Bump version for 1.9.0

Downgrade from Kamal 2 to 1.9

.github/workflows/ci.yml

@@ -3,31 +3,34 @@ on:
   push:
     branches:
       - main
+      - 1-8-stable
   pull_request:
 jobs:
+  rubocop:
+    name: RuboCop
+    runs-on: ubuntu-latest
+    env:
+      BUNDLE_ONLY: rubocop
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+      - name: Setup Ruby and install gems
+        uses: ruby/setup-ruby@v1
+        with:
+          ruby-version: 3.3.0
+          bundler-cache: true
+      - name: Run Rubocop
+        run: bundle exec rubocop --parallel
   tests:
     strategy:
       matrix:
         ruby-version:
-          - "2.7"
           - "3.1"
           - "3.2"
           - "3.3"
         gemfile:
           - Gemfile
-          - gemfiles/ruby_2.7.gemfile
           - gemfiles/rails_edge.gemfile
-        exclude:
-          - ruby-version: "2.7"
-            gemfile: Gemfile
-          - ruby-version: "2.7"
-            gemfile: gemfiles/rails_edge.gemfile
-          - ruby-version: "3.1"
-            gemfile: gemfiles/ruby_2.7.gemfile
-          - ruby-version: "3.2"
-            gemfile: gemfiles/ruby_2.7.gemfile
-          - ruby-version: "3.3"
-            gemfile: gemfiles/ruby_2.7.gemfile
     name: ${{ format('Tests (Ruby {0})', matrix.ruby-version) }}
     runs-on: ubuntu-latest
     continue-on-error: true

.rubocop.yml

@@ -0,0 +1,2 @@
+inherit_gem:
+  rubocop-rails-omakase: rubocop.yml

Dockerfile

@@ -1,7 +1,7 @@
 # Use the official Ruby 3.2.0 Alpine image as the base image
 FROM ruby:3.2.0-alpine
 
-# Install  docker/buildx-bin
+# Install docker/buildx-bin
 COPY --from=docker/buildx-bin /buildx /usr/libexec/docker/cli-plugins/docker-buildx
 
 # Set the working directory to /kamal
@@ -14,7 +14,7 @@ COPY Gemfile Gemfile.lock kamal.gemspec ./
 COPY lib/kamal/version.rb /kamal/lib/kamal/version.rb
 
 # Install system dependencies
-RUN apk add --no-cache --update build-base git docker openrc openssh-client-default \
+RUN apk add --no-cache build-base git docker openrc openssh-client-default \
     && rc-update add docker boot \
     && gem install bundler --version=2.4.3 \
     && bundle install

Gemfile

@@ -1,4 +1,8 @@
-source 'https://rubygems.org'
+source "https://rubygems.org"
 git_source(:github) { |repo| "https://github.com/#{repo}.git" }
 
 gemspec
+
+group :rubocop do
+  gem "rubocop-rails-omakase", require: false
+end

Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    kamal (1.4.0)
+    kamal (1.9.0)
       activesupport (>= 7.0)
       base64 (~> 0.2)
       bcrypt_pbkdf (~> 1.0)
@@ -9,7 +9,7 @@ PATH
       dotenv (~> 2.8)
       ed25519 (~> 1.2)
       net-ssh (~> 7.0)
-      sshkit (~> 1.21)
+      sshkit (>= 1.23.0, < 2.0)
       thor (~> 1.2)
       zeitwerk (~> 2.5)
 
@@ -42,6 +42,7 @@ GEM
       minitest (>= 5.1)
       mutex_m
       tzinfo (~> 2.0)
+    ast (2.4.2)
     base64 (0.2.0)
     bcrypt_pbkdf (1.1.0)
     bigdecimal (3.1.5)
@@ -63,6 +64,8 @@ GEM
     irb (1.11.0)
       rdoc
       reline (>= 0.3.8)
+    json (2.7.1)
+    language_server-protocol (3.17.0.3)
     loofah (2.22.0)
       crass (~> 1.0.2)
       nokogiri (>= 1.12.0)
@@ -72,6 +75,8 @@ GEM
     mutex_m (0.2.0)
     net-scp (4.0.0)
       net-ssh (>= 2.6.5, < 8.0.0)
+    net-sftp (4.0.0)
+      net-ssh (>= 5.0.0, < 8.0.0)
     net-ssh (7.2.1)
     nokogiri (1.16.0-arm64-darwin)
       racc (~> 1.4)
@@ -79,6 +84,10 @@ GEM
       racc (~> 1.4)
     nokogiri (1.16.0-x86_64-linux)
       racc (~> 1.4)
+    parallel (1.24.0)
+    parser (3.3.0.5)
+      ast (~> 2.4.1)
+      racc
     psych (5.1.2)
       stringio
     racc (1.7.3)
@@ -105,20 +114,55 @@ GEM
       rake (>= 12.2)
       thor (~> 1.0, >= 1.2.2)
       zeitwerk (~> 2.6)
+    rainbow (3.1.1)
     rake (13.1.0)
     rdoc (6.6.2)
       psych (>= 4.0.0)
+    regexp_parser (2.9.0)
     reline (0.4.2)
       io-console (~> 0.5)
+    rexml (3.2.6)
+    rubocop (1.62.1)
+      json (~> 2.3)
+      language_server-protocol (>= 3.17.0)
+      parallel (~> 1.10)
+      parser (>= 3.3.0.2)
+      rainbow (>= 2.2.2, < 4.0)
+      regexp_parser (>= 1.8, < 3.0)
+      rexml (>= 3.2.5, < 4.0)
+      rubocop-ast (>= 1.31.1, < 2.0)
+      ruby-progressbar (~> 1.7)
+      unicode-display_width (>= 2.4.0, < 3.0)
+    rubocop-ast (1.31.2)
+      parser (>= 3.3.0.4)
+    rubocop-minitest (0.35.0)
+      rubocop (>= 1.61, < 2.0)
+      rubocop-ast (>= 1.31.1, < 2.0)
+    rubocop-performance (1.20.2)
+      rubocop (>= 1.48.1, < 2.0)
+      rubocop-ast (>= 1.30.0, < 2.0)
+    rubocop-rails (2.24.0)
+      activesupport (>= 4.2.0)
+      rack (>= 1.1)
+      rubocop (>= 1.33.0, < 2.0)
+      rubocop-ast (>= 1.31.1, < 2.0)
+    rubocop-rails-omakase (1.0.0)
+      rubocop
+      rubocop-minitest
+      rubocop-performance
+      rubocop-rails
+    ruby-progressbar (1.13.0)
     ruby2_keywords (0.0.5)
-    sshkit (1.21.7)
+    sshkit (1.23.0)
-      mutex_m
+      base64
       net-scp (>= 1.1.2)
+      net-sftp (>= 2.1.2)
       net-ssh (>= 2.8.0)
     stringio (3.1.0)
     thor (1.3.0)
     tzinfo (2.0.6)
       concurrent-ruby (~> 1.0)
+    unicode-display_width (2.5.0)
     webrick (1.8.1)
     zeitwerk (2.6.12)
 
@@ -132,6 +176,7 @@ DEPENDENCIES
   kamal!
   mocha
   railties
+  rubocop-rails-omakase
 
 BUNDLED WITH
    2.4.3

bin/docs

@@ -0,0 +1,134 @@
+#!/usr/bin/env ruby
+require "stringio"
+
+def usage
+  puts "Usage: #{$0} <kamal_site_repo>"
+  exit 1
+end
+
+usage if ARGV.size != 1
+
+kamal_site_repo = ARGV[0]
+
+if !File.directory?(kamal_site_repo)
+  puts "Error: #{kamal_site_repo} is not a directory"
+  exit 1
+end
+
+DOCS = {
+  "accessory" => "Accessories",
+  "boot" => "Booting",
+  "builder" => "Builders",
+  "configuration" => "Configuration overview",
+  "env" => "Environment variables",
+  "healthcheck" => "Healthchecks",
+  "logging" => "Logging",
+  "registry" => "Docker Registry",
+  "role" => "Roles",
+  "servers" => "Servers",
+  "ssh" => "SSH",
+  "sshkit" => "SSHKit",
+  "traefik" => "Traefik"
+}
+
+class DocWriter
+  attr_reader :from_file, :to_file, :key, :heading, :body, :output, :in_yaml
+
+  def initialize(from_file, to_dir)
+    @from_file = from_file
+    @key = File.basename(from_file, ".yml")
+    @to_file = File.join(to_dir, "#{linkify(DOCS[key])}.md")
+    @body = File.readlines(from_file)
+    @heading = body.shift.chomp("\n")
+    @output = nil
+  end
+
+  def write
+    puts "Writing #{to_file}"
+    generate_markdown
+    File.write(to_file, output.string)
+  end
+
+  private
+    def generate_markdown
+      @output = StringIO.new
+
+      generate_header
+
+      place = :in_section
+
+      loop do
+        line = body.shift&.chomp("\n")
+        break if line.nil?
+
+        case place
+        when :new_section, :in_section
+          if line.empty?
+            output.puts
+            place = :new_section
+          elsif line =~ /^ *#/
+            generate_line(line, place: place)
+            place = :in_section
+          else
+            output.puts "```yaml"
+            output.print line
+            place = :in_yaml
+          end
+        when :in_yaml
+          if line =~ /^ *#/
+            output.puts "```"
+            generate_line(line, place: :new_section)
+            place = :in_section
+          else
+            output.puts
+            output.print line
+          end
+        end
+      end
+
+      output.puts "\n```" if place == :in_yaml
+    end
+
+    def generate_header
+      output.puts "---"
+      output.puts "title: #{heading[2..-1]}"
+      output.puts "---"
+      output.puts
+      output.puts heading
+      output.puts
+    end
+
+    def generate_line(line, place: :in_section)
+      line = line.gsub(/^ *#\s?/, "")
+
+      if line =~ /(.*)kamal docs ([a-z]*)(.*)/
+        line = "#{$1}[#{DOCS[$2]}](../#{linkify(DOCS[$2])})#{$3}"
+      end
+
+      if line =~ /(.*)https:\/\/kamal-deploy.org([a-z\/-]*)(.*)/
+        line = "#{$1}[#{titlify($2.split("/").last)}](#{$2})#{$3}"
+      end
+
+      if place == :new_section
+        output.puts "## [#{line}](##{linkify(line)})"
+      else
+        output.puts line
+      end
+    end
+
+    def linkify(text)
+      text.downcase.gsub(" ", "-")
+    end
+
+    def titlify(text)
+      text.capitalize.gsub("-", " ")
+    end
+end
+
+from_dir = File.join(File.dirname(__FILE__), "../lib/kamal/configuration/docs")
+to_dir = File.join(kamal_site_repo, "docs/configuration")
+Dir.glob("#{from_dir}/*") do |from_file|
+  key = File.basename(from_file, ".yml")
+
+  DocWriter.new(from_file, to_dir).write
+end

gemfiles/ruby_2.7.gemfile

@@ -1,6 +0,0 @@
-source 'https://rubygems.org'
-git_source(:github) { |repo| "https://github.com/#{repo}.git" }
-
-gemspec path: "../"
-
-gem "nokogiri", "~> 1.15.0"

kamal.gemspec

@@ -12,7 +12,7 @@ Gem::Specification.new do |spec|
   spec.executables = %w[ kamal ]
 
   spec.add_dependency "activesupport", ">= 7.0"
-  spec.add_dependency "sshkit", "~> 1.21"
+  spec.add_dependency "sshkit", ">= 1.23.0", "< 2.0"
   spec.add_dependency "net-ssh", "~> 7.0"
   spec.add_dependency "thor", "~> 1.2"
   spec.add_dependency "dotenv", "~> 2.8"

lib/kamal.rb

@@ -1,10 +1,12 @@
 module Kamal
+  class ConfigurationError < StandardError; end
 end
 
 require "active_support"
 require "zeitwerk"
+require "yaml"
 
 loader = Zeitwerk::Loader.for_gem
-loader.ignore("#{__dir__}/kamal/sshkit_with_ext.rb")
+loader.ignore(File.join(__dir__, "kamal", "sshkit_with_ext.rb"))
 loader.setup
 loader.eager_load # We need all commands loaded.

