Stack Trace Rules
Learn how to use stack trace rules to group incoming events based on matchers.
This feature is only applicable for error issues with debug information files, excluding source maps. Other categories of issues (such as performance issues) do not support this feature.
A subset of stack trace rules, when defined, will also be applied to incoming profiles in order to improve the frames classification (in-app vs system).
More specifically, only the rules with any of these matchers and actions (below) will be applied to profiles.
Matchers allowed: stack.abs_path
stack.module
stack.function
stack.package
Actions allowed: +app
-app
If you use stack traces for grouping, the stack trace rules (previously known as grouping enhancements) influence the data that's fed into the grouping algorithm. These rules can be configured on a per-project basis in [Project] > Settings > Issue Grouping > Stack Trace Rules.
Each line is a single rule; one or multiple match expressions are followed by one or multiple actions to be executed when all expressions match. All rules are executed from top to bottom on all frames in the stack trace.
The syntax for stack trace rules is similar to:
matcher-name:expression other-matcher:expression ... action1 action2 ...
If you want to negate the match, prefix the expression with an exclamation mark (!
). If a line is prefixed with a hash (#
), it's ignored and treated as a comment.
The following is a practical example of how this looks:
# mark all functions in the std namespace to be outside the app
family:native stack.function:std::* -app
# mark all code in node modules not to be in app
stack.abs_path:**/node_modules/** -app
# remove all generated javascript code from all grouping
stack.abs_path:**/generated/**.js -group
Multiple matchers can be defined in a line. The following matchers are available:
Matches on the general platform family, which currently includes javascript
, native
and other
. Comma separate rules to apply them to multiple platforms.
family:javascript,native stack.abs_path:**/generated/** -group
alias: path
This matcher is case insensitive with Unix glob behavior on a path in a stack trace. The path separators are normalized to /
. As a special rule, if the filename is relative, it still matches on **/
.
# match on all files under `project` with a `.c` extension
stack.abs_path:**/project/**.c +app
# matches on vendor/foo without sub folders
stack.abs_path:**/vendor/foo/*.c -app
# matches on `foo.gen.c` as well as `foo/bar.gen.c`.
stack.abs_path:**/*.gen.c -group
alias: module
Module is similar to path
but matches on the module
. This is not used for Native but it is used for JavaScript, Python, and similar platforms. Matches are case-sensitive, and normal globbing is available. Note that modules are not packages, which can be confusing for Native environments.
alias: function
Matches on a function in a stack trace, and is case-sensitive with normal globbing.
stack.function:myproject_* +app
stack.function:malloc -group
alias: package
Matches on a package in a stack trace. The package is the container that contains a function or module. This is a .jar
, a .dylib
or similar. The same matching rules as for path
apply. For example, this is typically an absolute path.
stack.package:**/libcurl.dylib -group
Matches on the current state of the in-app flag of a stack trace frame. yes
means the frame is in-app, no
means it's not.
Matches on a built-in or a user-defined frame category. See variable actions on how to set a category.
category:telemetry -group
Frame categorization is heavily used by our newest grouping algorithm. You can match on a variety of categories, including:
system
- detected system librariesstd
- detected standard librariesui
- UI frameworksdriver
- graphics drivers, and so ontelemetry
- crash reporting and analytics frameworks
See our source code for a full list of built-in categories.
If information about the surrounding frames is necessary to apply a rule to a frame, use caller/callee matching syntax. For example:
# Ignore in-app frames if they are called by telemetry
[ category:telemetry ] | app:yes -group
# Ignore system frames if they call app frames
category:system | [ app:yes ] -group
There are two types of actions: flag and variables setting.
A flag identifies the action to be taken if all matchers match, and it uses these prefixes:
+
sets the flag-
unsets the flag^
applies to frames above the matching frame (toward the crash)v
applies to frames below the matching frame (away from the crash)
As an example, -group ^-group
removes the matching frame and all frames above it from the grouping.
app
: marks or unmarks a frame in-appgroup
: adds or removes a frame from grouping
A limited set of variables can be defined (variable=value
):
max-frames
: Sets the total number of frames to be considered for grouping. The default is0
, which means "all frames". If set to3
, only the top three frames are considered.
stack.abs_path:**/node_modules/** -group
stack.abs_path:**/app/utils/requestError.jsx -group
stack.abs_path:**src/getsentry/src/getsentry/** +app
family:native max-frames=3
stack.function:fetchSavedSearches v-group
stack.abs_path:**/app/views/**.jsx stack.function:fetchData ^-group
family:native stack.function:SpawnThread v-app -app
family:native stack.function:_NSRaiseError ^-group -app
family:native stack.function:std::* -app
family:native stack.function:core::* -app
These recommendations will greatly improve your out-of-the-box grouping experience.
To proactively improve your experience, help Sentry determine which frames in your stack trace are "in-app" (part of your own application) and which are not. The SDK defines the default rules, but in many cases, this can be improved on the server as well. In particular, for languages where server-side processing is necessary (for example, Native C, C++, or JavaScript), it's better to override this on the server.
For instance, the following marks all frames that are below a specific C++ namespace are in-app:
stack.function:myapplication::* +app
See in-app frames for Apple to find out how the sentry-cocoa
SDK marks frames as in-app.
The following marks frames from libdispatch starting with _dispatch_
as inApp
.
stack.function:_dispatch_* +app
You can also achieve the same result by marking other frames "not in-app." However, if that's the case, you should ensure that first all frames are set to "in-app" to override the defaults:
app:no +app
stack.function:std::* -app
stack.function:boost::* -app
You need to force all frames to be in-app first because there might already have been some defaults set by the client SDK or earlier processing.
If you're using the sentry-cocoa
SDK, you can also achieve the same result by marking other frames not inApp
. The following sample marks functions from the classes DataRequest
and DownloadRequest
as not inApp
.
stack.function:DataRequest* -app
stack.function:DownloadRequest* -app
In many cases, you want to remove the top or bottom of the stack trace. For instance, many code bases use a common function to generate an error. In this case, the error machinery will appear as part of the stack trace.
For example, if you use Rust, you likely want to remove some frames that are related to panic handling:
stack.function:std::panicking::begin_panic ^-app -app ^-group -group
stack.function:core::panicking::begin_panic ^-app -app ^-group -group
Here we tell the system that all frames from begin-panic
to the crash location are not part of the application (including the panic frame itself). All frames above are, in all cases, irrelevant for grouping.
Likewise, you can also remove the base of a stack trace. This is particularly useful if you have different main loops that drive an application:
stack.function:myapp::LinuxMainLoop v-group -group
stack.function:myapp::MacMainLoop v-group -group
stack.function:myapp::WinMainLoop v-group -group
This isn't useful for all projects, but it can work well for large applications with many crashes. The default strategy is to consider most of the stack trace relevant for grouping. This means that every different stack trace that leads to a crashing function will cause a different group to be created. If you do not want that, you can force the groups to be much larger by limiting how many frames should be considered.
For instance, if any of the frames in the stack trace refer to a common external library, you could tell the system only to consider the top N frames:
# always only consider the top 1 frame for all native events
family:native max-frames=1
# if the bug is in proprietarymodule.so, only consider top 2 frames
family:native stack.package:**/proprietarymodule.so max-frames=2
# these are functions we want to consider much more of the stack trace for
family:native stack.function:KnownBadFunction1 max-frames=5
family:native stack.function:KnownBadFunction2 max-frames=5
Our documentation is open source and available on GitHub. Your contributions are welcome, whether fixing a typo (drat!) or suggesting an update ("yeah, this would be better").