rails--rails/guides/source/debugging_rails_applications.md

DO NOT READ THIS FILE ON GITHUB, GUIDES ARE PUBLISHED ON http://guides.rubyonrails.org.

Debugging Rails Applications

This guide introduces techniques for debugging Ruby on Rails applications.

After reading this guide, you will know:

• The purpose of debugging.
• How to track down problems and issues in your application that your tests aren't identifying.
• The different ways of debugging.
• How to analyze the stack trace.

---

View Helpers for Debugging

One common task is to inspect the contents of a variable. Rails provides three different ways to do this:

• debug
• to_yaml
• inspect

debug

The debug helper will return a <pre> tag that renders the object using the YAML format. This will generate human-readable data from any object. For example, if you have this code in a view:

<%= debug @article %>
<p>
  <b>Title:</b>
  <%= @article.title %>
</p>

You'll see something like this:

--- !ruby/object Article
attributes:
  updated_at: 2008-09-05 22:55:47
  body: It's a very helpful guide for debugging your Rails app.
  title: Rails debugging guide
  published: t
  id: "1"
  created_at: 2008-09-05 22:55:47
attributes_cache: {}


Title: Rails debugging guide

to_yaml

Alternatively, calling to_yaml on any object converts it to YAML. You can pass this converted object into the simple_format helper method to format the output. This is how debug does its magic.

<%= simple_format @article.to_yaml %>
<p>
  <b>Title:</b>
  <%= @article.title %>
</p>

The above code will render something like this:

--- !ruby/object Article
attributes:
updated_at: 2008-09-05 22:55:47
body: It's a very helpful guide for debugging your Rails app.
title: Rails debugging guide
published: t
id: "1"
created_at: 2008-09-05 22:55:47
attributes_cache: {}

Title: Rails debugging guide

inspect

Another useful method for displaying object values is inspect, especially when working with arrays or hashes. This will print the object value as a string. For example:

<%= [1, 2, 3, 4, 5].inspect %>
<p>
  <b>Title:</b>
  <%= @article.title %>
</p>

Will render:

[1, 2, 3, 4, 5]

Title: Rails debugging guide

The Logger

It can also be useful to save information to log files at runtime. Rails maintains a separate log file for each runtime environment.

What is the Logger?

Rails makes use of the ActiveSupport::Logger class to write log information. Other loggers, such as Log4r, may also be substituted.

You can specify an alternative logger in environment.rb or any other environment file, for example:

Rails.logger = Logger.new(STDOUT)
Rails.logger = Log4r::Logger.new("Application Log")

Or in the Initializer section, add any of the following

config.logger = Logger.new(STDOUT)
config.logger = Log4r::Logger.new("Application Log")

TIP: By default, each log is created under Rails.root/log/ and the log file is named after the environment in which the application is running.

Log Levels

When something is logged, it's printed into the corresponding log if the log level of the message is equal to or higher than the configured log level. If you want to know the current log level, you can call the Rails.logger.level method.

The available log levels are: :debug, :info, :warn, :error, :fatal, and :unknown, corresponding to the log level numbers from 0 up to 5, respectively. To change the default log level, use

config.log_level = :warn # In any environment initializer, or
Rails.logger.level = 0 # at any time

This is useful when you want to log under development or staging without flooding your production log with unnecessary information.

TIP: The default Rails log level is debug in all environments.

Sending Messages

To write in the current log use the logger.(debug|info|warn|error|fatal) method from within a controller, model or mailer:

logger.debug "Person attributes hash: #{@person.attributes.inspect}"
logger.info "Processing the request..."
logger.fatal "Terminating application, raised unrecoverable error!!!"

Here's an example of a method instrumented with extra logging:

class ArticlesController < ApplicationController
  # ...

  def create
    @article = Article.new(params[:article])
    logger.debug "New article: #{@article.attributes.inspect}"
    logger.debug "Article should be valid: #{@article.valid?}"

    if @article.save
      flash[:notice] =  'Article was successfully created.'
      logger.debug "The article was saved and now the user is going to be redirected..."
      redirect_to(@article)
    else
      render action: "new"
    end
  end

  # ...
end

Here's an example of the log generated when this controller action is executed:

