Serverless Sentry Plugin

homepage icon https://github.com/arabold/serverless-sentry-plugin

Tracked

NPM Downloads Last Month
1305
Issues
6
Stars
36
Forks
3
Watchers
36

Repo README Contents:

⚡️ Serverless Sentry Plugin

serverless npm license dependencies

About

This Serverless plugin simplifies integration of Sentry with the popular Serverless Framework and AWS Lambda.

Currently we support Node.js 4.3.2, Node.js 6.10.2 as well as Python running on AWS Lambda. Other platforms can be added by providing a respective integration library. Pull Requests are welcome!

What is Raven and Sentry?

It’s a bit confusing, but Raven is the official name of the error reporting SDK that will forward errors, exceptions and messages to the Sentry server. For more details of what Raven and Sentry actually is, refer to the official Sentry documentation: https://docs.sentry.io/.

The Serverless Sentry plugin and library is not affiliated with either Sentry or Serverless but developed independently and in my spare time.

Benefits

Overview

Sentry integration splits into two components:

  1. This plugin which simplifies installation with the Serverless Framework
  2. The serverless-sentry-lib which performs the runtime monitoring and error reporting.

For a detailed overview of how to use the serverless-sentry-lib refer to its README.md.

Installation

Usage

The Serverless Sentry Plugin allows configuration of the library through the serverless.yml and will create release and deployment information for you (if wanted). This is the recommended way of using the serverless-sentry-lib library.

▶️ Step 1: Load the Plugin

The plugin determines your environment during deployment and adds the SENTRY_DSN environment variables to your Lambda function. All you need to do is to load the plugin and set the dsn configuration option as follows:

service: my-serverless-project
provider:
  # ...
plugins:
  serverless-sentry
custom:
  sentry:
    dsn: https://xxxx:yyyy@sentry.io/zzzz # URL provided by Sentry

▶️ Step 2: Wrap Your Function Handler Code

The actual reporting to Sentry happens in platform specific libraries. Currently only Node.js and Python are supported.

Each library provides a RavenLambdaWrapper helper that act as decorators around your original AWS Lambda handler code and is configured via this plugin or manually through environment variables.

For more details refer to the individual libraries’ repositories:

Node.js

Import RavenLambdaWrapper from the serverless-sentry-lib Node.js module and pass in your Raven client as shown below - that’s it. Passing in your own client is necessary to ensure that the wrapper uses the same environment as the rest of your code. In the rare circumstances that this isn’t desired, you can pass in null instead.

Original Lambda Handler Code Before Adding RavenLambdaWrapper:

"use strict";

module.exports.hello = function(event, context, callback) {
  callback(null, { message: 'Go Serverless! Your function executed successfully!', event });
};

New Lambda Handler Code With RavenLambdaWrapper For Sentry Reporting

"use strict";

const Raven = require("raven"); // Official `raven` module
const RavenLambdaWrapper = require("serverless-sentry-lib");

module.exports.hello = RavenLambdaWrapper.handler(Raven, (event, context, callback) => {
  callback(null, { message: 'Go Serverless! Your function executed successfully!', event });
});

For more details about the different configuration options available refer to the serverless-sentry-lib documentation.

Python 🐍

Import RavenLambdaWrapper from the raven-python-lambda module as shown below:

Original Lambda Handler Code Before Adding RavenLambdaWrapper:

def handler(event, context):
    print("Go Serverless! Your function executed successfully")

New Lambda Handler Code With RavenLambdaWrapper For Sentry Reporting

from raven import Client # Offical `raven` module
from raven-python-lambda import RavenLambdaWrapper

@RavenLambdaWrapper()
def handler(event, context):
    print("Go Serverless! Your function executed successfully")

For more details about the Python integration refer to official repository at Netflix-Skunkworks/raven-python-lambda

Plugin Configuration Options

Configure the Sentry plugin using the following options in your serverless.yml:

Sentry API access

In order for some features such as releases and deployments to work, you need to grant API access to this plugin by setting the following options:

👉 Important: You need to make sure you’re using Auth Tokens not API Keys, which are deprecated.

Releases

Releases are used by Sentry to provide you with additional context when determining the cause of an issue. The plugin can automatically create releases for you and tag all messages accordingly. To find out more about releases in Sentry, refer to the official documentation.

