Track errors in Ruby and Rails

• Ruby 3.0 or later.
• An Observe key that starts with obs_. Read Get an ingest key.
• Version 0.2.0 of the gem. It has no runtime dependencies.

# Gemfile
gem "postdeploy", "~> 0.2"

Run bundle install, then set POSTDEPLOY_API_KEY in the environment of every process that reports errors.

The gem's Railtie adds PostDeploy::RackMiddleware inside ActionDispatch::DebugExceptions and reports errors from Rails.error. It also hooks into ActiveJob and Sidekiq when they are present. Do not add the middleware again.

class ApplicationController < ActionController::Base
  before_action do
    PostDeploy.set_user(id: current_user.id, email: current_user.email) if current_user
  end
end

Add the middleware first. It reports the error with the request method, URL without query string, parameters and allow-listed headers, then raises the error again so your error page still renders.

require "postdeploy"

use PostDeploy::RackMiddleware

run lambda { |env|
  PostDeploy.set_user(id: "user_rack_1")
  raise "Checkout failed" if env["PATH_INFO"] == "/boom"
  [200, { "content-type" => "text/plain" }, ["ok"]]
}

Call PostDeploy.flush before a short process exits. Version 0.2.0 does not flush at exit, so a script that ends without it loses its events.

require "postdeploy"

PostDeploy.configure do |config|
  config.environment = "production"
  config.release = ENV["GIT_SHA"]
end

PostDeploy.set_user(id: "user_456", email: "[email protected]")
PostDeploy.set_tag("plan", "team")
PostDeploy.set_context("order", { order_id: "ord_789" })

begin
  raise ArgumentError, "Invoice total is negative"
rescue => e
  PostDeploy.capture_exception(e, context: { feature: "billing" })
end

PostDeploy.capture_message("Low disk space", level: "warning")
PostDeploy.flush

1. 1

   Step 1. Run the script.

   Terminal shell

   Copy

   POSTDEPLOY_API_KEY=obs_your_key ruby report.rb

2. 2

   Step 2. Open Observe. The table shows "ArgumentError: Invoice total is negative" as an error and "Low disk space" as a warning.

A rake task or script reports nothing

Call PostDeploy.flush before the process exits.

Errors in development never arrive

Keep the Railtie's position. A second middleware at the top of the stack runs after DebugExceptions has handled the error.

Events go to the wrong host

Leave config.endpoint and POSTDEPLOY_INGEST_URL unset in production. The default is the PostDeploy ingest host.

The error page still renders

The middleware raises the error again after it reports it. This is expected.