lib/kamal/cli.rb

@@ -1,6 +1,6 @@
 module Kamal::Cli
-  class LockError < StandardError; end
   class HookError < StandardError; end
+  class LockError < StandardError; end
 end
 
 # SSHKit uses instance eval, so we need a global const for ergonomics

lib/kamal/cli/accessory.rb

@@ -1,7 +1,7 @@
 class Kamal::Cli::Accessory < Kamal::Cli::Base
   desc "boot [NAME]", "Boot new accessory service on host (use NAME=all to boot all accessories)"
   def boot(name, login: true)
-    mutating do
+    with_lock do
       if name == "all"
         KAMAL.accessory_names.each { |accessory_name| boot(accessory_name) }
       else
@@ -21,7 +21,7 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
 
   desc "upload [NAME]", "Upload accessory files to host", hide: true
   def upload(name)
-    mutating do
+    with_lock do
       with_accessory(name) do |accessory, hosts|
         on(hosts) do
           accessory.files.each do |(local, remote)|
@@ -38,7 +38,7 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
 
   desc "directories [NAME]", "Create accessory directories on host", hide: true
   def directories(name)
-    mutating do
+    with_lock do
       with_accessory(name) do |accessory, hosts|
         on(hosts) do
           accessory.directories.keys.each do |host_path|
@@ -51,7 +51,7 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
 
   desc "reboot [NAME]", "Reboot existing accessory on host (stop container, remove container, start new container; use NAME=all to boot all accessories)"
   def reboot(name)
-    mutating do
+    with_lock do
       if name == "all"
         KAMAL.accessory_names.each { |accessory_name| reboot(accessory_name) }
       else
@@ -70,7 +70,7 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
 
   desc "start [NAME]", "Start existing accessory container on host"
   def start(name)
-    mutating do
+    with_lock do
       with_accessory(name) do |accessory, hosts|
         on(hosts) do
           execute *KAMAL.auditor.record("Started #{name} accessory"), verbosity: :debug
@@ -82,7 +82,7 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
 
   desc "stop [NAME]", "Stop existing accessory container on host"
   def stop(name)
-    mutating do
+    with_lock do
       with_accessory(name) do |accessory, hosts|
         on(hosts) do
           execute *KAMAL.auditor.record("Stopped #{name} accessory"), verbosity: :debug
@@ -94,7 +94,7 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
 
   desc "restart [NAME]", "Restart existing accessory container on host"
   def restart(name)
-    mutating do
+    with_lock do
       with_accessory(name) do
         stop(name)
         start(name)
@@ -107,8 +107,9 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
     if name == "all"
       KAMAL.accessory_names.each { |accessory_name| details(accessory_name) }
     else
+      type = "Accessory #{name}"
       with_accessory(name) do |accessory, hosts|
-        on(hosts) { puts capture_with_info(*accessory.info) }
+        on(hosts) { puts_by_host host, capture_with_info(*accessory.info), type: type }
       end
     end
   end
@@ -148,23 +149,25 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
   option :since, aliases: "-s", desc: "Show logs since timestamp (e.g. 2013-01-02T13:23:37Z) or relative (e.g. 42m for 42 minutes)"
   option :lines, type: :numeric, aliases: "-n", desc: "Number of log lines to pull from each server"
   option :grep, aliases: "-g", desc: "Show lines with grep match only (use this to fetch specific requests by id)"
+  option :grep_options, aliases: "-o", desc: "Additional options supplied to grep"
   option :follow, aliases: "-f", desc: "Follow logs on primary server (or specific host set by --hosts)"
   def logs(name)
     with_accessory(name) do |accessory, hosts|
       grep = options[:grep]
+      grep_options = options[:grep_options]
 
       if options[:follow]
         run_locally do
           info "Following logs on #{hosts}..."
-          info accessory.follow_logs(grep: grep)
+          info accessory.follow_logs(grep: grep, grep_options: grep_options)
-          exec accessory.follow_logs(grep: grep)
+          exec accessory.follow_logs(grep: grep, grep_options: grep_options)
         end
       else
         since = options[:since]
         lines = options[:lines].presence || ((since || grep) ? nil : 100) # Default to 100 lines if since or grep isn't set
 
         on(hosts) do
-          puts capture_with_info(*accessory.logs(since: since, lines: lines, grep: grep))
+          puts capture_with_info(*accessory.logs(since: since, lines: lines, grep: grep, grep_options: grep_options))
         end
       end
     end
@@ -173,17 +176,12 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
   desc "remove [NAME]", "Remove accessory container, image and data directory from host (use NAME=all to remove all accessories)"
   option :confirmed, aliases: "-y", type: :boolean, default: false, desc: "Proceed without confirmation question"
   def remove(name)
-    mutating do
+    confirming "This will remove all containers, images and data directories for #{name}. Are you sure?" do
-      if name == "all"
+      with_lock do
-        KAMAL.accessory_names.each { |accessory_name| remove(accessory_name) }
+        if name == "all"
-      else
+          KAMAL.accessory_names.each { |accessory_name| remove_accessory(accessory_name) }
-        if options[:confirmed] || ask("This will remove all containers, images and data directories for #{name}. Are you sure?", limited_to: %w( y N ), default: "N") == "y"
+        else
-          with_accessory(name) do
+          remove_accessory(name)
-            stop(name)
-            remove_container(name)
-            remove_image(name)
-            remove_service_directory(name)
-          end
         end
       end
     end
@@ -191,7 +189,7 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
 
   desc "remove_container [NAME]", "Remove accessory container from host", hide: true
   def remove_container(name)
-    mutating do
+    with_lock do
       with_accessory(name) do |accessory, hosts|
         on(hosts) do
           execute *KAMAL.auditor.record("Remove #{name} accessory container"), verbosity: :debug
@@ -203,7 +201,7 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
 
   desc "remove_image [NAME]", "Remove accessory image from host", hide: true
   def remove_image(name)
-    mutating do
+    with_lock do
       with_accessory(name) do |accessory, hosts|
         on(hosts) do
           execute *KAMAL.auditor.record("Removed #{name} accessory image"), verbosity: :debug
@@ -215,7 +213,7 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
 
   desc "remove_service_directory [NAME]", "Remove accessory directory used for uploaded files and data directories from host", hide: true
   def remove_service_directory(name)
-    mutating do
+    with_lock do
       with_accessory(name) do |accessory, hosts|
         on(hosts) do
           execute *accessory.remove_service_directory
@@ -224,9 +222,29 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
     end
   end
 
+  desc "downgrade", "Downgrade accessories from Kamal 2 to 1.9"
+  option :rolling, type: :boolean, default: false, desc: "Upgrade one host at a time"
+  option :confirmed, aliases: "-y", type: :boolean, default: false, desc: "Proceed without confirmation question"
+  def downgrade(name)
+    confirming "This will restart all accessories" do
+      with_lock do
+        host_groups = options[:rolling] ? KAMAL.accessory_hosts : [ KAMAL.accessory_hosts ]
+        host_groups.each do |hosts|
+          host_list = Array(hosts).join(",")
+          KAMAL.with_specific_hosts(hosts) do
+            say "Downgrading #{name} accessories on #{host_list}...", :magenta
+            reboot name
+            say "Downgraded #{name} accessories on #{host_list}...", :magenta
+          end
+        end
+      end
+    end
+  end
+
   private
     def with_accessory(name)
-      if accessory = KAMAL.accessory(name)
+      if KAMAL.config.accessory(name)
+        accessory = KAMAL.accessory(name)
         yield accessory, accessory_hosts(accessory)
       else
         error_on_missing_accessory(name)
@@ -248,4 +266,13 @@ class Kamal::Cli::Accessory < Kamal::Cli::Base
         accessory.hosts
       end
     end
+
+    def remove_accessory(name)
+      with_accessory(name) do
+        stop(name)
+        remove_container(name)
+        remove_image(name)
+        remove_service_directory(name)
+      end
+    end
 end

lib/kamal/cli/app.rb

@@ -1,78 +1,45 @@
 class Kamal::Cli::App < Kamal::Cli::Base
   desc "boot", "Boot app on servers (or reboot app if already running)"
   def boot
-    mutating do
+    with_lock do
-      hold_lock_on_error do
+      say "Get most recent version available as an image...", :magenta unless options[:version]
-        say "Get most recent version available as an image...", :magenta unless options[:version]
+      using_version(version_or_latest) do |version|
-        using_version(version_or_latest) do |version|
+        say "Start container with version #{version} using a #{KAMAL.config.readiness_delay}s readiness delay (or reboot if already running)...", :magenta
-          say "Start container with version #{version} using a #{KAMAL.config.readiness_delay}s readiness delay (or reboot if already running)...", :magenta
 
-          on(KAMAL.hosts) do
+        # Assets are prepared in a separate step to ensure they are on all hosts before booting
-            execute *KAMAL.auditor.record("Tagging #{KAMAL.config.absolute_image} as the latest image"), verbosity: :debug
+        on(KAMAL.hosts) do
-            execute *KAMAL.app.tag_current_image_as_latest
+          KAMAL.roles_on(host).each do |role|
-
+            Kamal::Cli::App::PrepareAssets.new(host, role, self).run
-            KAMAL.roles_on(host).each do |role|
-              app = KAMAL.app(role: role)
-
-              if role.assets?
-                execute *app.extract_assets
-                old_version = capture_with_info(*app.current_running_version, raise_on_non_zero_exit: false).strip
-                execute *app.sync_asset_volumes(old_version: old_version)
-              end
-            end
           end
+        end
 
-          on(KAMAL.hosts, **KAMAL.boot_strategy) do |host|
+        # Primary hosts and roles are returned first, so they can open the barrier
-            KAMAL.roles_on(host).each do |role|
+        barrier = Kamal::Cli::Healthcheck::Barrier.new
-              app = KAMAL.app(role: role)
-              auditor = KAMAL.auditor(role: role)
 
-              if capture_with_info(*app.container_id_for_version(version), raise_on_non_zero_exit: false).present?
+        on(KAMAL.hosts, **KAMAL.boot_strategy) do |host|
-                tmp_version = "#{version}_replaced_#{SecureRandom.hex(8)}"
+          KAMAL.roles_on(host).each do |role|
-                info "Renaming container #{version} to #{tmp_version} as already deployed on #{host}"
+            Kamal::Cli::App::Boot.new(host, role, self, version, barrier).run
-                execute *auditor.record("Renaming container #{version} to #{tmp_version}"), verbosity: :debug
-                execute *app.rename_container(version: version, new_version: tmp_version)
-              end
-
-              old_version = capture_with_info(*app.current_running_version, raise_on_non_zero_exit: false).strip
-
-              execute *app.tie_cord(role.cord_host_file) if role.uses_cord?
-
-              execute *auditor.record("Booted app version #{version}"), verbosity: :debug
-
-              execute *app.run(hostname: "#{host}-#{SecureRandom.hex(6)}")
-
-              Kamal::Cli::Healthcheck::Poller.wait_for_healthy(pause_after_ready: true) { capture_with_info(*app.status(version: version)) }
-
-              if old_version.present?
-                if role.uses_cord?
-                  cord = capture_with_info(*app.cord(version: old_version), raise_on_non_zero_exit: false).strip
-                  if cord.present?
-                    execute *app.cut_cord(cord)
-                    Kamal::Cli::Healthcheck::Poller.wait_for_unhealthy(pause_after_ready: true) { capture_with_info(*app.status(version: old_version)) }
-                  end
-                end
-
-                execute *app.stop(version: old_version), raise_on_non_zero_exit: false
-
-                execute *app.clean_up_assets if role.assets?
-              end
-            end
           end
         end
+
+        # Tag once the app booted on all hosts
+        on(KAMAL.hosts) do |host|
+          execute *KAMAL.auditor.record("Tagging #{KAMAL.config.absolute_image} as the latest image"), verbosity: :debug
+          execute *KAMAL.app.tag_latest_image
+        end
       end
     end
   end
 
   desc "start", "Start existing app container on servers"
   def start
-    mutating do
+    with_lock do
       on(KAMAL.hosts) do |host|
         roles = KAMAL.roles_on(host)
 
         roles.each do |role|
           execute *KAMAL.auditor.record("Started app version #{KAMAL.config.version}"), verbosity: :debug
-          execute *KAMAL.app(role: role).start, raise_on_non_zero_exit: false
+          execute *KAMAL.app(role: role, host: host).start, raise_on_non_zero_exit: false
         end
       end
     end
@@ -80,13 +47,13 @@ class Kamal::Cli::App < Kamal::Cli::Base
 
   desc "stop", "Stop app container on servers"
   def stop
-    mutating do
+    with_lock do
       on(KAMAL.hosts) do |host|
         roles = KAMAL.roles_on(host)
 
         roles.each do |role|
           execute *KAMAL.auditor.record("Stopped app", role: role), verbosity: :debug
-          execute *KAMAL.app(role: role).stop, raise_on_non_zero_exit: false
+          execute *KAMAL.app(role: role, host: host).stop, raise_on_non_zero_exit: false
         end
       end
     end
@@ -99,21 +66,23 @@ class Kamal::Cli::App < Kamal::Cli::Base
       roles = KAMAL.roles_on(host)
 
       roles.each do |role|
-        puts_by_host host, capture_with_info(*KAMAL.app(role: role).info)
+        puts_by_host host, capture_with_info(*KAMAL.app(role: role, host: host).info)
       end
     end
   end
 