Processing ArticlesController#create (for 127.0.0.1 at 2008-09-08 11:52:54) [POST]
  Session ID: BAh7BzoMY3NyZl9pZCIlMDY5MWU1M2I1ZDRjODBlMzkyMWI1OTg2NWQyNzViZjYiCmZsYXNoSUM6J0FjdGl
vbkNvbnRyb2xsZXI6OkZsYXNoOjpGbGFzaEhhc2h7AAY6CkB1c2VkewA=--b18cd92fba90eacf8137e5f6b3b06c4d724596a4
  Parameters: {"commit"=>"Create", "article"=>{"title"=>"Debugging Rails",
 "body"=>"I'm learning how to print in logs!!!", "published"=>"0"},
 "authenticity_token"=>"2059c1286e93402e389127b1153204e0d1e275dd", "action"=>"create", "controller"=>"articles"}
New article: {"updated_at"=>nil, "title"=>"Debugging Rails", "body"=>"I'm learning how to print in logs!!!",
 "published"=>false, "created_at"=>nil}
Article should be valid: true
  Article Create (0.000443)   INSERT INTO "articles" ("updated_at", "title", "body", "published",
 "created_at") VALUES('2008-09-08 14:52:54', 'Debugging Rails',
 'I''m learning how to print in logs!!!', 'f', '2008-09-08 14:52:54')
The article was saved and now the user is going to be redirected...
Redirected to # Article:0x20af760>
Completed in 0.01224 (81 reqs/sec) | DB: 0.00044 (3%) | 302 Found [http://localhost/articles]

Adding extra logging like this makes it easy to search for unexpected or unusual behavior in your logs. If you add extra logging, be sure to make sensible use of log levels to avoid filling your production logs with useless trivia.

Tagged Logging

When running multi-user, multi-account applications, it's often useful to be able to filter the logs using some custom rules. TaggedLogging in Active Support helps you do exactly that by stamping log lines with subdomains, request ids, and anything else to aid debugging such applications.

logger = ActiveSupport::TaggedLogging.new(Logger.new(STDOUT))
logger.tagged("BCX") { logger.info "Stuff" }                            # Logs "[BCX] Stuff"
logger.tagged("BCX", "Jason") { logger.info "Stuff" }                   # Logs "[BCX] [Jason] Stuff"
logger.tagged("BCX") { logger.tagged("Jason") { logger.info "Stuff" } } # Logs "[BCX] [Jason] Stuff"

Impact of Logs on Performance

Logging will always have a small impact on the performance of your Rails app, particularly when logging to disk. Additionally, there are a few subtleties:

Using the :debug level will have a greater performance penalty than :fatal, as a far greater number of strings are being evaluated and written to the log output (e.g. disk).

Another potential pitfall is too many calls to Logger in your code:

logger.debug "Person attributes hash: #{@person.attributes.inspect}"

In the above example, there will be a performance impact even if the allowed output level doesn't include debug. The reason is that Ruby has to evaluate these strings, which includes instantiating the somewhat heavy String object and interpolating the variables. Therefore, it's recommended to pass blocks to the logger methods, as these are only evaluated if the output level is the same as — or included in — the allowed level (i.e. lazy loading). The same code rewritten would be:

logger.debug {"Person attributes hash: #{@person.attributes.inspect}"}

The contents of the block, and therefore the string interpolation, are only evaluated if debug is enabled. This performance savings are only really noticeable with large amounts of logging, but it's a good practice to employ.

Debugging with the byebug gem

When your code is behaving in unexpected ways, you can try printing to logs or the console to diagnose the problem. Unfortunately, there are times when this sort of error tracking is not effective in finding the root cause of a problem. When you actually need to journey into your running source code, the debugger is your best companion.

The debugger can also help you if you want to learn about the Rails source code but don't know where to start. Just debug any request to your application and use this guide to learn how to move from the code you have written into the underlying Rails code.

Setup

You can use the byebug gem to set breakpoints and step through live code in Rails. To install it, just run:

$ gem install byebug

Inside any Rails application you can then invoke the debugger by calling the byebug method.