In order to enable release tagging, you need to set the release option in your serverless.yml:

custom:
  sentry:
    dsn: https://xxxx:yyyy@sentry.io/zzzz
    organization: my-sentry-organziation
    project: my-sentry-project
    authToken: my-sentry-api-key
    release:
      version: <RELEASE VERSION>
      refs:
        - repository: <REPOSITORY NAME>
          commit: <COMMIT HASH>
          previousCommit: <COMMIT HASH>

If you don’t specify any refs, you can also use the short notation for release and simply set it to the desired release version as follows:

custom:
  sentry:
    dsn: https://xxxx:yyyy@sentry.io/zzzz
    release: <RELEASE VERSION>

If you don’t need or want the plugin to create releases and deployments, you can omit the authToken, organization and project options. Messages and exceptions sent by your Lambda functions will still be tagged with the release version and show up grouped in Sentry nonetheless.

👉 Pro Tip: The possibility to use a fixed string in combination with Serverless variables allows you to inject your release version through the command line, e.g. when running on your continuous integration machine.

custom:
  sentry:
    dsn: https://xxxx:yyyy@sentry.io/zzzz
    organization: my-sentry-organziation
    project: my-sentry-project
    authToken: my-sentry-api-key
    release:
      version: ${opt:sentryVersion}
      refs:
        - repository: ${opt:sentryRepository}
          commit: ${opt:sentryCommit}

And then deploy your project using the command line options from above:

sls deploy --sentryVersion 1.0.0 --sentryRepository foo/bar --sentryCommit 2da95dfb

👉 Tip when using Sentry with multiple projects: Releases in Sentry are specific to the organization and can span multiple projects. Take this in consideration when choosing a version name. If your version applies to the current project only, you should prefix it with your project name.

If no option for release is provided, releases and deployments are disabled.

Enabling and Disabling Error Reporting Features

In addition you can configure the Sentry error reporting on a service as well as a per-function level. For more details about the individual configuration options see the serverless-sentry-lib documentation.

Example Configuration

service: my-serverless-project
provider:
  # ...
plugins:
  serverless-sentry
custom:
  sentry:
    dsn: https://xxxx:yyyy@sentry.io/zzzz # URL provided by Sentry
    captureTimeoutWarnings: false # disable timeout warnings globally for all functions
functions:
  FuncFoo:
    handler: Foo.handler
    description: Hello World
    sentry:
      captureErrors: false # Disable error capturing for this specific function only
      captureTimeoutWarnings: true # Turn timeout warnings back on
  FuncBar:
    handler: Bar.handler
    sentry: false # completely turn off Sentry reporting

Troubleshooting

No errors are reported in Sentry

Double check the DSN settings in your serverless.yml and compare it with what Sentry shows you in your project settings under “Client Keys (DSN)”. You need a URL in the following format - see the Sentry Quick Start:

{PROTOCOL}://{PUBLIC_KEY}:{SECRET_KEY}@{HOST}/{PATH}{PROJECT_ID}

Also make sure to add the plugin to your plugins list in the serverless.yml:

plugins:
  serverless-sentry
custom:
  sentry:
    dsn: https://xxxx:yyyy@sentry.io/zzzz # URL provided by Sentry

The plugin doesn’t create any releases or deployments

Make sure to set the authToken, organization as well as project options in your serverless.yml, and set release to a non-empty value as shown in the example below:

plugins:
  serverless-sentry
custom:
  sentry:
    dsn: https://xxxx:yyyy@sentry.io/zzzz # URL provided by Sentry
    organization: my-sentry-organziation
    project: my-sentry-project
    authToken: my-sentry-api-key
    release: git

When using Raven in my code, exception logs are missing details or an error is thrown during runtime

This typically happens if you forgot to pass in the original Raven instance to the RavenLambdaWrapper. If you pass in null instead, then the serverless-sentry-lib will ultimately use a different instance of Raven than the rest of your code. This is typically not what you want. Make sure to pass in the Raven instance during initialization of the RavenLambdaWrapper as shown in the examples above.

Raven throws an uncaught error: Cannot read property ‘user’ of undefined

Raven is not initialized correctly. If you use raven-node directly you must either pass in the Raven class to the RavenLambdaWrapper.handler() function as shown in step 2 above, or manually initialize your instance of the raven module.

I’m testing my Raven integration locally but no errors or messages are reported

Check out the filterLocal configuration setting. If you test Raven locally and want to make sure your messages are sent, set this flag to false. Once done testing, don’t forget to switch it back to true as otherwise you’ll spam your Sentry projects with meaningless errors of local code changes.

Version History

1.0.0

1.0.0-rc.4

1.0.0-rc.3

1.0.0-rc.2

1.0.0-rc.1

To-Dos