-  desc "exec [CMD]", "Execute a custom command on servers (use --help to show options)"
+  desc "exec [CMD]", "Execute a custom command on servers within the app container (use --help to show options)"
   option :interactive, aliases: "-i", type: :boolean, default: false, desc: "Execute command over ssh for an interactive shell (use for console/bash)"
   option :reuse, type: :boolean, default: false, desc: "Reuse currently running container instead of starting a new one"
+  option :env, aliases: "-e", type: :hash, desc: "Set environment variables for the command"
   def exec(cmd)
+    env = options[:env]
     case
     when options[:interactive] && options[:reuse]
       say "Get current version of running container...", :magenta unless options[:version]
       using_version(options[:version] || current_running_version) do |version|
         say "Launching interactive command with version #{version} via SSH from existing container on #{KAMAL.primary_host}...", :magenta
-        run_locally { exec KAMAL.app(role: KAMAL.primary_role).execute_in_existing_container_over_ssh(cmd, host: KAMAL.primary_host) }
+        run_locally { exec KAMAL.app(role: KAMAL.primary_role, host: KAMAL.primary_host).execute_in_existing_container_over_ssh(cmd, env: env) }
       end
 
     when options[:interactive]
@@ -121,7 +90,7 @@ class Kamal::Cli::App < Kamal::Cli::Base
       using_version(version_or_latest) do |version|
         say "Launching interactive command with version #{version} via SSH from new container on #{KAMAL.primary_host}...", :magenta
         run_locally do
-          exec KAMAL.app(role: KAMAL.primary_role).execute_in_new_container_over_ssh(cmd, host: KAMAL.primary_host)
+          exec KAMAL.app(role: KAMAL.primary_role, host: KAMAL.primary_host).execute_in_new_container_over_ssh(cmd, env: env)
         end
       end
 
@@ -135,7 +104,7 @@ class Kamal::Cli::App < Kamal::Cli::Base
 
           roles.each do |role|
             execute *KAMAL.auditor.record("Executed cmd '#{cmd}' on app version #{version}", role: role), verbosity: :debug
-            puts_by_host host, capture_with_info(*KAMAL.app(role: role).execute_in_existing_container(cmd))
+            puts_by_host host, capture_with_info(*KAMAL.app(role: role, host: host).execute_in_existing_container(cmd, env: env))
           end
         end
       end
@@ -149,7 +118,7 @@ class Kamal::Cli::App < Kamal::Cli::Base
 
           roles.each do |role|
             execute *KAMAL.auditor.record("Executed cmd '#{cmd}' on app version #{version}"), verbosity: :debug
-            puts_by_host host, capture_with_info(*KAMAL.app(role: role).execute_in_new_container(cmd))
+            puts_by_host host, capture_with_info(*KAMAL.app(role: role, host: host).execute_in_new_container(cmd, env: env))
           end
         end
       end
@@ -164,19 +133,21 @@ class Kamal::Cli::App < Kamal::Cli::Base
   desc "stale_containers", "Detect app stale containers"
   option :stop, aliases: "-s", type: :boolean, default: false, desc: "Stop the stale containers found"
   def stale_containers
-    mutating do
+    stop = options[:stop]
-      stop = options[:stop]
-
-      cli = self
 
+    with_lock_if_stopping do
       on(KAMAL.hosts) do |host|
         roles = KAMAL.roles_on(host)
 
         roles.each do |role|
-          cli.send(:stale_versions, host: host, role: role).each do |version|
+          app = KAMAL.app(role: role, host: host)
+          versions = capture_with_info(*app.list_versions, raise_on_non_zero_exit: false).split("\n")
+          versions -= [ capture_with_info(*app.current_running_version, raise_on_non_zero_exit: false).strip ]
+
+          versions.each do |version|
             if stop
               puts_by_host host, "Stopping stale container for role #{role} with version #{version}"
-              execute *KAMAL.app(role: role).stop(version: version), raise_on_non_zero_exit: false
+              execute *app.stop(version: version), raise_on_non_zero_exit: false
             else
               puts_by_host host,  "Detected stale container for role #{role} with version #{version} (use `kamal app stale_containers --stop` to stop)"
             end
@@ -195,23 +166,27 @@ class Kamal::Cli::App < Kamal::Cli::Base
   option :since, aliases: "-s", desc: "Show lines since timestamp (e.g. 2013-01-02T13:23:37Z) or relative (e.g. 42m for 42 minutes)"
   option :lines, type: :numeric, aliases: "-n", desc: "Number of lines to show from each server"
   option :grep, aliases: "-g", desc: "Show lines with grep match only (use this to fetch specific requests by id)"
+  option :grep_options, aliases: "-o", desc: "Additional options supplied to grep"
   option :follow, aliases: "-f", desc: "Follow log on primary server (or specific host set by --hosts)"
   def logs
     # FIXME: Catch when app containers aren't running
 
     grep = options[:grep]
+    grep_options = options[:grep_options]
     since = options[:since]
+
     if options[:follow]
       lines = options[:lines].presence || ((since || grep) ? nil : 10) # Default to 10 lines if since or grep isn't set
 
       run_locally do
         info "Following logs on #{KAMAL.primary_host}..."
 
-        KAMAL.specific_roles ||= ["web"]
+        KAMAL.specific_roles ||= [ "web" ]
         role = KAMAL.roles_on(KAMAL.primary_host).first
 
-        info KAMAL.app(role: role).follow_logs(host: KAMAL.primary_host, lines: lines, grep: grep)
+        app = KAMAL.app(role: role, host: host)
-        exec KAMAL.app(role: role).follow_logs(host: KAMAL.primary_host, lines: lines, grep: grep)
+        info app.follow_logs(host: KAMAL.primary_host, lines: lines, grep: grep, grep_options: grep_options)
+        exec app.follow_logs(host: KAMAL.primary_host, lines: lines, grep: grep, grep_options: grep_options)
       end
     else
       lines = options[:lines].presence || ((since || grep) ? nil : 100) # Default to 100 lines if since or grep isn't set
@@ -221,7 +196,7 @@ class Kamal::Cli::App < Kamal::Cli::Base
 
         roles.each do |role|
           begin
-            puts_by_host host, capture_with_info(*KAMAL.app(role: role).logs(since: since, lines: lines, grep: grep))
+            puts_by_host host, capture_with_info(*KAMAL.app(role: role, host: host).logs(since: since, lines: lines, grep: grep, grep_options: grep_options))
           rescue SSHKit::Command::Failed
             puts_by_host host, "Nothing found"
           end
@@ -232,7 +207,7 @@ class Kamal::Cli::App < Kamal::Cli::Base
 
   desc "remove", "Remove app containers and images from servers"
   def remove
-    mutating do
+    with_lock do
       stop
       remove_containers
       remove_images
@@ -241,13 +216,13 @@ class Kamal::Cli::App < Kamal::Cli::Base
 
   desc "remove_container [VERSION]", "Remove app container with given version from servers", hide: true
   def remove_container(version)
-    mutating do
+    with_lock do
       on(KAMAL.hosts) do |host|
         roles = KAMAL.roles_on(host)
 
         roles.each do |role|
           execute *KAMAL.auditor.record("Removed app container with version #{version}", role: role), verbosity: :debug
-          execute *KAMAL.app(role: role).remove_container(version: version)
+          execute *KAMAL.app(role: role, host: host).remove_container(version: version)
         end
       end
     end
@@ -255,13 +230,13 @@ class Kamal::Cli::App < Kamal::Cli::Base
 
   desc "remove_containers", "Remove all app containers from servers", hide: true
   def remove_containers
-    mutating do
+    with_lock do
       on(KAMAL.hosts) do |host|
         roles = KAMAL.roles_on(host)
 
         roles.each do |role|
           execute *KAMAL.auditor.record("Removed all app containers", role: role), verbosity: :debug
-          execute *KAMAL.app(role: role).remove_containers
+          execute *KAMAL.app(role: role, host: host).remove_containers
         end
       end
     end
@@ -269,7 +244,7 @@ class Kamal::Cli::App < Kamal::Cli::Base
 
   desc "remove_images", "Remove all app images from servers", hide: true
   def remove_images
-    mutating do
+    with_lock do
       on(KAMAL.hosts) do
         execute *KAMAL.auditor.record("Removed all app images"), verbosity: :debug
         execute *KAMAL.app.remove_images
@@ -281,7 +256,7 @@ class Kamal::Cli::App < Kamal::Cli::Base
   def version
     on(KAMAL.hosts) do |host|
       role = KAMAL.roles_on(host).first
-      puts_by_host host, capture_with_info(*KAMAL.app(role: role).current_running_version).strip
+      puts_by_host host, capture_with_info(*KAMAL.app(role: role, host: host).current_running_version).strip
     end
   end
 
@@ -304,23 +279,20 @@ class Kamal::Cli::App < Kamal::Cli::Base
       version = nil
       on(host) do
         role = KAMAL.roles_on(host).first
-        version = capture_with_info(*KAMAL.app(role: role).current_running_version).strip
+        version = capture_with_info(*KAMAL.app(role: role, host: host).current_running_version).strip
       end
       version.presence
     end
 
-    def stale_versions(host:, role:)
+    def version_or_latest
-      versions = nil
+      options[:version] || KAMAL.config.latest_tag
-      on(host) do
-        versions = \
-          capture_with_info(*KAMAL.app(role: role).list_versions, raise_on_non_zero_exit: false)
-          .split("\n")
-          .drop(1)
-      end
-      versions
     end
 
