Hunting issues in a Java app with 31G heap and 500G off-heap: an Apache Ignite case study
summary
A case study of diagnosing and fixing incidents in a Java-based distributed database (Platform V DataGrid / Apache Ignite style) that uses a 31 GB Java heap plus 500 GB off-heap. Covers how to find root causes of performance drops and node losses, and describes Java (JVM) and Linux configuration and tuning techniques to prevent and mitigate such incidents.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 3:12What is Apache Ignite
- 7:42Real error examples
- 11:36Server node structure
- 16:16Heap memory issues
- 20:37Process loss and dumps
- 27:32Disk and performance problems
- 33:25Long pauses and causes
- 37:06Garbage collection issues
- 41:07JVM pause analysis
- 44:18Prevention checklist
- 45:21Summary and recommendations
related talks
41mDistributed Java DBs Under the Hood: Main Components and Interactions Between Them
51mA Healthy diet for your Java application
49mBeyond the Data Grid: Fast Data Processing with Apache Ignite
01h 28mThe dark art of performance tuning or how to become a performance hero ...
40mDebugging JVM performance issues
46mTuning Ignite in Cloud Infrastructure
52mJava 9: The Quest for Very Large Heaps
33mDistributed Database DevOps Dilemmas? Kubernetes to the rescue!