Movatterモバイル変換


[0]ホーム

URL:


YourKit
Products
Docs
Forum
Company
Fully featured low overhead profiler for Java EE and Java SE platforms.
Easy-to-use performance and memory .NET profiler for Windows, Linux and macOS.
Secure and easy profiling in cloud, containers and clustered environments.
Performance monitoring and profiling of Jenkins, Bamboo, TeamCity, Gradle, Maven, Ant and JUnit.
YourKit Java Profiler
Overview & Features
Benefits
Download
Buy & Upgrade
What's New
Docs

Changes in version 2025.3

JVM support

  • feature Java 24 is supported.

Alpine Linux support

  • caveat Minimal supported version ofmusl C standard library is now 1.2.4.

macOS support

  • improvement Version and dashes were removed from the macOS application name. The app is now always calledYourKit Java Profiler, following macOS naming guidelines.
  • fix Profiler may not start if the installation directory contains spaces.

Networking

  • feature The profiler can now use theAWS CLI to connect to a remote AWS instance and profile your remote Java app by specifying the AWS instance ID.

    Profiling via AWS SSM
  • improvement When profiling a remote application over SSH, the profiler now detects if SSH port forwarding is disabled on the remote server or if the user has no home directory. This helps identify configuration issues early and prevents confusion when the user cannot attach the profiler to their application.

User interface

  • improvement Navigation drawers now use space more efficiently and do not take up more horizontal space than necessary.
  • improvement Capturing a snapshot no longer stops profiling. To stop profiling, use theStop profiling action in the toolbar.
  • improvement Horizontal toolbars inTelemetry charts andThread states were replaced with vertical ones to save vertical space.

CPU profiling

  • fix Fixed a deadlock that could occur in the OpenJ9 JVM when using virtual threads.

Thread profiling

  • feature Since deadlock analysis is part of thread profiling, we have moved theDeadlocks tab inside theThreads tab.Deadlocks is now a tool withinThreads. This change simplifies UI and improves user experience.
  • feature The profiler agent now storesJava-level deadlock information in snapshots, enabling post-mortem analysis. In previous versions, deadlocks were only available in live profiling sessions.
  • feature The profiler now displays the time spent in a deadlock for deadlocked threads.

    Time spent in lock
  • feature An optionFocus on selected threadFocus on selected thread has been added. It allows you toestimate CPU usage of only the selected thread.

    CPU estimation for selected thread
  • feature A newFrozen threads view has been added to theThreads tab. The profiler identifies threads that do not change their stacks or consume CPU time and reports them as frozen. These frozen threads may be in a deadlock state or simply long-sleeping or waiting.
  • feature You can now ignore "false positive" frozen threads by hiding threads with similar stacks.

    Ignore false positive frozen threads
  • feature New toolbar actionszoom inZoom in andzoom outZoom out have been added allowing to choose an arbitrary scale for the thread states timeline.
  • feature A newTelemetry section in theThreads tab displays various telemetry charts related to thread activity.

    Thread telemetry
  • feature Deadlock profiling is now available in IBM Java. Previously, this feature was only available in HotSpot-based JVMs.
  • feature The profiler agent now provides a new option,deadlocks_check_period, to control the frequency of deadlock detection.
  • improvement The deadlock detector has been optimized - it runs faster, does not block the UI thread, and consumes fewer resources.

Memory profiling

  • fix Fixed an issue where the thread state was incorrectly loaded from an HPROF snapshot captured on Java 21.
  • fix DisplayingQuick info for certain arrays could cause a crash.

Object allocation profiling

  • featureAllocation profiling has been moved from theMemory tab to a dedicatedAllocations tab. YourKit profiler now provides more tools for inspecting object allocations, available in live profiling sessions and when analyzing memory snapshots.

    Allocation profiling
  • feature The new action,Selected Objects Allocation, lets you open objects with recorded allocation details in a newAllocations tab for further analysis. This action is available in the popup menu in allMemory views.

    Selected objects allocations
  • caveat Instead of memory snapshot, performance snapshot iscaptured automatically on application exit, if object allocation profiling was running.
  • caveat The option to record object allocations with "estimated" stacks has been removed. It often produced confusing and hard-to-interpret results.
  • fix Fixed an issue where recorded allocation data was cleared when allocation profiling was stopped.
  • fix Fixed an issue in live mode where popup menus could close unexpectedly when theMemory tab was selected.
  • fix TheCallee list view no longer loses its selection after a refresh inlive profiling session.
  • fix Fixed issues inobject counting mode where the allocated object count was incorrect.

Exception profiling

We have completely redesigned the UI for exception profiling. It now resembles the CPU profiling UI and introduces powerful new analysis tools, including anException tree that allows grouping exceptions by the threads where they were created, and anException flame graph for visualizing exception occurrences.

  • feature TheException tree, with per-thread grouping, displays stack traces in a hierarchical "tree" format, showing where exceptions were thrown.

    Exception tree
  • feature TheException flame graph visualizes exception occurrences as a flame graph, highlighting where exceptions were thrown.

    Exception flame graph
  • feature A newTelemetry section in theExceptions tab displays telemetry chart of thrown exceptions.

    Exception telemetry
  • feature Exception profiling is available in IBM Java.
  • improvement Profiling overhead has been reduced when exception profiling is not running.

Monitor profiling

The user interface ofMonitors tab has been significantly redesigned, improved, and unified with other parts of the profiler. New tools have been introduced for analyzing waiting and blocking events:Wait tree, with thread-based grouping,Monitors flame graph, and tools for analyzing monitors owned by threads and those where waiting and blocking occurred.

  • feature TheMonitors wait tree, with per-thread grouping, displays stack traces in a hierarchical "tree" format, showing where threads were waiting or blocked.

    Monitors wait tree
  • feature We have introduced theMonitor flame graph, a new visualization tool that highlights where threads were waiting or blocked. It enables faster analysis of contention issues.

    Monitor flame graph
  • feature NewOwner thread view displays the number of times and the total duration a thread has been blocked while attempting to acquire a Java monitor.

    Monitor owner thread
  • feature NewMonitor class view shows how many times and for how long threads waited to acquire a specific monitor object.

    Monitor class
  • improvement Average wait time is now shown for threads.

Telemetry

  • feature New toolbar actionszoom inZoom in andzoom outZoom out have been added allowing to choose an arbitrary scale for the telemetry charts.
  • feature A tooltip displaying uptime and clock time is shown above the telemetry chart.

    Telemetry uptime

IDE integration

  • feature Eclipse 2025-03 is supported.
  • feature IntelliJ IDEA 2025.1 is supported.
  • feature NetBeans 25 is supported.

EE server support

  • feature Apache Tomcat 11 is supported.
  • feature Apache TomEE 10.0 is supported.
  • feature Open Liberty 25 is supported.
  • feature WebSphere Liberty 25 is supported.
  • feature WildFly 35 is supported.

Triggers

  • caveat Thetrigger actions deprecated in previous releases have been removed.

YourKit uses cookies and other tracking technologies to improve your browsing experience on our website, to show you personalized content, to analyze our website traffic, and to understand where our visitors are coming from.

By browsing our website, you consent to our use of cookies and other tracking technologies in accordance with thePrivacy Policy.


[8]ページ先頭

©2009-2025 Movatter.jp