-    def version_or_latest
+    def with_lock_if_stopping
-      options[:version] || "latest"
+      if options[:stop]
+        with_lock { yield }
+      else
+        yield
+      end
     end
 end

lib/kamal/cli/app/boot.rb

@@ -0,0 +1,119 @@
+class Kamal::Cli::App::Boot
+  attr_reader :host, :role, :version, :barrier, :sshkit
+  delegate :execute, :capture_with_info, :capture_with_pretty_json, :info, :error, to: :sshkit
+  delegate :uses_cord?, :assets?, :running_traefik?, to: :role
+
+  def initialize(host, role, sshkit, version, barrier)
+    @host = host
+    @role = role
+    @version = version
+    @barrier = barrier
+    @sshkit = sshkit
+  end
+
+  def run
+    old_version = old_version_renamed_if_clashing
+
+    wait_at_barrier if queuer?
+
+    begin
+      start_new_version
+    rescue => e
+      close_barrier if gatekeeper?
+      stop_new_version
+      raise
+    end
+
+    release_barrier if gatekeeper?
+
+    if old_version
+      stop_old_version(old_version)
+    end
+  end
+
+  private
+    def old_version_renamed_if_clashing
+      if capture_with_info(*app.container_id_for_version(version), raise_on_non_zero_exit: false).present?
+        renamed_version = "#{version}_replaced_#{SecureRandom.hex(8)}"
+        info "Renaming container #{version} to #{renamed_version} as already deployed on #{host}"
+        audit("Renaming container #{version} to #{renamed_version}")
+        execute *app.rename_container(version: version, new_version: renamed_version)
+      end
+
+      capture_with_info(*app.current_running_version, raise_on_non_zero_exit: false).strip.presence
+    end
+
+    def start_new_version
+      audit "Booted app version #{version}"
+
+      execute *app.tie_cord(role.cord_host_file) if uses_cord?
+      hostname = "#{host.to_s[0...51].gsub(/\.+$/, '')}-#{SecureRandom.hex(6)}"
+      execute *app.run(hostname: hostname)
+      Kamal::Cli::Healthcheck::Poller.wait_for_healthy(pause_after_ready: true) { capture_with_info(*app.status(version: version)) }
+    end
+
+    def stop_new_version
+      execute *app.stop(version: version), raise_on_non_zero_exit: false
+    end
+
+    def stop_old_version(version)
+      if uses_cord?
+        cord = capture_with_info(*app.cord(version: version), raise_on_non_zero_exit: false).strip
+        if cord.present?
+          execute *app.cut_cord(cord)
+          Kamal::Cli::Healthcheck::Poller.wait_for_unhealthy(pause_after_ready: true) { capture_with_info(*app.status(version: version)) }
+        end
+      end
+
+      execute *app.stop(version: version), raise_on_non_zero_exit: false
+
+      execute *app.clean_up_assets if assets?
+    end
+
+    def release_barrier
+      if barrier.open
+        info "First #{KAMAL.primary_role} container is healthy on #{host}, booting any other roles"
+      end
+    end
+
+    def wait_at_barrier
+      info "Waiting for the first healthy #{KAMAL.primary_role} container before booting #{role} on #{host}..."
+      barrier.wait
+      info "First #{KAMAL.primary_role} container is healthy, booting #{role} on #{host}..."
+    rescue Kamal::Cli::Healthcheck::Error
+      info "First #{KAMAL.primary_role} container is unhealthy, not booting #{role} on #{host}"
+      raise
+    end
+
+    def close_barrier
+      if barrier.close
+        info "First #{KAMAL.primary_role} container is unhealthy on #{host}, not booting any other roles"
+        error capture_with_info(*app.logs(version: version))
+        error capture_with_info(*app.container_health_log(version: version))
+      end
+    end
+
+    def barrier_role?
+      role == KAMAL.primary_role
+    end
+
+    def app
+      @app ||= KAMAL.app(role: role, host: host)
+    end
+
+    def auditor
+      @auditor = KAMAL.auditor(role: role)
+    end
+
+    def audit(message)
+      execute *auditor.record(message), verbosity: :debug
+    end
+
+    def gatekeeper?
+      barrier && barrier_role?
+    end
+
+    def queuer?
+      barrier && !barrier_role?
+    end
+end

lib/kamal/cli/app/prepare_assets.rb

@@ -0,0 +1,24 @@
+class Kamal::Cli::App::PrepareAssets
+  attr_reader :host, :role, :sshkit
+  delegate :execute, :capture_with_info, :info, to: :sshkit
+  delegate :assets?, to: :role
+
+  def initialize(host, role, sshkit)
+    @host = host
+    @role = role
+    @sshkit = sshkit
+  end
+
+  def run
+    if assets?
+      execute *app.extract_assets
+      old_version = capture_with_info(*app.current_running_version, raise_on_non_zero_exit: false).strip
+      execute *app.sync_asset_volumes(old_version: old_version)
+    end
+  end
+
+  private
+    def app
+      @app ||= KAMAL.app(role: role, host: host)
+    end
+end

lib/kamal/cli/base.rb

@@ -25,12 +25,17 @@ module Kamal::Cli
     def initialize(*)
       super
       @original_env = ENV.to_h.dup
-      load_envs
+      load_env
       initialize_commander(options_with_subcommand_class_options)
     end
 
     private
-      def load_envs
+      def reload_env
+        reset_env
+        load_env
+      end
+
+      def load_env
         if destination = options[:destination]
           Dotenv.load(".env.#{destination}", ".env")
         else
@@ -38,10 +43,27 @@ module Kamal::Cli
         end
       end
 
-      def reload_envs
+      def reset_env
+        replace_env @original_env
+      end
+
+      def replace_env(env)
         ENV.clear
-        ENV.update(@original_env)
+        ENV.update(env)
-        load_envs
+      end
+
+      def with_original_env
+        keeping_current_env do
+          reset_env
+          yield
+        end
+      end
+
+      def keeping_current_env
+        current_env = ENV.to_h.dup
+        yield
+      ensure
+        replace_env(current_env)
       end
 
       def options_with_subcommand_class_options
@@ -73,34 +95,43 @@ module Kamal::Cli
       def print_runtime
         started_at = Time.now
         yield
-        return Time.now - started_at
+        Time.now - started_at
       ensure
         runtime = Time.now - started_at
         puts "  Finished all in #{sprintf("%.1f seconds", runtime)}"
       end
 
-      def mutating
+      def with_lock
-        return yield if KAMAL.holding_lock?
+        if KAMAL.holding_lock?
-
-        run_hook "pre-connect"
-
-        ensure_run_directory
-
-        acquire_lock
-
-        begin
           yield
-        rescue
+        else
-          if KAMAL.hold_lock_on_error?
+          ensure_run_and_locks_directory
-            error "  \e[31mDeploy lock was not released\e[0m"
+
-          else
+          acquire_lock
-            release_lock
+
+          begin
+            yield
+          rescue
+            begin
+              release_lock
+            rescue => e
+              say "Error releasing the deploy lock: #{e.message}", :red
+            end
+            raise
           end
 
-          raise
+          release_lock
         end
+      end
 
-        release_lock
+      def confirming(question)
+        return yield if options[:confirmed]
+
+        if ask(question, limited_to: %w[ y N ], default: "N") == "y"
+          yield
+        else
+          say "Aborted", :red
+        end
       end
 
       def acquire_lock
@@ -131,29 +162,28 @@ module Kamal::Cli
         end
       end
 
-      def hold_lock_on_error
-        if KAMAL.hold_lock_on_error?
-          yield
-        else
-          KAMAL.hold_lock_on_error = true
-          yield
-          KAMAL.hold_lock_on_error = false
-        end
-      end
-
       def run_hook(hook, **extra_details)
         if !options[:skip_hooks] && KAMAL.hook.hook_exists?(hook)
           details = { hosts: KAMAL.hosts.join(","), command: command, subcommand: subcommand }
 
           say "Running the #{hook} hook...", :magenta
           run_locally do
-            KAMAL.with_verbosity(:debug) { execute *KAMAL.hook.run(hook, **details, **extra_details) }
+            execute *KAMAL.hook.run(hook, **details, **extra_details)
-          rescue SSHKit::Command::Failed
+          rescue SSHKit::Command::Failed => e
-            raise HookError.new("Hook `#{hook}` failed")
+            raise HookError.new("Hook `#{hook}` failed:\n#{e.message}")
           end
         end
       end
 
+      def on(*args, &block)
+        if !KAMAL.connected?
+          run_hook "pre-connect"
+          KAMAL.connected = true
+        end
+
+        super
+      end
+
       def command
         @kamal_command ||= begin
           invocation_class, invocation_commands = *first_invocation
@@ -176,10 +206,18 @@ module Kamal::Cli
         instance_variable_get("@_invocations").first
       end
 
-      def ensure_run_directory
+      def reset_invocation(cli_class)
+        instance_variable_get("@_invocations")[cli_class].pop
+      end
+
+      def ensure_run_and_locks_directory
         on(KAMAL.hosts) do
           execute(*KAMAL.server.ensure_run_directory)
         end
+
+        on(KAMAL.primary_host) do
+          execute(*KAMAL.lock.ensure_locks_directory)
+        end
       end
-    end
+  end
 end

lib/kamal/cli/build.rb

@@ -5,74 +5,87 @@ class Kamal::Cli::Build < Kamal::Cli::Base
 
   desc "deliver", "Build app and push app image to registry then pull image on servers"
   def deliver
-    mutating do
+    push
-      push
+    pull
-      pull
-    end
   end
 
   desc "push", "Build and push app image to registry"
   def push
-    mutating do
+    cli = self
-      cli = self
 
-      verify_local_dependencies
+    verify_local_dependencies
-      run_hook "pre-build"
+    run_hook "pre-build"
 
-      if (uncommitted_changes = Kamal::Git.uncommitted_changes).present?
+    uncommitted_changes = Kamal::Git.uncommitted_changes
-        say "The following paths have uncommitted changes:\n #{uncommitted_changes}", :yellow
+
+    if KAMAL.config.builder.git_clone?
+      if uncommitted_changes.present?
+        say "Building from a local git clone, so ignoring these uncommitted changes:\n #{uncommitted_changes}", :yellow
       end
 
       run_locally do
-        begin
+        Clone.new(self).prepare
-          KAMAL.with_verbosity(:debug) do
+      end
-            execute *KAMAL.builder.push
+    elsif uncommitted_changes.present?
-          end
+      say "Building with uncommitted changes:\n #{uncommitted_changes}", :yellow
-        rescue SSHKit::Command::Failed => e
+    end
-          if e.message =~ /(no builder)|(no such file or directory)/
-            error "Missing compatible builder, so creating a new one first"
 
-            if cli.create
+    # Get the command here to ensure the Dir.chdir doesn't interfere with it
-              KAMAL.with_verbosity(:debug) { execute *KAMAL.builder.push }
+    push = KAMAL.builder.push
-            end
+
-          else
+    run_locally do
-            raise
+      begin
-          end
+        context_hosts = capture_with_info(*KAMAL.builder.context_hosts).split("\n")
+
+        if context_hosts != KAMAL.builder.config_context_hosts
+          warn "Context hosts have changed, so re-creating builder, was: #{context_hosts.join(", ")}], now: #{KAMAL.builder.config_context_hosts.join(", ")}"
+          cli.remove
+          cli.create
         end
+      rescue SSHKit::Command::Failed => e
+        if e.message =~ /(context not found|no builder|does not exist)/
+          warn "Missing compatible builder, so creating a new one first"
+          cli.create
+        else
+          raise
+        end
+      end
+
+      KAMAL.with_verbosity(:debug) do
+        Dir.chdir(KAMAL.config.builder.build_directory) { execute *push }
       end
     end
   end
 
   desc "pull", "Pull app image from registry onto servers"
   def pull
-    mutating do
+    if (first_hosts = mirror_hosts).any?
-      on(KAMAL.hosts) do
+      #  Pull on a single host per mirror first to seed them
-        execute *KAMAL.auditor.record("Pulled image with version #{KAMAL.config.version}"), verbosity: :debug
+      say "Pulling image on #{first_hosts.join(", ")} to seed the #{"mirror".pluralize(first_hosts.count)}...", :magenta
-        execute *KAMAL.builder.clean, raise_on_non_zero_exit: false
+      pull_on_hosts(first_hosts)
-        execute *KAMAL.builder.pull
+      say "Pulling image on remaining hosts...", :magenta
-        execute *KAMAL.builder.validate_image
+      pull_on_hosts(KAMAL.hosts - first_hosts)
-      end
+    else
+      pull_on_hosts(KAMAL.hosts)
     end
   end
 
   desc "create", "Create a build setup"
   def create
-    mutating do
+    if (remote_host = KAMAL.config.builder.remote_host)
-      if (remote_host = KAMAL.config.builder.remote_host)
+      connect_to_remote_host(remote_host)
-        connect_to_remote_host(remote_host)
+    end
-      end
 
-      run_locally do
+    run_locally do
-        begin
+      begin
-          debug "Using builder: #{KAMAL.builder.name}"
+        debug "Using builder: #{KAMAL.builder.name}"
-          execute *KAMAL.builder.create
+        execute *KAMAL.builder.create
-        rescue SSHKit::Command::Failed => e
+      rescue SSHKit::Command::Failed => e
-          if e.message =~ /stderr=(.*)/
+        if e.message =~ /stderr=(.*)/
-            error "Couldn't create remote builder: #{$1}"
+          error "Couldn't create remote builder: #{$1}"
-            false
+          false
-          else
+        else
-            raise
+          raise
-          end
         end
       end
     end
@@ -80,11 +93,9 @@ class Kamal::Cli::Build < Kamal::Cli::Base
 
   desc "remove", "Remove build setup"
   def remove
-    mutating do
+    run_locally do
-      run_locally do
+      debug "Using builder: #{KAMAL.builder.name}"
-        debug "Using builder: #{KAMAL.builder.name}"
+      execute *KAMAL.builder.remove
-        execute *KAMAL.builder.remove
-      end
     end
   end
 
