Filter the flame graph

Note: Profiler samples describes several Go language samples stored in a GitHub repository. You can download, and then run these samples in the Cloud Shell, on Google Cloud, or on a local Linux box. The images on this page are for a Service named docdemo-service. For information on how to generate your own data with the same configuration, see the sample hotapp.

Cloud Profiler lets you add filters to control how the information in the selected profiles is displayed. For example, you can add a filter to hide particular frames or call stacks. Adding and removing filters doesn't change your set of selected profiles.

Each filter is specified by a predefined FILTER-OPTION that is paired with a user-defined VALUE:

[FILTER-OPTION] : [VALUE]

Each filter that you add is displayed on the filter bar. In this example, there is one filter that displays Metric : Bytes:

Profiler automatically creates a filter with a FILTER-OPTION of Metric and a VALUE based on your selected profile type. You can change VALUE for some profile types. You cannot remove this filter.

To add a filter, use one of the following approaches:

• Click Filters , select an option from the list, and then enter the value.
• Click the gray text Add profile data filter in the filter bar, and then enter the filter option and value.
• For the Focus, Show from frame, and Show stacks filter options, you can also place your pointer on the frame, and then select the option from the frame tooltip.

To remove a filter, click Close on the filter.

When preparing the data to display, Profiler searches for matches between a frame and a filter. When a match occurs Profiler uses the FILTER-OPTION to determine what action to execute. A frame matches the filter when the frame's function name or the filename of the function's source contains VALUE. A case-sensitive comparison is performed. For example, if the filter is Hide frames : oo, then frames with functions named foo, foo1, and busyloop match are hidden from the flame graph.

Metric

To set the profile type aggregation mode, use the Metric filter. For example, if you select a Heap profile type, you have the choice between visualizing the data in terms of Bytes and Objects.

The choices available for the Metric filter depend on the programming language and the selected Profile type:

• For CPU time profiles, the only choice is CPU time.

• For Heap profiles, the choices are:

  • Bytes
  • Objects

• For Allocated Heap profiles, the choices are:

  • Total alloc bytes
  • Total alloc objects

• For Wall time profiles, the choices are:

  • Count
  • Wall time

• For Threads profiles, the only choice is Goroutine.

• For Contention profiles, the choices are:

  • Delay
  • Contentions

For more information about types of profiling metrics, see Profiling concepts.

For example, the following screenshot shows the CPU consumption of a program:

Here, you can see that the busyloop routine calls foo1 and foo2, both of which call various other routines. You can add filters to further restrict the graph only to the data of interest.

Focus

When you use the Focus filter, you select a single function and the flame graph displays the code paths that flow into, and out of, that specific function.

For details how to focus the graph and interpret the results, see Focusing the flame graph.

Show stacks

To display all call stacks that contain a frame that matches the filter value and to hide all other call stacks, use the Show stacks filter. The graph shows the callers and callees of the function, that is, everything that calls the matching function, and everything it calls.

This filter performs a case sensitive substring test. A match occurs if the frame function contains the filter value.

To restrict the CPU-usage graph from the previous example to show only the call stacks that involve the function foo1, set a Show stacks filter for foo1:

Hide stacks

Hide all call stacks that contain a frame that matches the filter value. This filter is often useful when you want to hide uninteresting stacks. For example, with Java applications, adding a Hide stacks: unsafe.park filter is common.

This filter performs a case sensitive substring test. A match occurs if the frame function contains the filter value.

Show from frame

To display all call stacks, starting from the frame that matches the filter value and to hide all other call stacks, use the Show from frame filter. The resulting graph shows the call stacks from the named function down. This filter is useful if your function is called from many places, and you want to see the total consumption attributable to it.

For example, to only show calls originating from the baz function, set a Show from frame filter for baz:

This filter performs a case sensitive substring test. A match occurs if the frame function contains the filter value.

Hide frames

To hide from view all frames that match the filter value, use the Hide frames filter. The graph shows the callers of the function, and any callees of the function are collected together. This filter is useful for removing irrelevant frames from the graph.

For example, to hide the frames for both foo1 and foo2, set a Hide frames filter for foo. Both foo1 and foo2 match, so both are removed from the graph. Because both of them call the bar and baz routines, the data for each of those functions is aggregated together.

This filter performs a case sensitive substring test. A match occurs if the frame function contains the filter value.

Highlight

To highlight all frames whose function names match the filter value, use the Highlight filter. The function remains in normal color mode, but the call sequences are colored in more subdued tones.

For example, here is a graph with no highlighting:

Here's the same graph with highlighting requested for the baz function:

This filter performs a case sensitive substring test. A match occurs if the frame function contains the filter value.

Color mode

By default, the frame color corresponds, where possible, to the function's package. If package information is unavailable, as with Node.js, the names of the source files are used to color the function blocks. With the default setting, a change in a call stack frame color means a transition from one package to another. The default option corresponds to the Color mode filter with a value of Name.

To color the frames in the flame graph by the consumption of a function and its children, add a Color mode filter with the value of Total. If a function is called through multiple call stacks, then the color is determined by the metric consumption for all call stacks. For example, the main and busyloop are colored red. These two frames consume the most CPU time. The frames labeled foo2 and baz are a deep orange, while the frame labeled foo1 is a lighter orange. The frames for bar and load are the lightest. This flame graph illustrates that foo2 consumes more CPU time than foo1 but less than busyloop:

To color the frames in the flame graph by the function's metric consumption but exclude the metric consumption of its children, add a Color mode filter with the value of Self. For example, this filter shows that the baz function consumes more CPU time than any other function:

What's next

• Focus the flame graph on a function.
• View historical trends.
• Compare profiles.
• Download profile data.