Backtrace captures and reports handled and unhandled exceptions in your production software so you can manage application quality through the complete product lifecycle.
The @backtrace/node SDK connects your JavaScript application to Backtrace. The basic integration is quick and easy, after which you can explore the rich set of Backtrace features.
$ npm install @backtrace/node
Add the following code to your application before all other scripts to report node errors to Backtrace.
// Import the BacktraceClient from @backtrace/node with your favorite package manager.
import { BacktraceClient, BacktraceConfiguration } from '@backtrace/node';
// Configure client options
const options: BacktraceConfiguration = {
// Submission url
// <universe> is the subdomain of your Backtrace instance (<universe>.backtrace.io)
// <token> can be found in Project Settings/Submission tokens
url: 'https://submit.backtrace.io/<universe>/<token>/json',
};
// Initialize the client with the options
const client = BacktraceClient.initialize(options);
// By default, Backtrace will send an error for Uncaught Exceptions and Unhandled Promise Rejections
// Manually send an error
client.send(new Error('Something broke!'));
Client-side error reports are based on minified code. Upload source maps and source code to resolve minified code to your original source identifiers.
(Source Map feature documentation)
Custom attributes are key-value pairs that can be added to your error reports. They are used in report aggregation, sorting and filtering, can provide better contextual data for an error, and much more. They are foundational to many of the advanced Backtrace features detailed in Error Reporting documentation. By default attributes such as application name and version are populated automatically based on your package.json information. If Backtrace cannot find them, you need to provide them manually via userAttributes attributes.
There are several places where attributes can be added, modified or deleted.
It is possible to include an attributes object during BacktraceClient initialization. This list of attributes will be included with every error report, referred to as global attributes.
// Create an attributes object that can be modified throughout runtime
const attributes: Record<string, unknown> = {
release: 'PROD',
};
// BacktraceClientOptions
const options: BacktraceConfiguration = {
url: 'https://submit.backtrace.io/<universe>/<token>/json',
// Attach the attributes object
userAttributes: attributes,
};
// Initialize the client
const client = BacktraceClient.initialize(options);
You can also include attributes that will be resolved when creating a report:
// BacktraceClientOptions
const options: BacktraceConfiguration = {
url: 'https://submit.backtrace.io/<universe>/<token>/json',
// Attach the attributes object
userAttributes: () => ({
attribute: getAttributeValue(),
}),
};
// Initialize the client
const client = BacktraceClient.initialize(options);
Global attributes can be set during the runtime once specific data has be loaded (e.g. a user has logged in).
const client = BacktraceClient.initialize(options);
...
client.addAttribute({
"clientID": "de6faf4d-d5b5-486c-9789-318f58a14476"
})
You can also add attributes that will be resolved when creating a report:
const client = BacktraceClient.initialize(options);
...
client.addAttribute(() => ({
"clientID": resolveCurrentClientId()
}))
The attributes list of a BacktraceReport object can be directly modified.
const report: BacktraceReport = new BacktraceReport('My error message', { myReportKey: 'myValue' });
report.attributes['myReportKey'] = 'New value';
Files can be attached to error reports. This can be done when initalizing the BacktraceClient, updating the BacktraceClient, or dynamically for specific reports. When including attachments in BacktraceClient, all files will be uploaded with each report.
// Import attachment types from @backtrace/node
import { BacktraceStringAttachment, BacktraceUint8ArrayAttachment, BacktraceFileAttachment } from "@backtrace/node";
// BacktraceStringAttachment should be used for text object like a log file, for example
const stringAttachment = new BacktraceStringAttachment("logfile.txt", "This is the start of my log")
// Buffer attachment is an attachment type dedicated to store buffer data
const bufferAttachment = new BacktraceBufferAttachment('buffer-attachment.txt', Buffer.from('sample'));
// File attachment is an attachment type dedicated for streaming files
const fileAttachment = new BacktraceFileAttachment('/path/to/sample/file');
// BacktraceClientOptions
const options = {
url: "https://submit.backtrace.io/<universe>/<token>/json",
// Attach the files to all reports
attachments: [path.join('/path/to/attachment'), stringAttachment],
}
const client = BacktraceClient.initialize(options);
// Later decide to add an attachment to all reports
client.addAttachment(bufferAttachment)
// After catching an exception and generating a report
try {
throw new Error("Caught exception!")