@@ -114,10 +125,37 @@ class Kamal::Cli::Build < Kamal::Cli::Base
     def connect_to_remote_host(remote_host)
       remote_uri = URI.parse(remote_host)
       if remote_uri.scheme == "ssh"
-        options = { user: remote_uri.user, port: remote_uri.port }.compact
+        host = SSHKit::Host.new(
-        on(remote_uri.host, options) do
+          hostname: remote_uri.host,
+          ssh_options: { user: remote_uri.user, port: remote_uri.port }.compact
+        )
+        on(host, options) do
           execute "true"
         end
       end
     end
+
+    def mirror_hosts
+      if KAMAL.hosts.many?
+        mirror_hosts = Concurrent::Hash.new
+        on(KAMAL.hosts) do |host|
+          first_mirror = capture_with_info(*KAMAL.builder.first_mirror).strip.presence
+          mirror_hosts[first_mirror] ||= host.to_s if first_mirror
+        rescue SSHKit::Command::Failed => e
+          raise unless e.message =~ /error calling index: reflect: slice index out of range/
+        end
+        mirror_hosts.values
+      else
+        []
+      end
+    end
+
+    def pull_on_hosts(hosts)
+      on(hosts) do
+        execute *KAMAL.auditor.record("Pulled image with version #{KAMAL.config.version}"), verbosity: :debug
+        execute *KAMAL.builder.clean, raise_on_non_zero_exit: false
+        execute *KAMAL.builder.pull
+        execute *KAMAL.builder.validate_image
+      end
+    end
 end

lib/kamal/cli/build/clone.rb

@@ -0,0 +1,61 @@
+require "uri"
+
+class Kamal::Cli::Build::Clone
+  attr_reader :sshkit
+  delegate :info, :error, :execute, :capture_with_info, to: :sshkit
+
+  def initialize(sshkit)
+    @sshkit = sshkit
+  end
+
+  def prepare
+    begin
+      clone_repo
+    rescue SSHKit::Command::Failed => e
+      if e.message =~ /already exists and is not an empty directory/
+        reset
+      else
+        raise Kamal::Cli::Build::BuildError, "Failed to clone repo: #{e.message}"
+      end
+    end
+
+    validate!
+  rescue Kamal::Cli::Build::BuildError => e
+    error "Error preparing clone: #{e.message}, deleting and retrying..."
+
+    FileUtils.rm_rf KAMAL.config.builder.clone_directory
+    clone_repo
+    validate!
+  end
+
+  private
+    def clone_repo
+      info "Cloning repo into build directory `#{KAMAL.config.builder.build_directory}`..."
+
+      FileUtils.mkdir_p KAMAL.config.builder.clone_directory
+      execute *KAMAL.builder.clone
+    end
+
+    def reset
+      info "Resetting local clone as `#{KAMAL.config.builder.build_directory}` already exists..."
+
+      KAMAL.builder.clone_reset_steps.each { |step| execute *step }
+    rescue SSHKit::Command::Failed => e
+      raise Kamal::Cli::Build::BuildError, "Failed to clone repo: #{e.message}"
+    end
+
+    def validate!
+      status = capture_with_info(*KAMAL.builder.clone_status).strip
+
+      unless status.empty?
+        raise Kamal::Cli::Build::BuildError, "Clone in #{KAMAL.config.builder.build_directory} is dirty, #{status}"
+      end
+
+      revision = capture_with_info(*KAMAL.builder.clone_revision).strip
+      if revision != Kamal::Git.revision
+        raise Kamal::Cli::Build::BuildError, "Clone in #{KAMAL.config.builder.build_directory} is not on the correct revision, expected `#{Kamal::Git.revision}` but got `#{revision}`"
+      end
+    rescue SSHKit::Command::Failed => e
+      raise Kamal::Cli::Build::BuildError, "Failed to validate clone: #{e.message}"
+    end
+end

lib/kamal/cli/env.rb

@@ -3,26 +3,26 @@ require "tempfile"
 class Kamal::Cli::Env < Kamal::Cli::Base
   desc "push", "Push the env file to the remote hosts"
   def push
-    mutating do
+    with_lock do
       on(KAMAL.hosts) do
         execute *KAMAL.auditor.record("Pushed env files"), verbosity: :debug
 
         KAMAL.roles_on(host).each do |role|
-          execute *KAMAL.app(role: role).make_env_directory
+          execute *KAMAL.app(role: role, host: host).make_env_directory
-          upload! StringIO.new(role.env_file), role.host_env_file_path, mode: 400
+          upload! role.env(host).secrets_io, role.env(host).secrets_file, mode: 400
         end
       end
 
       on(KAMAL.traefik_hosts) do
         execute *KAMAL.traefik.make_env_directory
-        upload! StringIO.new(KAMAL.traefik.env_file), KAMAL.traefik.host_env_file_path, mode: 400
+        upload! KAMAL.traefik.env.secrets_io, KAMAL.traefik.env.secrets_file, mode: 400
       end
 
       on(KAMAL.accessory_hosts) do
         KAMAL.accessories_on(host).each do |accessory|
           accessory_config = KAMAL.config.accessory(accessory)
           execute *KAMAL.accessory(accessory).make_env_directory
-          upload! StringIO.new(accessory_config.env_file), accessory_config.host_env_file_path, mode: 400
+          upload! accessory_config.env.secrets_io, accessory_config.env.secrets_file, mode: 400
         end
       end
     end
@@ -30,12 +30,12 @@ class Kamal::Cli::Env < Kamal::Cli::Base
 
   desc "delete", "Delete the env file from the remote hosts"
   def delete
-    mutating do
+    with_lock do
       on(KAMAL.hosts) do
         execute *KAMAL.auditor.record("Deleted env files"), verbosity: :debug
 
         KAMAL.roles_on(host).each do |role|
-          execute *KAMAL.app(role: role).remove_env_file
+          execute *KAMAL.app(role: role, host: host).remove_env_file
         end
       end
 

lib/kamal/cli/healthcheck.rb

@@ -1,21 +0,0 @@
-class Kamal::Cli::Healthcheck < Kamal::Cli::Base
-  default_command :perform
-
-  desc "perform", "Health check current app version"
-  def perform
-    raise "The primary host is not configured to run Traefik" unless KAMAL.config.role(KAMAL.config.primary_role).running_traefik?
-    on(KAMAL.primary_host) do
-      begin
-        execute *KAMAL.healthcheck.run
-        Poller.wait_for_healthy { capture_with_info(*KAMAL.healthcheck.status) }
-      rescue Poller::HealthcheckError => e
-        error capture_with_info(*KAMAL.healthcheck.logs)
-        error capture_with_pretty_json(*KAMAL.healthcheck.container_health_log)
-        raise
-      ensure
-        execute *KAMAL.healthcheck.stop, raise_on_non_zero_exit: false
-        execute *KAMAL.healthcheck.remove, raise_on_non_zero_exit: false
-      end
-    end
-  end
-end

lib/kamal/cli/healthcheck/barrier.rb

@@ -0,0 +1,31 @@
+class Kamal::Cli::Healthcheck::Barrier
+  def initialize
+    @ivar = Concurrent::IVar.new
+  end
+
+  def close
+    set(false)
+  end
+
+  def open
+    set(true)
+  end
+
+  def wait
+    unless opened?
+      raise Kamal::Cli::Healthcheck::Error.new("Halted at barrier")
+    end
+  end
+
+  private
+    def opened?
+      @ivar.value
+    end
+
+    def set(value)
+      @ivar.set(value)
+      true
+    rescue Concurrent::MultipleAssignmentError
+      false
+    end
+end

lib/kamal/cli/healthcheck/error.rb

@@ -0,0 +1,2 @@
+class Kamal::Cli::Healthcheck::Error < StandardError
+end

lib/kamal/cli/healthcheck/poller.rb

@@ -3,11 +3,10 @@ module Kamal::Cli::Healthcheck::Poller
 
   TRAEFIK_UPDATE_DELAY = 5
 
-  class HealthcheckError < StandardError; end
 
   def wait_for_healthy(pause_after_ready: false, &block)
     attempt = 1
-    max_attempts = KAMAL.config.healthcheck["max_attempts"]
+    max_attempts = KAMAL.config.healthcheck.max_attempts
 
     begin
       case status = block.call
@@ -16,9 +15,9 @@ module Kamal::Cli::Healthcheck::Poller
       when "running" # No health check configured
         sleep KAMAL.config.readiness_delay if pause_after_ready
       else
-        raise HealthcheckError, "container not ready (#{status})"
+        raise Kamal::Cli::Healthcheck::Error, "container not ready (#{status})"
       end
-    rescue HealthcheckError => e
+    rescue Kamal::Cli::Healthcheck::Error => e
       if attempt <= max_attempts
         info "#{e.message}, retrying in #{attempt}s (attempt #{attempt}/#{max_attempts})..."
         sleep attempt
@@ -34,16 +33,16 @@ module Kamal::Cli::Healthcheck::Poller
 
   def wait_for_unhealthy(pause_after_ready: false, &block)
     attempt = 1
-    max_attempts = KAMAL.config.healthcheck["max_attempts"]
+    max_attempts = KAMAL.config.healthcheck.max_attempts
 
     begin
       case status = block.call
       when "unhealthy"
         sleep TRAEFIK_UPDATE_DELAY if pause_after_ready
       else
-        raise HealthcheckError, "container not unhealthy (#{status})"
+        raise Kamal::Cli::Healthcheck::Error, "container not unhealthy (#{status})"
       end
-    rescue HealthcheckError => e
+    rescue Kamal::Cli::Healthcheck::Error => e
       if attempt <= max_attempts
         info "#{e.message}, retrying in #{attempt}s (attempt #{attempt}/#{max_attempts})..."
         sleep attempt

lib/kamal/cli/main.rb

@@ -3,13 +3,14 @@ class Kamal::Cli::Main < Kamal::Cli::Base
   option :skip_push, aliases: "-P", type: :boolean, default: false, desc: "Skip image build and push"
   def setup
     print_runtime do
-      mutating do
+      with_lock do
         invoke_options = deploy_options
 
         say "Ensure Docker is installed...", :magenta
         invoke "kamal:cli:server:bootstrap", [], invoke_options
 
-        say "Push env files...", :magenta
+        say "Evaluate and push env files...", :magenta
+        invoke "kamal:cli:main:envify", [], invoke_options
         invoke "kamal:cli:env:push", [], invoke_options
 
         invoke "kamal:cli:accessory:boot", [ "all" ], invoke_options
@@ -22,30 +23,25 @@ class Kamal::Cli::Main < Kamal::Cli::Base
   option :skip_push, aliases: "-P", type: :boolean, default: false, desc: "Skip image build and push"
   def deploy
     runtime = print_runtime do
-      mutating do
+      invoke_options = deploy_options
-        invoke_options = deploy_options
 
-        say "Log into image registry...", :magenta
+      say "Log into image registry...", :magenta
-        invoke "kamal:cli:registry:login", [], invoke_options
+      invoke "kamal:cli:registry:login", [], invoke_options.merge(skip_local: options[:skip_push])
 
-        if options[:skip_push]
+      if options[:skip_push]
-          say "Pull app image...", :magenta
+        say "Pull app image...", :magenta
-          invoke "kamal:cli:build:pull", [], invoke_options
+        invoke "kamal:cli:build:pull", [], invoke_options
-        else
+      else
-          say "Build and push app image...", :magenta
+        say "Build and push app image...", :magenta
-          invoke "kamal:cli:build:deliver", [], invoke_options
+        invoke "kamal:cli:build:deliver", [], invoke_options
-        end
+      end
 
+      with_lock do
         run_hook "pre-deploy"
 
         say "Ensure Traefik is running...", :magenta
         invoke "kamal:cli:traefik:boot", [], invoke_options
 
-        if KAMAL.config.role(KAMAL.config.primary_role).running_traefik?
-          say "Ensure app can pass healthcheck...", :magenta
-          invoke "kamal:cli:healthcheck:perform", [], invoke_options
-        end
-
         say "Detect stale containers...", :magenta
         invoke "kamal:cli:app:stale_containers", [], invoke_options.merge(stop: true)
 
