HP jmeter Software for -UX guide

with the number of bytes that would be reclaimed. It is up to you to check if the presented objects are still needed by the application, and how the critical references should be removed.

Using Heuristics to Locate Possible Hot Spots

79

Image 79

HP jmeter Software for -UX manual Using Heuristics to Locate Possible Hot Spots

Contents

HPjmeter 4.2 Users Guide Document Notice Table of Contents Monitoring Applications Analyzing Garbage Collection Data 116 Unfinalized Objects Live Objects Bytes 199 187192 197 Additional HPjmeter Documents Related Information About This DocumentIntended Audience Typographic Conventions HP Encourages Your Comments Publishing History Introducing HPjmeter Features JVM Agent ConceptsNode Agent Console Requirements Completing Installation of HPjmeterPlatform Support and System Requirements Agent Requirements Preparing to run Java Completing the installationFile Locations Attaching to the JVM Agent of a Running Application On Java Example Usage Showing Version Information JVM Agent OptionsSelecting Other JVM Agent Options Include=filter1filter2..., exclude=filter1filter2 To specify a file name Noalloc JVM Options Usage Examples Security AwarenessWorking with Firewalls Configuring User Access Getting Started Using HPjmeter to Monitor Applications To connect to a node agent via a SecureShell SSH tunnel Using HPjmeter to Monitor Applications Set Session Preferences When the Optional Port Box Is Not Used Getting Started Changing Session Preferences During a Session Using HPjmeter to Analyze Profiling DataView Monitoring Metrics During Your Open Session Configure your application Using HPjmeter to Analyze Garbage Collection Data Monitoring Demonstration Instructions Memory Leak Applications See also Heap Usage Notification Thread Deadlock Sample Thread Deadlock Example Display Monitoring Applications Setting Data Collection PreferencesControlling Data Collection and Display Verifying HP-UX Daemon is Running Managing Node AgentsManaging Node Agents On HP-UX Running Node Agent as a Daemon Stopping Node Agents Node Agent Access RestrictionsSaving Monitoring Metrics Information Running Multiple Node Agents Identifying Unexpected CPU Usage by Method Diagnosing Errors When Monitoring Running ApplicationsNaming Monitoring Data Files Saving Data from the Console Viewing the Application Load Checking for Application Paging ProblemsChecking for Long Garbage Collection Pauses Reviewing the Percentage of Time Spent in Garbage Collection Identifying Excessive Calls to System.gc Monitoring Applications Confirming Java Memory Leaks Checking for Proper Heap SizingDetermining the Severity of a Memory Leak Identifying Multiple Short-lived Threads Identifying Excessive Object AllocationIdentifying the Site of Excessive Object Allocation Identifying Abnormal Thread Termination Identifying Deadlocked Threads Identifying Excessive Lock ContentionIdentifying Excessive Thread Creation Example Metric Method Compilation Count Identifying Excessive Method Compilation Identifying Too Many Classes Loaded Using the JMX Viewer JMX Summary Tab Understanding the JMX Summary View JMX Memory Tab Appearance of the Summary JMX Viewer When First Opened JMX Viewer with Summary Memory Tab Selected JMX Viewer with Summary Threads Tab Open JMX Threads Tab JMX Runtime Tab Changing Mbean Values and Monitoring the ResultJMX Notifications Tab MBean Filter Using the Functions in the JMX Server View MBean Attribute Tab 10 MBean Operations Tab Open for Display MBean Operations Tab MBean Information Tab MBean Notifications Tab 12 MBean Information Tab Open for Display Profiling Overview Profiling Applications Tracing Tuning PerformanceSampling Preparing a Benchmark Profiling with -Xeprof Collecting Profile Data Profiling with Zero Preparation If timeon=sig... has not been specified, the default is Following options are useful Profiling with -agentlibhprof Supported -agentlibhprofoptions Generate more complete call graphs Naming Profile Data Files Inclusive Method CPU Times Profiling with -Xeprof Profiling with -agentlibhprof Looking at the Data from the Top Down Approaches to Analyzing Performance DataConsiderations in Interpreting the Data Looking at the Data from the Bottom Up CPU Versus Clock Time Locating Summary Information for Saved Data SetsInclusive Versus Exclusive Time Time Units Estimate menus Adjusting Scope Comparing Profiling Data Files Scaling Comparison Data Key to Thread States Reported by Reading Profiling Histograms Interpreting the Histogram Presentation Interpreting Call Graph Data Using Call Graph Trees Example of Node Color Display Auto-Expanding the Call Tree Using Sub-TreesOptions for Manipulating the Call Tree Display Tree Pruning Searching the Trees Using Heuristics to Locate Possible Hot Spots Using Heuristics to Locate Possible Hot Spots Obtaining Garbage Collection Data Analyzing Garbage Collection DataData Collection with -Xverbosegc Java ... -Xverbosegchelp 13, Old generation, in bytes. These spaces are reported Collecting Glance Data for Viewing in HPjmeter Collecting GC Data with Zero Preparation Related Topics Data Collection with -Xloggc Naming GC Data Files Using -XX+PrintGCDetails and -XX+PrintHeapAtGC Together Comparing Garbage Collection Data Files Key to Garbage Collection Types Recognized by HPjmeter Basic Garbage Collection Concepts Common Garbage Collection Types Recognized by HPjmeter Understanding the Summary Presentation of GC Data Summary Panel Showing Garbage Collection Statistics Understanding the System Details Captured with GC Data Basic Garbage Collection Concepts Analyzing Garbage Collection Data Using the Main Window Functions Using the ConsoleStarting the Console Main Console Window Data Representation Node Agent Icons and Their MeaningJVM Agent Time Slice Entries Open and Cached SessionsSaving Data Console Menu Choices Console Tool Bar Buttons Remember Main Window Location Monitor Menu Console Guide Location and First Screen Console Guide Status Bar Setting Monitoring Session PreferencesSpecifying Metrics to Collect for Monitoring Enabling Monitoring Alerts Editing Filters Specifying Filters for MonitoringAdding Filters Viewing Monitoring Data in HPjmeter Using AlertsSpecifying Filter Sets Appearance of Alert Threshold in GC Duration Visualizer Appearance of Alert Notices in the Main Console Pane Using the Alert Controller Alert Controller Window Viewing a Log of the Alert History Editing E-mail Notification Attributes Abnormal Thread Termination Alert Responding to Alerts Identifying Abnormal Thread Termination Thread Histogram Excessive Compilation Alert Expected Out Of Memory Error AlertGC Duration Notification Java Collection Leak Locations Alert Heap Usage Notification 13 Java Collection Leak Locations Visualizer Array Leak Locations Alert 14 Array Leak Locations Visualizer Process CPU Usage Alert System CPU Usage Alert Unfinalized Queue Growth Thread Deadlock Alert Profiling Data Viewer Using Visualizer Functions Locating Information About a JVM and its Environment Using Monitoring Displays Java Method HotSpots Monitor Code and/or CPU Activity Menu Details Thrown Exceptions with Stack Traces Thrown Exceptions Monitoring Metric Thrown Exceptions with Stack Traces Heap Monitor Monitor Memory and/or Heap Activity Menu Monitoring Metric Garbage Collections Garbage Collections GC Duration Percentage of Time Spent in Garbage Collection Unfinalized Objects Allocated Object Statistics by Class Allocating Method Statistics Current Live Heap Objects Name of class to which object belongs Thread Histogram Monitor Threads and/or Locks Menu Already acquired by another thread 14 Monitoring Metric Lock Contention Lock Contention Method Compilation Count Monitor JVM and/or System Activity Menu Loaded Classes Method Compilation Frequency 17 Monitoring Metric Loaded Classes Related Topics