raygun4go adds Raygun-based error handling to your golang code. It catches all occuring errors, extracts as much information as possible and sends the error to Raygun via their REST-API.
$ go get github.com/MindscapeHQ/raygun4go
Include the package and then defer the HandleError-method as soon as possible in a context as global as possible. In webservers, this will probably be your request handling method, in all other programs it should be your main-method. It'll automatically capture any panic and report it. Having found the right spot, just add the following example code:
raygun, err := raygun4go.New("appName", "apiKey")
if err != nil {
log.Println("Unable to create Raygun client:", err.Error())
}
raygun.Silent(true)
defer raygun.HandleError()
where appName
is the name of your app and apiKey
is your
Raygun-API-key. If your program runs into a panic now (which you can easily
test by adding panic("foo")
after the call to defer
), the handler will
print the resulting error message. If you remove the line
raygun.Silent(true)
the error will be sent to Raygun using your API-key.
To send errors manually you can use CreateError(message string)
, or SendError(error error)
. CreateError
creates an
error and immediately reports it, SendError
immediately reports the error.
if err := raygun.CreateError("something bad happened"); err != nil {
log.Printf("failed to report error to Raygun: %v\n", err)
}
err := something.Do()
if err := raygun.SendError(err); err != nil {
log.Printf("failed to report error to Raygun: %v\n", err)
}
The client returned by New
has several chainable option-setting methods
Method | Description |
---|---|
Silent(bool) |
If set to true, this prevents the handler from sending the error to Raygun, printing it instead. |
Request(*http.Request) |
Adds the responsible http.Request to the error. |
Version(string) |
If your program has a version, you can add it here. |
Tags([]string) |
Adds the given tags to the error. These can be used for filtering later. |
CustomData(interface{}) |
Add arbitrary custom data to you error. Will only reach Raygun if it works with json.Marshal() . |
User(string) |
Add the name of the affected user to the error. |
By default, the Raygun service will group errors together based on stacktrace content. If you have any cases where you want to control the error grouping yourself, then you can provide a custom-grouping-key callback function. Below is a simple example of this, that returns a hard-coded grouping key, which would cause all errors to be grouped together:
raygun.CustomGroupingKeyFunction(func(error, raygun4go.PostData)string{return "customGroupingKey"})
The callback takes the original error, and the Raygun PostData payload structure that is about to be serialized and sent to Raygun. In your callback, you can check these values to help build your own grouping key logic based on different cases that you want to control. For any error you don't want to group yourself, return an empty string - Raygun will then use the default grouping.
Have a bug or a feature request? Please first check the list of issues.
If your problem or idea is not addressed yet, please open a new issue.