@@ -63,22 +59,19 @@ class Kamal::Cli::Main < Kamal::Cli::Base
   option :skip_push, aliases: "-P", type: :boolean, default: false, desc: "Skip image build and push"
   def redeploy
     runtime = print_runtime do
-      mutating do
+      invoke_options = deploy_options
-        invoke_options = deploy_options
 
-        if options[:skip_push]
+      if options[:skip_push]
-          say "Pull app image...", :magenta
+        say "Pull app image...", :magenta
-          invoke "kamal:cli:build:pull", [], invoke_options
+        invoke "kamal:cli:build:pull", [], invoke_options
-        else
+      else
-          say "Build and push app image...", :magenta
+        say "Build and push app image...", :magenta
-          invoke "kamal:cli:build:deliver", [], invoke_options
+        invoke "kamal:cli:build:deliver", [], invoke_options
-        end
+      end
 
+      with_lock do
         run_hook "pre-deploy"
 
-        say "Ensure app can pass healthcheck...", :magenta
-        invoke "kamal:cli:healthcheck:perform", [], invoke_options
-
         say "Detect stale containers...", :magenta
         invoke "kamal:cli:app:stale_containers", [], invoke_options.merge(stop: true)
 
@@ -93,7 +86,7 @@ class Kamal::Cli::Main < Kamal::Cli::Base
   def rollback(version)
     rolled_back = false
     runtime = print_runtime do
-      mutating do
+      with_lock do
         invoke_options = deploy_options
 
         KAMAL.config.version = version
@@ -134,6 +127,18 @@ class Kamal::Cli::Main < Kamal::Cli::Base
     end
   end
 
+  desc "docs", "Show Kamal documentation for configuration setting"
+  def docs(section = nil)
+    case section
+    when NilClass
+      puts Kamal::Configuration.validation_doc
+    else
+      puts Kamal::Configuration.const_get(section.titlecase.to_sym).validation_doc
+    end
+  rescue NameError
+    puts "No documentation found for #{section}"
+  end
+
   desc "init", "Create config stub in config/deploy.yml and env stub in .env"
   option :bundle, type: :boolean, default: false, desc: "Add Kamal to the Gemfile and create a bin/kamal binstub"
   def init
@@ -185,23 +190,60 @@ class Kamal::Cli::Main < Kamal::Cli::Base
       env_path          = ".env"
     end
 
-    File.write(env_path, ERB.new(File.read(env_template_path), trim_mode: "-").result, perm: 0600)
+    if Pathname.new(File.expand_path(env_template_path)).exist?
+      # Ensure existing env doesn't pollute template evaluation
+      content = with_original_env { ERB.new(File.read(env_template_path), trim_mode: "-").result }
+      File.write(env_path, content, perm: 0600)
 
-    unless options[:skip_push]
+      unless options[:skip_push]
-      reload_envs
+        reload_env
-      invoke "kamal:cli:env:push", options
+        invoke "kamal:cli:env:push", options
+      end
+    else
+      puts "Skipping envify (no #{env_template_path} exist)"
     end
   end
 
   desc "remove", "Remove Traefik, app, accessories, and registry session from servers"
   option :confirmed, aliases: "-y", type: :boolean, default: false, desc: "Proceed without confirmation question"
   def remove
-    mutating do
+    confirming "This will remove all containers and images. Are you sure?" do
-      if options[:confirmed] || ask("This will remove all containers and images. Are you sure?", limited_to: %w( y N ), default: "N") == "y"
+      with_lock do
         invoke "kamal:cli:traefik:remove", [], options.without(:confirmed)
         invoke "kamal:cli:app:remove", [], options.without(:confirmed)
         invoke "kamal:cli:accessory:remove", [ "all" ], options
-        invoke "kamal:cli:registry:logout", [], options.without(:confirmed)
+        invoke "kamal:cli:registry:logout", [], options.without(:confirmed).merge(skip_local: true)
+      end
+    end
+  end
+
+  desc "downgrade", "Downgrade from Kamal 2 to 1.9"
+  option :confirmed, aliases: "-y", type: :boolean, default: false, desc: "Proceed without confirmation question"
+  option :rolling, type: :boolean, default: false, desc: "Downgrade one host at a time"
+  def downgrade
+    confirming "This will replace Traefik with kamal-proxy and restart all accessories" do
+      with_lock do
+        if options[:rolling]
+          (KAMAL.hosts | KAMAL.accessory_hosts).each do |host|
+            KAMAL.with_specific_hosts(host) do
+              say "Downgrading #{host}...", :magenta
+              if KAMAL.hosts.include?(host)
+                invoke "kamal:cli:traefik:downgrade", [], options.merge(confirmed: true, rolling: false)
+                reset_invocation(Kamal::Cli::Traefik)
+              end
+              if KAMAL.accessory_hosts.include?(host)
+                invoke "kamal:cli:accessory:downgrade", [ "all" ], options.merge(confirmed: true, rolling: false)
+                reset_invocation(Kamal::Cli::Accessory)
+              end
+              say "Downgraded #{host}", :magenta
+            end
+          end
+        else
+          say "Downgrading all hosts...", :magenta
+          invoke "kamal:cli:traefik:downgrade", [], options.merge(confirmed: true)
+          invoke "kamal:cli:accessory:downgrade", [ "all" ], options.merge(confirmed: true)
+          say "Downgraded all hosts", :magenta
+        end
       end
     end
   end
@@ -223,9 +265,6 @@ class Kamal::Cli::Main < Kamal::Cli::Base
   desc "env", "Manage environment files"
   subcommand "env", Kamal::Cli::Env
 
-  desc "healthcheck", "Healthcheck application"
-  subcommand "healthcheck", Kamal::Cli::Healthcheck
-
   desc "lock", "Manage the deploy lock"
   subcommand "lock", Kamal::Cli::Lock
 
@@ -246,11 +285,11 @@ class Kamal::Cli::Main < Kamal::Cli::Base
       begin
         on(KAMAL.hosts) do
           KAMAL.roles_on(host).each do |role|
-            container_id = capture_with_info(*KAMAL.app(role: role).container_id_for_version(version))
+            container_id = capture_with_info(*KAMAL.app(role: role, host: host).container_id_for_version(version))
             raise "Container not found" unless container_id.present?
           end
         end
-      rescue SSHKit::Runner::ExecuteError => e
+      rescue SSHKit::Runner::ExecuteError, SSHKit::Runner::MultipleExecuteError => e
         if e.message =~ /Container not found/
           say "Error looking for container version #{version}: #{e.message}"
           return false

lib/kamal/cli/prune.rb

@@ -1,7 +1,7 @@
 class Kamal::Cli::Prune < Kamal::Cli::Base
   desc "all", "Prune unused images and stopped containers"
   def all
-    mutating do
+    with_lock do
       containers
       images
     end
@@ -9,7 +9,7 @@ class Kamal::Cli::Prune < Kamal::Cli::Base
 
   desc "images", "Prune unused images"
   def images
-    mutating do
+    with_lock do
       on(KAMAL.hosts) do
         execute *KAMAL.auditor.record("Pruned images"), verbosity: :debug
         execute *KAMAL.prune.dangling_images
@@ -24,7 +24,7 @@ class Kamal::Cli::Prune < Kamal::Cli::Base
     retain = options.fetch(:retain, KAMAL.config.retain_containers)
     raise "retain must be at least 1" if retain < 1
 
-    mutating do
+    with_lock do
       on(KAMAL.hosts) do
         execute *KAMAL.auditor.record("Pruned containers"), verbosity: :debug
         execute *KAMAL.prune.app_containers(retain: retain)

lib/kamal/cli/registry.rb

@@ -1,18 +1,17 @@
 class Kamal::Cli::Registry < Kamal::Cli::Base
   desc "login", "Log in to registry locally and remotely"
+  option :skip_local, aliases: "-L", type: :boolean, default: false, desc: "Skip local login"
+  option :skip_remote, aliases: "-R", type: :boolean, default: false, desc: "Skip remote login"
   def login
-    run_locally    { execute *KAMAL.registry.login }
+    run_locally    { execute *KAMAL.registry.login } unless options[:skip_local]
-    on(KAMAL.hosts) { execute *KAMAL.registry.login }
+    on(KAMAL.hosts) { execute *KAMAL.registry.login } unless options[:skip_remote]
-  # FIXME: This rescue needed?
-  rescue ArgumentError => e
-    puts e.message
   end
 
-  desc "logout", "Log out of registry remotely"
+  desc "logout", "Log out of registry locally and remotely"
+  option :skip_local, aliases: "-L", type: :boolean, default: false, desc: "Skip local login"
+  option :skip_remote, aliases: "-R", type: :boolean, default: false, desc: "Skip remote login"
   def logout
-    on(KAMAL.hosts) { execute *KAMAL.registry.logout }
+    run_locally    { execute *KAMAL.registry.logout } unless options[:skip_local]
-  # FIXME: This rescue needed?
+    on(KAMAL.hosts) { execute *KAMAL.registry.logout } unless options[:skip_remote]
-  rescue ArgumentError => e
-    puts e.message
   end
 end

lib/kamal/cli/server.rb

@@ -1,25 +1,49 @@
 class Kamal::Cli::Server < Kamal::Cli::Base
+  desc "exec", "Run a custom command on the server (use --help to show options)"
+  option :interactive, type: :boolean, aliases: "-i", default: false, desc: "Run the command interactively (use for console/bash)"
+  def exec(cmd)
+    hosts = KAMAL.hosts | KAMAL.accessory_hosts
+
+    case
+    when options[:interactive]
+      host = KAMAL.primary_host
+
+      say "Running '#{cmd}' on #{host} interactively...", :magenta
+
+      run_locally { exec KAMAL.server.run_over_ssh(cmd, host: host) }
+    else
+      say "Running '#{cmd}' on #{hosts.join(', ')}...", :magenta
+
+      on(hosts) do |host|
+        execute *KAMAL.auditor.record("Executed cmd '#{cmd}' on #{host}"), verbosity: :debug
+        puts_by_host host, capture_with_info(cmd)
+      end
+    end
+  end
+
   desc "bootstrap", "Set up Docker to run Kamal apps"
   def bootstrap
-    missing = []
+    with_lock do
+      missing = []
 
-    on(KAMAL.hosts | KAMAL.accessory_hosts) do |host|
+      on(KAMAL.hosts | KAMAL.accessory_hosts) do |host|
-      unless execute(*KAMAL.docker.installed?, raise_on_non_zero_exit: false)
+        unless execute(*KAMAL.docker.installed?, raise_on_non_zero_exit: false)
-        if execute(*KAMAL.docker.superuser?, raise_on_non_zero_exit: false)
+          if execute(*KAMAL.docker.superuser?, raise_on_non_zero_exit: false)
-          info "Missing Docker on #{host}. Installing…"
+            info "Missing Docker on #{host}. Installing…"
-          execute *KAMAL.docker.install
+            execute *KAMAL.docker.install
-        else
+          else
-          missing << host
+            missing << host
+          end
         end
+
+        execute(*KAMAL.server.ensure_run_directory)
       end
 
-      execute(*KAMAL.server.ensure_run_directory)
+      if missing.any?
-    end
+        raise "Docker is not installed on #{missing.join(", ")} and can't be automatically installed without having root access and either `wget` or `curl`. Install Docker manually: https://docs.docker.com/engine/install/"
+      end
 
-    if missing.any?
+      run_hook "docker-setup"
-      raise "Docker is not installed on #{missing.join(", ")} and can't be automatically installed without having root access and either `wget` or `curl`. Install Docker manually: https://docs.docker.com/engine/install/"
     end
-
-    run_hook "docker-setup"
   end
 end

lib/kamal/cli/templates/deploy.yml

@@ -63,7 +63,7 @@ registry:
 #     directories:
 #       - data:/data
 
-# Configure custom arguments for Traefik
+# Configure custom arguments for Traefik. Be sure to reboot traefik when you modify it.
 # traefik:
 #   args:
 #     accesslog: true

lib/kamal/cli/templates/sample_hooks/docker-setup.sample

@@ -2,6 +2,12 @@
 
 # A sample docker-setup hook
 #
-# Sets up a Docker network which can then be used by the application’s containers
+# Sets up a Docker network on defined hosts which can then be used by the application’s containers
 
