An unofficial Crystal-language client and integration layer for the Sentry error reporting API.

Based on fine raven-ruby gem from folks at @getsentry.

• Crystal error tracking with Sentry
• Sentry error tracking for Kemal

• Processors (data scrubbers)
• Interfaces (Message, Exception, Stacktrace, User, HTTP, ...)
• Contexts (user, tags, extra, os, runtime)
• Breadcrumbs
• Integrations (Kemal, Amber, Lucky, Sidekiq.cr, action-controller)
• Async support
• User Feedback
• Source code context for stack traces
• Dedicated Log backend
• Crash Handler

Add this to your application's shard.yml:

dependencies:
  raven:
    github: Sija/raven.cr

require "raven"

Raven will capture and send exceptions to the Sentry server whenever its DSN is set. This makes environment-based configuration easy - if you don't want to send errors in a certain environment, just don't set the DSN in that environment!

# Set your SENTRY_DSN environment variable.
export SENTRY_DSN=https://[email protected]/project-id

# Or you can configure the client in the code (not recommended - keep your DSN secret!)
Raven.configure do |config|
  config.dsn = "https://[email protected]/project-id"
end

If used with integrations, Raven ignores some exceptions by default - most of these are related to 404s or controller actions not being found.

Raven doesn't report POST, PUT, PATCH data or cookies by default. In addition, it will attempt to remove any obviously sensitive data, such as credit card or Social Security numbers. For more information about how Sentry processes your data, check out the documentation on the processors config setting.

Raven supports two methods of capturing exceptions:

Raven.capture do
  # capture any exceptions which happen during execution of this block
  1 // 0
end

begin
  1 // 0
rescue ex : DivisionByZeroError
  Raven.capture(ex)
end

You're all set - but there are a few more settings you may want to know about too!

While we advise that you set your Sentry DSN through the SENTRY_DSN environment variable, there are two other configuration settings for controlling Raven:

# DSN can be configured as a config setting instead.
# Place in config/initializers or similar.
Raven.configure do |config|
  config.dsn = "your_dsn"
end

And, while not necessary if using SENTRY_DSN, you can also provide an environments setting. Raven will only capture events when SENTRY_ENVIRONMENT matches an environment on the list.

Raven.configure do |config|
  config.environments = %w(staging production)
end

When an error or message occurs, the notification is immediately sent to Sentry. Raven can be configured to send asynchronously:

# define your own handler
config.async = ->(event : Raven::Event) {
  spawn { Raven.send_event(event) }
}
# or use default implementation based on fibers (i.e. the one above)
config.async = true

If the async callback raises an exception, Raven will attempt to send synchronously.

We recommend creating a background job, using your background job processor, that will send Sentry notifications in the background. Rather than enqueuing an entire Raven::Event object, we recommend providing the Hash representation of an event as a job argument.

Here’s an example for Sidekiq.cr:

config.async = ->(event : Raven::Event) {
  # enqueue the job with a hash...
  SentryJob.async.perform(event.to_hash)
  # or with JSON string
  # SentryJob.async.perform(event.to_json)
}

class SentryJob
  include Sidekiq::Worker

  sidekiq_options do |job|
    job.queue = "sentry"
    job.retry = true
  end

  def perform(event : Raven::Event::HashType)
    Raven.send_event(event)
  end
end

If Raven fails to send an event to Sentry for any reason (either the Sentry server has returned a 4XX or 5XX response), this Proc will be called.

config.transport_failure_callback = ->(event : Raven::Event::HashType) {
  AdminMailer.async.perform("Oh god, it's on fire!", event)
}

Much of the usefulness of Sentry comes from additional context data with the events. Raven makes this very convenient by providing methods to set context data that is then submitted automatically with all events.

There are three primary methods for providing request context:

# bind the logged in user
Raven.user_context email: "[email protected]"

# tag the request with something interesting
Raven.tags_context interesting: "yes"

# provide a bit of additional context
Raven.extra_context happiness: "very"

For more information, see Context.

Raven::LogBackend allows for intercepting log entries, and takes following options:

• record_breadcrumbs - records every log entry as Breadcrumbs
• capture_exceptions - captures exceptions attached to the log entry
• capture_all - captures every log entry

Every captured Exception or a Breadcrumb will have corresponding fields mapped directly from the original Log::Entry.

Metadata will be passed as Event#tags and Breadcrumb#data, respectively.

# append it to the existing bindings
Log.builder.bind "*", :info, Raven::LogBackend.new(
  record_breadcrumbs: true,
  capture_exceptions: false,
  capture_all: false,
)

# or bind it within the `Log.setup` block
Log.setup do |c|
  # bind the regular io-based backend
  c.bind "*", :info, Log::IOBackend.new

  # bind raven's backend
  c.bind "*", :info, Raven::LogBackend.new(record_breadcrumbs: true)
  c.bind "*", :warn, Raven::LogBackend.new(capture_exceptions: true)
  c.bind "*", :fatal, Raven::LogBackend.new(capture_all: true)
end

See more in Crystal's Log documentation.

Since Crystal doesn't provide native handlers for unhandled exceptions and segfaults, raven.cr introduces its own crash handler compiled as external binary.

The easiest way of using it is by adding the appropriate entry to the project's shard.yml:

targets:
  # other target definitions if any...

  sentry.crash_handler:
    main: lib/raven/src/crash_handler.cr

With the above entry defined in targets, running shards build should result in binary built in bin/sentry.crash_handler.

NOTE: While building you might specify SENTRY_DSN env variable, which will be compiled into the binary (as plain-text) and used by the handler.

SENTRY_DSN=<your_dsn> shards build sentry.crash_handler

Pass --release flag to disable debug messages.

You need to run your app with previously built bin/sentry.crash_handler in front.

bin/sentry.crash_handler bin/your_app --some arguments --passed to your program

As one would expect, STDIN is passed to the original process, while STDOUT and STDERR are piped back from it.

NOTE: You can always pass SENTRY_DSN env variable during execution in case you didn't do it while building the wrapper.

• Documentation
• Bug Tracker
• Code Repository

1. Fork it
2. Create your feature branch (git checkout -b my-new-feature)
3. Commit your changes (git commit -am 'Add some feature')
4. Push to the branch (git push origin my-new-feature)
5. Create a new Pull Request

• @Sija Sijawusz Pur Rahnama - creator, maintainer