Here's an example:

class PeopleController < ApplicationController
  def new
    byebug
    @person = Person.new
  end
end

The Shell

As soon as your application calls the byebug method, the debugger will be started in a debugger shell inside the terminal window where you launched your application server, and you will be placed at the debugger's prompt (byebug). Before the prompt, the code around the line that is about to be run will be displayed and the current line will be marked by '=>', like this:

[1, 10] in /PathTo/project/app/controllers/articles_controller.rb
    3:
    4:   # GET /articles
    5:   # GET /articles.json
    6:   def index
    7:     byebug
=>  8:     @articles = Article.find_recent
    9:
   10:     respond_to do |format|
   11:       format.html # index.html.erb
   12:       format.json { render json: @articles }

(byebug)

If you got there by a browser request, the browser tab containing the request will be hung until the debugger has finished and the trace has finished processing the entire request.

For example:

=> Booting Puma
=> Rails 5.0.0 application starting in development on http://0.0.0.0:3000
=> Run `rails server -h` for more startup options
=> Ctrl-C to shutdown server
Puma 2.15.3 starting...
* Min threads: 0, max threads: 16
* Environment: development
* Listening on tcp://localhost:3000


Started GET "/" for 127.0.0.1 at 2014-04-11 13:11:48 +0200
  ActiveRecord::SchemaMigration Load (0.2ms)  SELECT "schema_migrations".* FROM "schema_migrations"
Processing by ArticlesController#index as HTML

[3, 12] in /PathTo/project/app/controllers/articles_controller.rb
    3:
    4:   # GET /articles
    5:   # GET /articles.json
    6:   def index
    7:     byebug
=>  8:     @articles = Article.find_recent
    9:
   10:     respond_to do |format|
   11:       format.html # index.html.erb
   12:       format.json { render json: @articles }

(byebug)

Now it's time to explore your application. A good place to start is by asking the debugger for help. Type: help

(byebug) help

  h[elp][ <cmd>[ <subcmd>]]

  help                -- prints this help.
  help <cmd>          -- prints help on command <cmd>.
  help <cmd> <subcmd> -- prints help on <cmd>'s subcommand <subcmd>.

To see the previous ten lines you should type list- (or l-).

(byebug) l-

[1, 10] in /PathTo/project/app/controllers/articles_controller.rb
   1  class ArticlesController < ApplicationController
   2    before_action :set_article, only: [:show, :edit, :update, :destroy]
   3
   4    # GET /articles
   5    # GET /articles.json
   6    def index
   7      byebug
   8      @articles = Article.find_recent
   9
   10      respond_to do |format|

This way you can move inside the file and see the code above the line where you added the byebug call. Finally, to see where you are in the code again you can type list=

(byebug) list=

[3, 12] in /PathTo/project/app/controllers/articles_controller.rb
    3:
    4:   # GET /articles
    5:   # GET /articles.json
    6:   def index
    7:     byebug
=>  8:     @articles = Article.find_recent
    9:
   10:     respond_to do |format|
   11:       format.html # index.html.erb
   12:       format.json { render json: @articles }

(byebug)

The Context

When you start debugging your application, you will be placed in different contexts as you go through the different parts of the stack.

The debugger creates a context when a stopping point or an event is reached. The context has information about the suspended program which enables the debugger to inspect the frame stack, evaluate variables from the perspective of the debugged program, and know the place where the debugged program is stopped.

At any time you can call the backtrace command (or its alias where) to print the backtrace of the application. This can be very helpful to know how you got where you are. If you ever wondered about how you got somewhere in your code, then backtrace will supply the answer.

(byebug) where
--> #0  ArticlesController.index
      at /PathTo/project/test_app/app/controllers/articles_controller.rb:8
    #1  ActionController::ImplicitRender.send_action(method#String, *args#Array)
      at /PathToGems/actionpack-5.0.0/lib/action_controller/metal/implicit_render.rb:4
    #2  AbstractController::Base.process_action(action#NilClass, *args#Array)
      at /PathToGems/actionpack-5.0.0/lib/abstract_controller/base.rb:189
    #3  ActionController::Rendering.process_action(action#NilClass, *args#NilClass)
      at /PathToGems/actionpack-5.0.0/lib/action_controller/metal/rendering.rb:10
...

The current frame is marked with -->. You can move anywhere you want in this trace (thus changing the context) by using the frame _n_ command, where n is the specified frame number. If you do that, byebug will display your new context.