} catch (error) {
const report = const report = new BacktraceReport(error, {}, [fileAttachment])
client.send(report);
}
Breadcrumbs are snippets of chronological data tracing runtime events. This SDK records a number of events by default, and manual breadcrumbs can also be added.
(Breadcrumbs feature documentation)
Option Name | Type | Description | Default | Required? |
---|---|---|---|---|
enable |
Boolean | Determines if the breadcrumbs support is enabled. By default the value is set to true. | true |
|
logLevel |
BreadcrumbLogLevel | Specifies which log level severity to include. By default all logs are included. | All Logs | |
eventType |
BreadcrumbType | Specifies which breadcrumb type to include. By default all types are included. | All Types | |
maximumBreadcrumbs |
Number | Specifies maximum number of breadcrumbs stored by the library. By default, only 100 breadcrumbs will be stored. | 100 |
|
intercept |
(breadcrumb: RawBreadcrumb) => RawBreadcrumb | undefined; | Inspects breadcrumb and allows to modify it. If the undefined value is being returned from the method, no breadcrumb will be added to the breadcrumb storage. | All Breadcrumbs |
import { BacktraceClient, BacktraceConfiguration } from '@backtrace/node';
// BacktraceClientOptions
const options: BacktraceConfiguration = {
// ignoring all but breadcrumbs config for simplicity
breadcrumbs: {
// breadcrumbs configuration
},
};
// Initialize the client
const client = BacktraceClient.initialize(options);
Type | Description |
---|---|
Console | Adds a breadcrumb every time console log is being used by the developer. |
If PII or other information needs to be filtered from a breadcrumb, you can use the intercept function to skip or filter out the sensitive information. Any RawBreadcrumb returned will be used for the breadcrumb. If undefined is returned, no breadcrumb will be added.
In addition to all of the default breadcrumbs that are automatically collected, you can also manually add breadcrumbs of your own.
client.breadcrumbs?.info('This is a manual breadcrumb.', {
customAttr: 'wow!',
});
The Backtrace Node SDK has the ability to send usage Metrics to be viewable in the Backtrace UI.
(Stability Metrics feature documentation)
Option Name | Type | Description | Default | Required? |
---|---|---|---|---|
metricsSubmissionUrl |
String | Metrics server hostname. By default the value is set to https://events.backtrace.io. | https://events.backtrace.io |
|
enable |
Boolean | Determines if the metrics support is enabled. By default the value is set to true. | true |
|
autoSendInterval |
Number | Indicates how often crash free metrics are sent to Backtrace. The interval is a value in ms. By default, session events are sent on application startup/finish, and every 30 minutes while the application is running. If the value is set to 0. The auto send mode is disabled. In this situation the application needs to maintain send mode manually. | On application startup/finish | |
size |
Number | Indicates how many events the metrics storage can store before auto submission. | 50 |
// metrics will be undefined if not enabled
client.metrics?.send();
The Backtrace Node SDK can cache generated reports and crashes to local disk before sending them to Backtrace. This is recommended; in certain configurations Node applications can crash before the SDK finishes submitting data, and under slow internet conditions your application might wait in a closing window until the HTTP submission finishes. In such an event cached reports will be sent on next application launch.
With offline database support you can:
- cache your reports when the user doesn't have an internet connection or the service is unavailable,
- capture crashes,
- manually decide whether or not to send reports, and when.
Offline database support is disabled by default. To enable it, please add "enable: true" and the path to the directory where Backtrace can store crash data.
const client = BacktraceClient.initialize({
// ignoring all but database config for simplicity
database: {
enable: true,
path: '/path/to/the/database/directory',
captureNativeCrashes: true,
},
});
// manually send and keep the data on connection issue
client.database.send();
// manually send and remove all data no matter if received success or not.
client.database.flush();
Option Name | Type | Description | Default | Required? |
---|---|---|---|---|
enabled |
Boolean | Enable/disable offline database support. | false | |
path |
String | Local storage path for crash data. | - | |
createDatabaseDirectory |
Boolean | Allow the SDK to create the offline database directory. | true | |
autoSend |
Boolean | Sends reports to the server based on the retry settings. If the value is set to 'false', you can use the Flush or Send methods as an alternative. | true | |
maximumNumberOfRecords |
Number | The maximum number of reports stored in the offline database. When the limit is reached, the oldest reports are removed. If the value is equal to '0', then no limit is set. | 8 | |
retryInterval |
Number | The amount of time (in ms) to wait between retries if the database is unable to send a report. | 60 000 | |
maximumRetries |
Number | The maximum number of retries to attempt if the database is unable to send a report. | 3 | |
captureNativeCrashes |
Boolean | Capture and symbolicate stack traces for native crashes if the runtime supports this. A crash report is generated, stored locally, and uploaded upon next start. | false |
The Backtrace Node SDK can capture native crashes generated by a Node application such as Assert/OOM crashes. In order
to collect them, the SDK uses the Node's process.report
API. After setting up the native crash support, your
process.report
settings may be overridden and your crash data might be created in the database directory.
