Analyze the performance of call stacks using call graphs

Learn what call graphs are and how they help you analyze and troubleshoot the performance of call stacks.

Analyze and troubleshoot the performance of call stacks using call graphs. Call graphs help you locate, diagnose, and debug performance issues with your service.

For instructions on how to access a call graph, see Access a call graph.

Prerequisite

To use call graphs, you must first コールグラフ プロファイリングへのデータの取得 .

Visualize method durations with the summary bar

The summary bar above the call graph is a visual representation of the method durations in the call stack. You can use the summary bar to understand and compare the proportion of time your methods contribute to the call stack.

The top 5 methods by duration are displayed as individual purple segments. The remaining methods are grouped in a single gray segment. The width of the segments in the summary bar is proportionate to the percentage of time the methods take up in the call stack.

Hover over a segment in the summary bar or the text in the legend below the summary bar to:

  • Display a panel with the full method name, percentage of time the method takes up in the call stack, and the duration of the method.

  • Highlight the corresponding method(s) in the call graph.

  • Highlight the corresponding rows in the call graph table.

Explore the methods in the call stack

The call graph view displays a hierarchical visualization of the methods in a call stack. Select a method to display a menu with options to copy the name of the method or expand/collapse the list of the child methods.

Monitor the resource usage of methods in the call stack

Use the Time (ms) column to view the duration of each method in the call stack.

Use the Percent % column to understand which methods in the call stack are the most time intensive. This value describes the percentage of time in the call stack used by each method. For example, if the total duration of the call stack is 100 ms and the duration of a method is 50 ms, the percentage shown for the method would be 50%.

The Exit Calls / Threads column lists the exit call(s) for each method. Select the exit call link to navigate to the next call graph.

Download the JSON object for the call graph

Select Download call graph to download a JSON object that contains the data on the call graph view.

Exclude packages from call graphs

Note: You must have the admin role to exclude packages, add custom packages, and remove custom packages.
To keep call graphs readable and focused on application code, certain frameworks and library packages (such as core Java, common third-party libraries, and app server classes) are excluded by default. You can use the APM call graphs page to view the list of packages excluded by default, add custom packages to exclude, and remove custom packages.

To navigate to the page, use the Splunk Observability Cloud main menu to select Settings > APM call graphs. The APM call graphs table lists the packages excluded from call graphs. On this page, you can:

  • Use the search bar to search for a package.

  • Select the All Languages drop-down menu to filter the table by a language.

  • Select Add package to exclude to add a custom package to exclude from call graphs.

  • Select the trash icon in the Actions column to remove a custom package. Default packages can't be removed.

By default, excluded packages aren't displayed in call graphs. To view the excluded packages in any call graph, select Show them in the call graph view. Select Hide them to filter out the excluded packages from the call graph.