(byebug) frame 2

[184, 193] in /PathToGems/actionpack-5.0.0/lib/abstract_controller/base.rb
   184:       # is the intended way to override action dispatching.
   185:       #
   186:       # Notice that the first argument is the method to be dispatched
   187:       # which is *not* necessarily the same as the action name.
   188:       def process_action(method_name, *args)
=> 189:         send_action(method_name, *args)
   190:       end
   191:
   192:       # Actually call the method associated with the action. Override
   193:       # this method if you wish to change how action methods are called,

(byebug)

The available variables are the same as if you were running the code line by line. After all, that's what debugging is.

You can also use up [n] (u for abbreviated) and down [n] commands in order to change the context n frames up or down the stack respectively. n defaults to one. Up in this case is towards higher-numbered stack frames, and down is towards lower-numbered stack frames.

Threads

The debugger can list, stop, resume and switch between running threads by using the thread command (or the abbreviated th). This command has a handful of options:

• thread: shows the current thread.
• thread list: is used to list all threads and their statuses. The plus + character and the number indicates the current thread of execution.
• thread stop _n_: stop thread n.
• thread resume _n_: resumes thread n.
• thread switch _n_: switches the current thread context to n.

This command is very helpful when you are debugging concurrent threads and need to verify that there are no race conditions in your code.

Inspecting Variables

Any expression can be evaluated in the current context. To evaluate an expression, just type it!

This example shows how you can print the instance variables defined within the current context:

[3, 12] in /PathTo/project/app/controllers/articles_controller.rb
    3:
    4:   # GET /articles
    5:   # GET /articles.json
    6:   def index
    7:     byebug
=>  8:     @articles = Article.find_recent
    9:
   10:     respond_to do |format|
   11:       format.html # index.html.erb
   12:       format.json { render json: @articles }

(byebug) instance_variables
[:@_action_has_layout, :@_routes, :@_headers, :@_status, :@_request,
 :@_response, :@_prefixes, :@_lookup_context, :@_action_name,
 :@_response_body, :@marked_for_same_origin_verification, :@_config]

As you may have figured out, all of the variables that you can access from a controller are displayed. This list is dynamically updated as you execute code. For example, run the next line using next (you'll learn more about this command later in this guide).

(byebug) next
[5, 14] in /PathTo/project/app/controllers/articles_controller.rb
   5     # GET /articles.json
   6     def index
   7       byebug
   8       @articles = Article.find_recent
   9
=> 10       respond_to do |format|
   11         format.html # index.html.erb
   12        format.json { render json: @articles }
   13      end
   14    end
   15
(byebug)

And then ask again for the instance_variables:

(byebug) instance_variables
[:@_action_has_layout, :@_routes, :@_headers, :@_status, :@_request,
 :@_response, :@_prefixes, :@_lookup_context, :@_action_name,
 :@_response_body, :@marked_for_same_origin_verification, :@_config,
 :@articles]

Now @articles is included in the instance variables, because the line defining it was executed.

TIP: You can also step into irb mode with the command irb (of course!). This will start an irb session within the context you invoked it. But be warned: this is an experimental feature.

The var method is the most convenient way to show variables and their values. Let's have byebug help us with it.

(byebug) help var
v[ar] cl[ass]                   show class variables of self
v[ar] const <object>            show constants of object
v[ar] g[lobal]                  show global variables
v[ar] i[nstance] <object>       show instance variables of object
v[ar] l[ocal]                   show local variables

This is a great way to inspect the values of the current context variables. For example, to check that we have no local variables currently defined:

(byebug) var local
(byebug)

You can also inspect for an object method this way:

(byebug) var instance Article.new
@_start_transaction_state = {}
@aggregation_cache = {}
@association_cache = {}
@attributes = {"id"=>nil, "created_at"=>nil, "updated_at"=>nil}
@attributes_cache = {}
@changed_attributes = nil
...

TIP: The commands p (print) and pp (pretty print) can be used to evaluate Ruby expressions and display the value of variables to the console.

You can use also display to start watching variables. This is a good way of tracking the values of a variable while the execution goes on.

(byebug) display @articles
1: @articles = nil

The variables inside the displayed list will be printed with their values after you move in the stack. To stop displaying a variable use undisplay _n_ where n is the variable number (1 in the last example).

Step by Step

Now you should know where you are in the running trace and be able to print the available variables. But let's continue and move on with the application execution.

Use step (abbreviated s) to continue running your program until the next logical stopping point and return control to the debugger.