Database records sent in the next session may not have some information about the crashing session such as attributes or breadcrumbs. To reduce database record size, attachment support was limited only to file attachments.
Database support is available in the client options with the BacktraceDatabase object. You can use it to manually operate on database records.
BacktraceClient is the main SDK class. Error monitoring starts when this singleton object is instantiated, and it will compose and send reports for unhandled errors and unhandled promise rejections. It can also be used to manually send reports from exceptions and rejection handlers. Do not create more than one instance of this object.
The following options are available for the BacktraceClientOptions passed when initializing the BacktraceClient.
Option Name | Type | Description | Default | Required? |
---|---|---|---|---|
url |
String | Submission URL to send errors to | ||
token |
String | The submission token for error injestion. This is required only if submitting directly to a Backtrace URL. (uncommon) | ||
userAttributes |
Dictionary | Additional attributes that can be filtered and aggregated against in the Backtrace UI. | ||
attachments |
BacktraceAttachment[] | Additional files to be sent with error reports. See File Attachments | ||
beforeSend |
(data: BacktraceData) => BacktraceData | undefined | Triggers an event every time an exception in the managed environment occurs, which allows you to skip the report (by returning a null value) or to modify data that library collected before sending the report. You can use the BeforeSend event to extend attributes or JSON object data based on data the application has at the time of exception. See Modify/skip error reports) | ||
skipReport |
(report: BacktraceReport) => boolean | If you want to ignore specific types of error reports, we recommend that you use the skipReport callback. By using it, based on the data generated in the report, you can decide to filter the report, or send it to Backtrace. | ||
captureUnhandledErrors |
Boolean | Enable unhandled errors | true |
|
captureUnhandledPromiseRejections |
Boolean | Enable unhandled promise rejection | true |
|
timeout |
Integer | How long to wait in ms before timing out the connection | 15000 |
|
ignoreSslCertificate |
Boolean | Ignore SSL Certificate errors | false |
|
rateLimit |
Integer | Limits the number of reports the client will send per minute. If set to '0', there is no limit. If set to a value greater than '0' and the value is reached, the client will not send any reports until the next minute. | 0 |
|
metrics |
BacktraceMetricsOptions | See Backtrace Stability Metrics | ||
breadcrumbs |
BacktraceBreadcrumbsSettings | See Backtrace Breadcrumbs | ||
database |
BacktraceDatabaseSettings | See Backtrace Database |
There are several ways to send an error to Backtrace. For more details on the definition of client.send()
see
Methods below.
// send as a string
await client.send('This is a string!');
// send as an Error
await client.send(new Error('This is an Error!'));
// as a BacktraceReport (string)
await client.send(new BacktraceReport('This is a report with a string!'));
// as a BacktraceReport (Error)
await client.send(new BacktraceReport(new Error('This is a report with a string!')));
A BeforeSend event is triggered when an exception in the managed environment occurs to which you can attach a handler. You can use the BeforeSend event to scrub PII, or extend attributes or JSON object data based on data your application has at the time of exception. A report can be skipped sompletely by returning a null value.
const client = BacktraceClient.initialize({
url: SUBMISSION_URL,
beforeSend: (data: BacktraceData) => {
// skip the report by returning a null from the callback
if (!shouldSendReportToBacktrace(data)) {
return undefined;
}
// apply custom attribute
data.attributes['new-attribute"] = 'apply-data-in-callback';
return data;
},
});
BacktraceClient.builder is used to override default BacktraceClient methods. File and http operation overrides, for example, can be used to implement custom encryption for data at rest or in motion.
Do not use these operations to modify the data objects. See Modify/skip error reports for the correct method to modify a report before sending it to Backtrace.
const client = BacktraceClient.builder(options)
.useRequestHandler(requestHandler)
.useBreadcrumbSubscriber(breadcrumbSubscriber)
.addAttributeProvider(attributeProvider)
.build();