difficult at compile time or costly at run-time” jlink: Command Line Tool to Link Modules • Modules in a jimage file • Generate runtime images
- 12. Jlink Packaging Legacy JDK image JDK > 9 generated image bin jre lib bin conf …….. Modular runtime image jlink
- 13. Maintain The Good Old Dockerfile with JLink The runtime stage build • From a JDK base image • Create your own custom JRE with jlink
- 14. Maintain The Good Old Dockerfile with JLink The runtime stage build • From a JDK base image • Create your own custom JRE with jlink The application stage build • From an OS base image • Copy the custom JRE from the runtime stage build • Copy the artifacts needed by your application • Run the application
- 15. The Runtime Stage Build
- 16. The Runtime Stage Build jdeps --ignore-missing-deps -q -recursive --multi-release 19 --print-module-deps --class-path 'target/libs/*’ target/spring-todo-app.jar
- 17. The Runtime Stage Build Strips debug information from the output image
- 18. The Runtime Stage Build Exclude man pages and header files
- 19. The Runtime Stage Build Enable compression of resources: 0: No compression 1: Constant string sharing 2: ZIP
- 20. The Application Stage Build Copy the custom JRE
- 21. The Application Stage Build Copy the artifacts needed by your application
- 22. The Application Stage Build Run the application
- 23. Fine Tuning JVM Flags
- 24. Tracking Native Memory With Jcmd Add jdk.jcmd module
- 25. Tracking Native Memory With Jcmd WARNING: Overusing jcmd to send diagnostic commands can affect the performance of the VM. Add -XX:NativeMemoryTracking={off|summary|detail} to JDK_JAVA_OPTIONS • kubectl set env deployment/app JDK_JAVA_OPTIONS=“-XX:NativeMemoryTracking=summary” • Please consider that having NMT enabled can add application performance overhead.
- 26. Tracking Native Memory With Jcmd WARNING: Overusing jcmd to send diagnostic commands can affect the performance of the VM. Add -XX:NativeMemoryTracking={off|summary|detail} to JDK_JAVA_OPTIONS Create a native memory baseline kubectl exec pod_name –it --/bin/bash -c "jcmd llvmid VM.native_memory baseline"
- 27. Tracking Native Memory With Jcmd WARNING: Overusing jcmd to send diagnostic commands can affect the performance of the VM. Add -XX:NativeMemoryTracking={off|summary|detail} to JDK_JAVA_OPTIONS Create a native memory baseline kubectl exec pod_name –it --/bin/bash -c "jcmd llvmid VM.native_memory baseline" Create a native memory summary diff kubectl exec pod_name –it --/bin/bash -c "jcmd llvmid VM.native_memory summary.diff"
- 28. Enabling JMX Monitoring and Remote Access to a Remote Host
- 29. Get Statistics About Running Java Processes
- 30. Obtain Statistics with Jstat and Jmap Obtain statistics about garbage collectors by running jstat kubectl exec pod_name –it --/bin/bash -c "jstat -gcutil llvmid 250 10" Print heap summaries kubectl exec pod_name –it --/bin/bash -c "jmap -histo llvmid"
- 31. Capture Performance Relevant JVM And Application Events
- 32. Add JFR To Your Container Image
- 33. Enable JFR Enable JFR in JDK_JAVA_OPTIONS kubectl set env deployment/app JDK_JAVA_OPTIONS=“--XX:StartFlightRecording:..” Start a Time Fixed Recording with profile settings -XX:StartFlightRecording=delay=10s,duration=10m,name=Profiling,settings=profile Start a Time Fixed Recording with default settings -XX:StartFlightRecording=delay=10s,duration=10m,name=Default,settings=default Start a Continous Recording -XX:StartFlightRecording:filename=/recordings/recording.jfr,settings=profile
- 34. Where to Use JFR Streaming Streaming selected metrics to your monitoring service Send an average, min, max, etc of a metric Expose JFR data through other management APIs
- 35. Where NOT to Use JFR Streaming “To consume the data today, a user must start a recording, stop it, dump the contents to disk and then parse the recording file. This works well for application profiling, where typically at least a minute of data is being recorded at a time, but not for monitoring purposes.“ source: JEP 349
- 36. Make your settings by extending jdk.jfr.SettingControl Recording Custom JFR Events
- 37. Make your settings by extending jdk.jfr.SettingControl Create your custom event by extending jdk.jfr.Event Recording Custom JFR Events
- 38. Recording Custom JFR Events Make your settings by extending jdk.jfr.SettingControl Create your custom event by extending jdk.jfr.Event Register the custom settings in your custom event via @jdk.jfr.SettingsDefinition
- 39. Recording Custom Events in Spring Boot (example) Capture the custom JFR events in a dedicated filter and register it
- 40. Recording Custom Events in Spring Boot (example) Listen the custom JFR events and record their duration
- 42. Useful Links • Article https://www.javaadvent.com/2022/12/a-sneak-peek-at-the-java-performance-toolbox.html • Tools https://docs.oracle.com/en/java/javase/19/docs/specs/man/index.html • Troubleshoot Memory Leaks: https://docs.oracle.com/en/java/javase/19/troubleshoot/troubleshooting-memory- leaks.html • JFR Runtime guide : https://docs.oracle.com/javacomponents/jmc-5-4/jfr-runtime-guide/about.htm#JFRUH170 • Innovative JFR use by @gunnarmorling: https://www.morling.dev/blog/finding-java-thread-leaks-with-jdk-flight- recorder-and-bit-of-sql/ • Great intro by @BillyKorando https://www.youtube.com/watch?v=K1ApBZGiT-Y • jlink https://dev.java/learn/jlink/ • Erik Gahlin’s health report app • @gunnarmorling article on custom JDK Flight Recorder Events