You may also use next which is similar to step, but function or method calls that appear within the line of code are executed without stopping.

TIP: You can also use step n or next n to move forwards n steps at once.

The difference between next and step is that step stops at the next line of code executed, doing just a single step, while next moves to the next line without descending inside methods.

For example, consider the following situation:

Started GET "/" for 127.0.0.1 at 2014-04-11 13:39:23 +0200
Processing by ArticlesController#index as HTML

[1, 8] in /home/davidr/Proyectos/test_app/app/models/article.rb
   1: class Article < ApplicationRecord
   2:
   3:   def self.find_recent(limit = 10)
   4:     byebug
=> 5:     where('created_at > ?', 1.week.ago).limit(limit)
   6:   end
   7:
   8: end

(byebug)

If we use next, we won't go deep inside method calls. Instead, byebug will go to the next line within the same context. In this case, it is the last line of the current method, so byebug will return to the next line of the caller method.

(byebug) next

Next advances to the next line (line 6: `end`), which returns to the next line
of the caller method:

[4, 13] in /PathTo/project/test_app/app/controllers/articles_controller.rb
    4:   # GET /articles
    5:   # GET /articles.json
    6:   def index
    7:     @articles = Article.find_recent
    8:
=>  9:     respond_to do |format|
   10:       format.html # index.html.erb
   11:       format.json { render json: @articles }
   12:     end
   13:   end

(byebug)

If we use step in the same situation, byebug will literally go to the next Ruby instruction to be executed -- in this case, Active Support's week method.

(byebug) step

[50, 59] in /PathToGems/activesupport-5.0.0/lib/active_support/core_ext/numeric/time.rb
   50:     ActiveSupport::Duration.new(self * 24.hours, [[:days, self]])
   51:   end
   52:   alias :day :days
   53:
   54:   def weeks
=> 55:     ActiveSupport::Duration.new(self * 7.days, [[:days, self * 7]])
   56:   end
   57:   alias :week :weeks
   58:
   59:   def fortnights

(byebug)

This is one of the best ways to find bugs in your code.

Breakpoints

A breakpoint makes your application stop whenever a certain point in the program is reached. The debugger shell is invoked in that line.

You can add breakpoints dynamically with the command break (or just b). There are 3 possible ways of adding breakpoints manually:

• break line: set breakpoint in the line in the current source file.
• break file:line [if expression]: set breakpoint in the line number inside the file. If an expression is given it must evaluated to true to fire up the debugger.
• break class(.|\#)method [if expression]: set breakpoint in method (. and # for class and instance method respectively) defined in class. The expression works the same way as with file:line.

For example, in the previous situation

[4, 13] in /PathTo/project/app/controllers/articles_controller.rb
    4:   # GET /articles
    5:   # GET /articles.json
    6:   def index
    7:     @articles = Article.find_recent
    8:
=>  9:     respond_to do |format|
   10:       format.html # index.html.erb
   11:       format.json { render json: @articles }
   12:     end
   13:   end

(byebug) break 11
Created breakpoint 1 at /PathTo/project/app/controllers/articles_controller.rb:11

Use info breakpoints _n_ or info break _n_ to list breakpoints. If you supply a number, it lists that breakpoint. Otherwise it lists all breakpoints.

(byebug) info breakpoints
Num Enb What
1   y   at /PathTo/project/app/controllers/articles_controller.rb:11

To delete breakpoints: use the command delete _n_ to remove the breakpoint number n. If no number is specified, it deletes all breakpoints that are currently active.

(byebug) delete 1
(byebug) info breakpoints
No breakpoints.

You can also enable or disable breakpoints:

• enable breakpoints: allow a breakpoints list or all of them if no list is specified, to stop your program. This is the default state when you create a breakpoint.
• disable breakpoints: the breakpoints will have no effect on your program.

Catching Exceptions

The command catch exception-name (or just cat exception-name) can be used to intercept an exception of type exception-name when there would otherwise be no handler for it.

To list all active catchpoints use catch.