Percent % CPU Utilization Viewing Profiling or GC Data in HPjmeter Using Profile Displays Using Visualizer Functions Inline Candidates Exceptions Thrown Memory Leaks Profile Code and/or CPU Activity Menu ChoicesMethod Call Count Exclusive Method Clock Times Exclusive Method Times CPUCall Graph Tree with Call Count Inclusive Method Clock Times Call Graph Tree with CPUCall Graph Tree with Clock Time Inclusive Method CPU Times Starvation by Method Average Exclusive Method Clock TimesAverage Inclusive Method CPU Times Average Inclusive Method Clock Times Exclusive Class CPU Times Profile Memory and/or Heap ActivityStarvation Ratio Methods with Loops Live Objects Count Objects Created by MethodCreated Objects Count Created Objects Bytes Reference Sub-Trees by Size Reference Graph Tree Using Profile Displays For string and char arrays, see String and char arrays 21 Long String Pop-up Box 23 Class Loaders Visualizer Class Loaders 24 Class Types Pop-up Window Guidelines Threads Histogram Profile by ThreadsResidual Objects Count Residual Objects Bytes Thread Groups Histogram Profile by Locks All Lock Claims by Method Lock Delay Method ExclusiveLock Delay Call Graph Tree Contested Lock Claims by Method Exclusive Method Lock Delay / Clock Time Lock Delay Method InclusiveLock Contention Ratio by Method Average Exclusive Method Lock Delay Exceptions Thrown Using Heuristic Metrics from the Estimate MenuExclusive Class Lock Delay Inline Candidates Heap Usage After GC Using Specialized Garbage Collection Displays 27 GC Metric Heap Usage After GC Related Topics Duration Stop the World 28 GC Metric Duration Related Topics Cumulative Allocation 29 GC Metric Cumulative Allocation Related Topics Creation Rate 30 GC Metric Creation Rate Related Topics Allocation Site Statistics 31 GC Metric Combined Sets of Allocation Site Statistics 32 GC Metric Separated Sets of Allocation Site Statistics Using Visualizer Functions Filter Field Descriptions Using Visualizer Functions User-Defined X-Y Axes 37 User-Defined Comparisons of GC Metrics Related Topics Multiple User-Defined 38 Multiple User-Defined Metrics Glance Data 40 GC Metric Glance Data 41 Glance Data Metric Configuration Box Run Queue Average Are private Glance System Call Data 42 GC Metric Glance System Call Data Common Tool Bar Buttons Using Visualizer Tool Bars Tool Bar Buttons for Manipulating Graphical Data Tool Bar Buttons for Manipulating Tabular DataTool Bar Buttons for Manipulating Garbage Collection Data Mark an Item for Search Special Button or Other Gadget Functions Find a Search Pattern 44 Locating an Item Changing Time Interval in GC Data Visualizers Pause or Resume Graphical Time-based Scrolling 46 The View Menu Selections in the GC Viewer Using Visualizer Tool Bars Changing Time Interval in Monitoring Visualizers 50 Time Interval Use in a Monitoring Visualizer Change Color Selection for Histogram Display Thread Histogram Profile by Threads Performance Overhead on Running Applications Understanding How HPjmeter Works Data Sampling Considerations Using Confidence Interval to Indicate Sample ValidityConsole Overhead How Memory Leak Detection Works Related Topics Page Identifying Version Numbers TroubleshootingInstallation Documentation and Support See also To Take Advantage of Dynamic Attach Console JVM agent If you see this JVM agent error on startup Zero Preparation Profiling Node agent Unexpected Behavior in Visualizers Quick References Connecting to the HPjmeter Node Agent Start and stop verbose GC data 0.14 0.02 Glossary Active mode Jvmpi SSL Index Symbols Alert Notification Editor, 107 alerts Restore Roots Alert Notification Editor, 108 count Collecting with zero preparation, 83 GC event Glance system call data, 174 graphical scrolling JVM Duration Stop the World, 158 enable, 99 User-Defined X-Y Axes for GC analysis, 167 viewing, 29 Program termination, 47 overhead Scale Special, 72 scaling data Xeprofoptions Xbootclasspath