-ssh user@example.com docker network create kamal
+hosts = ENV["KAMAL_HOSTS"].split(",")
+
+hosts.each do |ip|
+  destination = "root@#{ip}"
+  puts "Creating a Docker network \"kamal\" on #{destination}"
+  `ssh #{destination} docker network create kamal`
+end

lib/kamal/cli/traefik.rb

@@ -1,7 +1,7 @@
 class Kamal::Cli::Traefik < Kamal::Cli::Base
   desc "boot", "Boot Traefik on servers"
   def boot
-    mutating do
+    with_lock do
       on(KAMAL.traefik_hosts) do
         execute *KAMAL.registry.login
         execute *KAMAL.traefik.start_or_run
@@ -11,27 +11,30 @@ class Kamal::Cli::Traefik < Kamal::Cli::Base
 
   desc "reboot", "Reboot Traefik on servers (stop container, remove container, start new container)"
   option :rolling, type: :boolean, default: false, desc: "Reboot traefik on hosts in sequence, rather than in parallel"
+  option :confirmed, aliases: "-y", type: :boolean, default: false, desc: "Proceed without confirmation question"
   def reboot
-    mutating do
+    confirming "This will cause a brief outage on each host. Are you sure?" do
-      host_groups = options[:rolling] ? KAMAL.traefik_hosts : [KAMAL.traefik_hosts]
+      with_lock do
-      host_groups.each do |hosts|
+        host_groups = options[:rolling] ? KAMAL.traefik_hosts : [ KAMAL.traefik_hosts ]
-        host_list = Array(hosts).join(",")
+        host_groups.each do |hosts|
-        run_hook "pre-traefik-reboot", hosts: host_list
+          host_list = Array(hosts).join(",")
-        on(hosts) do
+          run_hook "pre-traefik-reboot", hosts: host_list
-          execute *KAMAL.auditor.record("Rebooted traefik"), verbosity: :debug
+          on(hosts) do
-          execute *KAMAL.registry.login
+            execute *KAMAL.auditor.record("Rebooted traefik"), verbosity: :debug
-          execute *KAMAL.traefik.stop, raise_on_non_zero_exit: false
+            execute *KAMAL.registry.login
-          execute *KAMAL.traefik.remove_container
+            execute *KAMAL.traefik.stop, raise_on_non_zero_exit: false
-          execute *KAMAL.traefik.run
+            execute *KAMAL.traefik.remove_container
+            execute *KAMAL.traefik.run
+          end
+          run_hook "post-traefik-reboot", hosts: host_list
         end
-        run_hook "post-traefik-reboot", hosts: host_list
       end
     end
   end
 
   desc "start", "Start existing Traefik container on servers"
   def start
-    mutating do
+    with_lock do
       on(KAMAL.traefik_hosts) do
         execute *KAMAL.auditor.record("Started traefik"), verbosity: :debug
         execute *KAMAL.traefik.start
@@ -41,7 +44,7 @@ class Kamal::Cli::Traefik < Kamal::Cli::Base
 
   desc "stop", "Stop existing Traefik container on servers"
   def stop
-    mutating do
+    with_lock do
       on(KAMAL.traefik_hosts) do
         execute *KAMAL.auditor.record("Stopped traefik"), verbosity: :debug
         execute *KAMAL.traefik.stop, raise_on_non_zero_exit: false
@@ -51,7 +54,7 @@ class Kamal::Cli::Traefik < Kamal::Cli::Base
 
   desc "restart", "Restart existing Traefik container on servers"
   def restart
-    mutating do
+    with_lock do
       stop
       start
     end
@@ -66,29 +69,31 @@ class Kamal::Cli::Traefik < Kamal::Cli::Base
   option :since, aliases: "-s", desc: "Show logs since timestamp (e.g. 2013-01-02T13:23:37Z) or relative (e.g. 42m for 42 minutes)"
   option :lines, type: :numeric, aliases: "-n", desc: "Number of log lines to pull from each server"
   option :grep, aliases: "-g", desc: "Show lines with grep match only (use this to fetch specific requests by id)"
+  option :grep_options, aliases: "-o", desc: "Additional options supplied to grep"
   option :follow, aliases: "-f", desc: "Follow logs on primary server (or specific host set by --hosts)"
   def logs
     grep = options[:grep]
+    grep_options = options[:grep_options]
 
     if options[:follow]
       run_locally do
         info "Following logs on #{KAMAL.primary_host}..."
-        info KAMAL.traefik.follow_logs(host: KAMAL.primary_host, grep: grep)
+        info KAMAL.traefik.follow_logs(host: KAMAL.primary_host, grep: grep, grep_options: grep_options)
-        exec KAMAL.traefik.follow_logs(host: KAMAL.primary_host, grep: grep)
+        exec KAMAL.traefik.follow_logs(host: KAMAL.primary_host, grep: grep, grep_options: grep_options)
       end
     else
       since = options[:since]
       lines = options[:lines].presence || ((since || grep) ? nil : 100) # Default to 100 lines if since or grep isn't set
 
       on(KAMAL.traefik_hosts) do |host|
-        puts_by_host host, capture(*KAMAL.traefik.logs(since: since, lines: lines, grep: grep)), type: "Traefik"
+        puts_by_host host, capture(*KAMAL.traefik.logs(since: since, lines: lines, grep: grep, grep_options: grep_options)), type: "Traefik"
       end
     end
   end
 
   desc "remove", "Remove Traefik container and image from servers"
   def remove
-    mutating do
+    with_lock do
       stop
       remove_container
       remove_image
@@ -97,7 +102,7 @@ class Kamal::Cli::Traefik < Kamal::Cli::Base
 
   desc "remove_container", "Remove Traefik container from servers", hide: true
   def remove_container
-    mutating do
+    with_lock do
       on(KAMAL.traefik_hosts) do
         execute *KAMAL.auditor.record("Removed traefik container"), verbosity: :debug
         execute *KAMAL.traefik.remove_container
@@ -107,11 +112,51 @@ class Kamal::Cli::Traefik < Kamal::Cli::Base
 
   desc "remove_image", "Remove Traefik image from servers", hide: true
   def remove_image
-    mutating do
+    with_lock do
       on(KAMAL.traefik_hosts) do
         execute *KAMAL.auditor.record("Removed traefik image"), verbosity: :debug
         execute *KAMAL.traefik.remove_image
       end
     end
   end
+
+  desc "downgrade", "Downgrade to Traefik on servers (stop container, remove container, start new container, reboot app)"
+  option :rolling, type: :boolean, default: false, desc: "Reboot proxy on hosts in sequence, rather than in parallel"
+  option :confirmed, aliases: "-y", type: :boolean, default: false, desc: "Proceed without confirmation question"
+  def downgrade
+    invoke_options = { "version" => KAMAL.config.latest_tag }.merge(options)
+
+    confirming "This will cause a brief outage on each host. Are you sure?" do
+      host_groups = options[:rolling] ? KAMAL.hosts : [ KAMAL.hosts ]
+      host_groups.each do |hosts|
+        host_list = Array(hosts).join(",")
+        say "Downgrading to Traefik on #{host_list}...", :magenta
+        run_hook "pre-traefik-reboot", hosts: host_list
+        on(hosts) do |host|
+          execute *KAMAL.auditor.record("Rebooted Traefik"), verbosity: :debug
+          execute *KAMAL.registry.login
+
+          "Stopping and removing kamal-proxy on #{host}, if running..."
+          execute *KAMAL.traefik.cleanup_kamal_proxy
+
+          "Stopping and removing Traefik on #{host}, if running..."
+          execute *KAMAL.traefik.stop, raise_on_non_zero_exit: false
+          execute *KAMAL.traefik.remove_container
+          execute *KAMAL.traefik.remove_image
+        end
+
+        KAMAL.with_specific_hosts(hosts) do
+          invoke "kamal:cli:traefik:boot", [], invoke_options
+          reset_invocation(Kamal::Cli::Traefik)
+          invoke "kamal:cli:app:boot", [], invoke_options
+          reset_invocation(Kamal::Cli::App)
+          invoke "kamal:cli:prune:all", [], invoke_options
+          reset_invocation(Kamal::Cli::Prune)
+        end
+
+        run_hook "post-traefik-reboot", hosts: host_list
+        say "Downgraded to Traefik on #{host_list}", :magenta
+      end
+    end
+  end
 end

lib/kamal/commander.rb

@@ -2,12 +2,14 @@ require "active_support/core_ext/enumerable"
 require "active_support/core_ext/module/delegation"
 
 class Kamal::Commander
-  attr_accessor :verbosity, :holding_lock, :hold_lock_on_error
+  attr_accessor :verbosity, :holding_lock, :connected
+  delegate :hosts, :roles, :primary_host, :primary_role, :roles_on, :traefik_hosts, :accessory_hosts, to: :specifics
 
   def initialize
     self.verbosity = :info
     self.holding_lock = false
-    self.hold_lock_on_error = false
+    self.connected = false
+    @specifics = nil
   end
 
   def config
@@ -24,10 +26,12 @@ class Kamal::Commander
   attr_reader :specific_roles, :specific_hosts
 
   def specific_primary!
+    @specifics = nil
     self.specific_hosts = [ config.primary_host ]
   end
 
   def specific_roles=(role_names)
+    @specifics = nil
     if role_names.present?
       @specific_roles = Kamal::Utils.filter_specific_items(role_names, config.roles)
 
@@ -40,6 +44,7 @@ class Kamal::Commander
   end
 
   def specific_hosts=(hosts)
+    @specifics = nil
     if hosts.present?
       @specific_hosts = Kamal::Utils.filter_specific_items(hosts, config.all_hosts)
 
@@ -51,37 +56,11 @@ class Kamal::Commander
     end
   end
 
-  def primary_host
+  def with_specific_hosts(hosts)
-    # Given a list of specific roles, make an effort to match up with the primary_role
+    original_hosts, self.specific_hosts = specific_hosts, hosts
-    specific_hosts&.first || specific_roles&.detect { |role| role == config.primary_role }&.primary_host || specific_roles&.first&.primary_host || config.primary_host
+    yield
-  end
+  ensure
-
+    self.specific_hosts = original_hosts
-  def primary_role
-    roles_on(primary_host).first
-  end
-
-  def roles
-    (specific_roles || config.roles).select do |role|
-      ((specific_hosts || config.all_hosts) & role.hosts).any?
-    end
-  end
-
-  def hosts
-    (specific_hosts || config.all_hosts).select do |host|
-      (specific_roles || config.roles).flat_map(&:hosts).include?(host)
-    end
-  end
-
-  def roles_on(host)
-    roles.select { |role| role.hosts.include?(host.to_s) }
-  end
-
-  def traefik_hosts
-    specific_hosts || config.traefik_hosts
-  end
-
-  def accessory_hosts
-    specific_hosts || config.accessories.flat_map(&:hosts)
   end
 
   def accessory_names
@@ -93,8 +72,8 @@ class Kamal::Commander
   end
 
 
-  def app(role: nil)
+  def app(role: nil, host: nil)
-    Kamal::Commands::App.new(config, role: role)
+    Kamal::Commands::App.new(config, role: role, host: host)
   end
 
   def accessory(name)
@@ -166,8 +145,8 @@ class Kamal::Commander
     self.holding_lock
   end
 