Resuming Execution

There are two ways to resume execution of an application that is stopped in the debugger:

• continue [line-specification] (or c): resume program execution, at the address where your script last stopped; any breakpoints set at that address are bypassed. The optional argument line-specification allows you to specify a line number to set a one-time breakpoint which is deleted when that breakpoint is reached.
• finish [frame-number] (or fin): execute until the selected stack frame returns. If no frame number is given, the application will run until the currently selected frame returns. The currently selected frame starts out the most-recent frame or 0 if no frame positioning (e.g up, down or frame) has been performed. If a frame number is given it will run until the specified frame returns.

Editing

Two commands allow you to open code from the debugger into an editor:

• edit [file:line]: edit file using the editor specified by the EDITOR environment variable. A specific line can also be given.

Quitting

To exit the debugger, use the quit command (abbreviated to q). Or, type q! to bypass the Really quit? (y/n) prompt and exit unconditionally.

A simple quit tries to terminate all threads in effect. Therefore your server will be stopped and you will have to start it again.

Settings

byebug has a few available options to tweak its behavior:

• set autoreload: Reload source code when changed (defaults: true).
• set autolist: Execute list command on every breakpoint (defaults: true).
• set listsize _n_: Set number of source lines to list by default to n (defaults: 10)
• set forcestep: Make sure the next and step commands always move to a new line.

You can see the full list by using help set. Use help set _subcommand_ to learn about a particular set command.

TIP: You can save these settings in an .byebugrc file in your home directory. The debugger reads these global settings when it starts. For example:

set forcestep
set listsize 25

Debugging with the web-console gem

Web Console is a bit like byebug, but it runs in the browser. In any page you are developing, you can request a console in the context of a view or a controller. The console would be rendered next to your HTML content.

Console

Inside any controller action or view, you can invoke the console by calling the console method.

For example, in a controller:

class PostsController < ApplicationController
  def new
    console
    @post = Post.new
  end
end

Or in a view:

<% console %>

<h2>New Post</h2>

This will render a console inside your view. You don't need to care about the location of the console call; it won't be rendered on the spot of its invocation but next to your HTML content.

The console executes pure Ruby code: You can define and instantiate custom classes, create new models and inspect variables.

NOTE: Only one console can be rendered per request. Otherwise web-console will raise an error on the second console invocation.

Inspecting Variables

You can invoke instance_variables to list all the instance variables available in your context. If you want to list all the local variables, you can do that with local_variables.

Settings

• config.web_console.whitelisted_ips: Authorized list of IPv4 or IPv6 addresses and networks (defaults: 127.0.0.1/8, ::1).
• config.web_console.whiny_requests: Log a message when a console rendering is prevented (defaults: true).

Since web-console evaluates plain Ruby code remotely on the server, don't try to use it in production.

Debugging Memory Leaks

A Ruby application (on Rails or not), can leak memory — either in the Ruby code or at the C code level.

In this section, you will learn how to find and fix such leaks by using tool such as Valgrind.

Valgrind

Valgrind is an application for detecting C-based memory leaks and race conditions.

There are Valgrind tools that can automatically detect many memory management and threading bugs, and profile your programs in detail. For example, if a C extension in the interpreter calls malloc() but doesn't properly call free(), this memory won't be available until the app terminates.

For further information on how to install Valgrind and use with Ruby, refer to Valgrind and Ruby by Evan Weaver.

Plugins for Debugging

There are some Rails plugins to help you to find errors and debug your application. Here is a list of useful plugins for debugging:

• Footnotes Every Rails page has footnotes that give request information and link back to your source via TextMate.
• Query Trace Adds query origin tracing to your logs.
• Query Reviewer This Rails plugin not only runs "EXPLAIN" before each of your select queries in development, but provides a small DIV in the rendered output of each page with the summary of warnings for each query that it analyzed.
• Exception Notifier Provides a mailer object and a default set of templates for sending email notifications when errors occur in a Rails application.
• Better Errors Replaces the standard Rails error page with a new one containing more contextual information, like source code and variable inspection.
• RailsPanel Chrome extension for Rails development that will end your tailing of development.log. Have all information about your Rails app requests in the browser — in the Developer Tools panel. Provides insight to db/rendering/total times, parameter list, rendered views and more.

References

• ruby-debug Homepage
• debugger Homepage
• byebug Homepage
• web-console Homepage
• Article: Debugging a Rails application with ruby-debug
• Ryan Bates' debugging ruby (revised) screencast
• Ryan Bates' stack trace screencast
• Ryan Bates' logger screencast
• Debugging with ruby-debug