-  def hold_lock_on_error?
+  def connected?
-    self.hold_lock_on_error
+    self.connected
   end
 
   private
@@ -181,4 +160,8 @@ class Kamal::Commander
       SSHKit.config.command_map[:docker] = "docker" # No need to use /usr/bin/env, just clogs up the logs
       SSHKit.config.output_verbosity = verbosity
     end
+
+    def specifics
+      @specifics ||= Kamal::Commander::Specifics.new(config, specific_hosts, specific_roles)
+    end
 end

lib/kamal/commander/specifics.rb

@@ -0,0 +1,49 @@
+class Kamal::Commander::Specifics
+  attr_reader :primary_host, :primary_role, :hosts, :roles
+  delegate :stable_sort!, to: Kamal::Utils
+
+  def initialize(config, specific_hosts, specific_roles)
+    @config, @specific_hosts, @specific_roles = config, specific_hosts, specific_roles
+
+    @roles, @hosts = specified_roles, specified_hosts
+
+    @primary_host = specific_hosts&.first || primary_specific_role&.primary_host || config.primary_host
+    @primary_role = primary_or_first_role(roles_on(primary_host))
+
+    stable_sort!(roles) { |role| role == primary_role ? 0 : 1 }
+    stable_sort!(hosts) { |host| roles_on(host).any? { |role| role == primary_role } ? 0 : 1 }
+  end
+
+  def roles_on(host)
+    roles.select { |role| role.hosts.include?(host.to_s) }
+  end
+
+  def traefik_hosts
+    config.traefik_hosts & specified_hosts
+  end
+
+  def accessory_hosts
+    config.accessories.flat_map(&:hosts) & specified_hosts
+  end
+
+  private
+    attr_reader :config, :specific_hosts, :specific_roles
+
+    def primary_specific_role
+      primary_or_first_role(specific_roles) if specific_roles.present?
+    end
+
+    def primary_or_first_role(roles)
+      roles.detect { |role| role == config.primary_role } || roles.first
+    end
+
+    def specified_roles
+      (specific_roles || config.roles) \
+        .select { |role| ((specific_hosts || config.all_hosts) & role.hosts).any? }
+    end
+
+    def specified_hosts
+      (specific_hosts || config.all_hosts) \
+        .select { |host| (specific_roles || config.roles).flat_map(&:hosts).include?(host) }
+    end
+end

lib/kamal/commands/accessory.rb

@@ -36,17 +36,17 @@ class Kamal::Commands::Accessory < Kamal::Commands::Base
   end
 
 
-  def logs(since: nil, lines: nil, grep: nil)
+  def logs(since: nil, lines: nil, grep: nil, grep_options: nil)
     pipe \
       docker(:logs, service_name, (" --since #{since}" if since), (" --tail #{lines}" if lines), "--timestamps", "2>&1"),
-      ("grep '#{grep}'" if grep)
+      ("grep '#{grep}'#{" #{grep_options}" if grep_options}" if grep)
   end
 
-  def follow_logs(grep: nil)
+  def follow_logs(grep: nil, grep_options: nil)
     run_over_ssh \
       pipe \
         docker(:logs, service_name, "--timestamps", "--tail", "10", "--follow", "2>&1"),
-        (%(grep "#{grep}") if grep)
+        (%(grep "#{grep}"#{" #{grep_options}" if grep_options}) if grep)
   end
 
 
@@ -99,11 +99,11 @@ class Kamal::Commands::Accessory < Kamal::Commands::Base
   end
 
   def make_env_directory
-    make_directory accessory_config.host_env_directory
+    make_directory accessory_config.env.secrets_directory
   end
 
   def remove_env_file
-    [:rm, "-f", accessory_config.host_env_file_path]
+    [ :rm, "-f", accessory_config.env.secrets_file ]
   end
 
   private

lib/kamal/commands/app.rb

@@ -3,11 +3,12 @@ class Kamal::Commands::App < Kamal::Commands::Base
 
   ACTIVE_DOCKER_STATUSES = [ :running, :restarting ]
 
-  attr_reader :role, :role
+  attr_reader :role, :host
 
-  def initialize(config, role: nil)
+  def initialize(config, role: nil, host: nil)
     super(config)
     @role = role
+    @host = host
   end
 
   def run(hostname: nil)
@@ -15,10 +16,10 @@ class Kamal::Commands::App < Kamal::Commands::Base
       "--detach",
       "--restart unless-stopped",
       "--name", container_name,
-      *(["--hostname", hostname] if hostname),
+      *([ "--hostname", hostname ] if hostname),
       "-e", "KAMAL_CONTAINER_NAME=\"#{container_name}\"",
       "-e", "KAMAL_VERSION=\"#{config.version}\"",
-      *role.env_args,
+      *role.env_args(host),
       *role.health_check_args,
       *role.logging_args,
       *config.volume_args,
@@ -49,7 +50,7 @@ class Kamal::Commands::App < Kamal::Commands::Base
 
 
   def current_running_container_id
-    docker :ps, "--quiet", *filter_args(statuses: ACTIVE_DOCKER_STATUSES), "--latest"
+    current_running_container(format: "--quiet")
   end
 
   def container_id_for_version(version, only_running: false)
@@ -57,22 +58,24 @@ class Kamal::Commands::App < Kamal::Commands::Base
   end
 
   def current_running_version
-    list_versions("--latest", statuses: ACTIVE_DOCKER_STATUSES)
+    pipe \
+      current_running_container(format: "--format '{{.Names}}'"),
+      extract_version_from_name
   end
 
   def list_versions(*docker_args, statuses: nil)
     pipe \
       docker(:ps, *filter_args(statuses: statuses), *docker_args, "--format", '"{{.Names}}"'),
-      %(while read line; do echo ${line##{role.container_prefix}-}; done) # Extract SHA from "service-role-dest-SHA"
+      extract_version_from_name
   end
 
 
   def make_env_directory
-    make_directory role.host_env_directory
+    make_directory role.env(host).secrets_directory
   end
 
   def remove_env_file
-    [ :rm, "-f", role.host_env_file_path ]
+    [ :rm, "-f", role.env(host).secrets_file ]
   end
 
 
@@ -81,12 +84,31 @@ class Kamal::Commands::App < Kamal::Commands::Base
       [ role.container_prefix, version || config.version ].compact.join("-")
     end
 
+    def latest_image_id
+      docker :image, :ls, *argumentize("--filter", "reference=#{config.latest_image}"), "--format", "'{{.ID}}'"
+    end
+
+    def current_running_container(format:)
+      pipe \
+        shell(chain(latest_image_container(format: format), latest_container(format: format))),
+        [ :head, "-1" ]
+    end
+
+    def latest_image_container(format:)
+      latest_container format: format, filters: [ "ancestor=$(#{latest_image_id.join(" ")})" ]
+    end
+
+    def latest_container(format:, filters: nil)
+      docker :ps, "--latest", *format, *filter_args(statuses: ACTIVE_DOCKER_STATUSES), argumentize("--filter", filters)
+    end
+
     def filter_args(statuses: nil)
       argumentize "--filter", filters(statuses: statuses)
     end
 
-    def service_role_dest
+    def extract_version_from_name
-      [ config.service, role, config.destination ].compact.join("-")
+      # Extract SHA from "service-role-dest-SHA"
+      %(while read line; do echo ${line##{role.container_prefix}-}; done)
     end
 
     def filters(statuses: nil)

lib/kamal/commands/app/assets.rb

@@ -4,8 +4,8 @@ module Kamal::Commands::App::Assets
 
     combine \
       make_directory(role.asset_extracted_path),
-      [*docker(:stop, "-t 1", asset_container, "2> /dev/null"), "|| true"],
+      [ *docker(:stop, "-t 1", asset_container, "2> /dev/null"), "|| true" ],
-      docker(:run, "--name", asset_container, "--detach", "--rm", config.latest_image, "sleep 1000000"),
+      docker(:run, "--name", asset_container, "--detach", "--rm", config.absolute_image, "sleep 1000000"),
       docker(:cp, "-L", "#{asset_container}:#{role.asset_path}/.", role.asset_extracted_path),
       docker(:stop, "-t 1", asset_container),
       by: "&&"
@@ -17,7 +17,7 @@ module Kamal::Commands::App::Assets
       old_extracted_path, old_volume_path = role.asset_extracted_path(old_version), role.asset_volume(old_version).host_path
     end
 
-    commands = [make_directory(new_volume_path), copy_contents(new_extracted_path, new_volume_path)]
+    commands = [ make_directory(new_volume_path), copy_contents(new_extracted_path, new_volume_path) ]
 
     if old_version.present?
       commands << copy_contents(new_extracted_path, old_volume_path, continue_on_error: true)
@@ -46,6 +46,6 @@ module Kamal::Commands::App::Assets
     end
 
     def copy_contents(source, destination, continue_on_error: false)
-      [ :cp, "-rnT", "#{source}", destination, *("|| true" if continue_on_error)]
+      [ :cp, "-rnT", "#{source}", destination, *("|| true" if continue_on_error) ]
     end
 end

lib/kamal/commands/app/containers.rb

@@ -1,4 +1,6 @@
 module Kamal::Commands::App::Containers
+  DOCKER_HEALTH_LOG_FORMAT    = "'{{json .State.Health}}'"
+
   def list_containers
     docker :container, :ls, "--all", *filter_args
   end
@@ -20,4 +22,10 @@ module Kamal::Commands::App::Containers
   def remove_containers
     docker :container, :prune, "--force", *filter_args
   end
+
+  def container_health_log(version:)
+    pipe \
+      container_id_for(container_name: container_name(version)),
+      xargs(docker(:inspect, "--format", DOCKER_HEALTH_LOG_FORMAT))
+  end
 end

lib/kamal/commands/app/cord.rb

@@ -2,7 +2,7 @@ module Kamal::Commands::App::Cord
   def cord(version:)
     pipe \
       docker(:inspect, "-f '{{ range .Mounts }}{{printf \"%s %s\\n\" .Source .Destination}}{{ end }}'", container_name(version)),
-      [:awk, "'$2 == \"#{role.cord_volume.container_path}\" {print $1}'"]
+      [ :awk, "'$2 == \"#{role.cord_volume.container_path}\" {print $1}'" ]
   end
 
   def tie_cord(cord)
@@ -17,6 +17,6 @@ module Kamal::Commands::App::Cord
     def create_empty_file(file)
       chain \
         make_directory_for(file),
-        [:touch, file]
+        [ :touch, file ]
     end
 end

lib/kamal/commands/app/execution.rb

@@ -1,27 +1,29 @@
 module Kamal::Commands::App::Execution
-  def execute_in_existing_container(*command, interactive: false)
+  def execute_in_existing_container(*command, interactive: false, env:)
     docker :exec,
       ("-it" if interactive),
+      *argumentize("--env", env),
       container_name,
       *command
   end
 
-  def execute_in_new_container(*command, interactive: false)
+  def execute_in_new_container(*command, interactive: false, env:)
     docker :run,
       ("-it" if interactive),
       "--rm",
-      *role&.env_args,
+      *role&.env_args(host),
+      *argumentize("--env", env),
       *config.volume_args,
       *role&.option_args,
       config.absolute_image,
       *command
   end
 
-  def execute_in_existing_container_over_ssh(*command, host:)
+  def execute_in_existing_container_over_ssh(*command,  env:)
-    run_over_ssh execute_in_existing_container(*command, interactive: true), host: host
+    run_over_ssh execute_in_existing_container(*command, interactive: true, env: env), host: host
   end
 
-  def execute_in_new_container_over_ssh(*command, host:)
+  def execute_in_new_container_over_ssh(*command, env:)
-    run_over_ssh execute_in_new_container(*command, interactive: true), host: host
+    run_over_ssh execute_in_new_container(*command, interactive: true, env: env), host: host
   end
 end

lib/kamal/commands/app/images.rb

@@ -7,7 +7,7 @@ module Kamal::Commands::App::Images
     docker :image, :prune, "--all", "--force", *filter_args
   end
 
-  def tag_current_image_as_latest
+  def tag_latest_image
     docker :tag, config.absolute_image, config.latest_image
   end
 end

lib/kamal/commands/app/logging.rb

@@ -1,17 +1,17 @@
 module Kamal::Commands::App::Logging
-  def logs(since: nil, lines: nil, grep: nil)
+  def logs(version: nil, since: nil, lines: nil, grep: nil, grep_options: nil)
     pipe \
-      current_running_container_id,
+      version ? container_id_for_version(version) : current_running_container_id,
       "xargs docker logs#{" --since #{since}" if since}#{" --tail #{lines}" if lines} 2>&1",
-      ("grep '#{grep}'" if grep)
+      ("grep '#{grep}'#{" #{grep_options}" if grep_options}" if grep)
   end
 
-  def follow_logs(host:, lines: nil, grep: nil)
+  def follow_logs(host:, lines: nil, grep: nil, grep_options: nil)
     run_over_ssh \
       pipe(
         current_running_container_id,
         "xargs docker logs --timestamps#{" --tail #{lines}" if lines} --follow 2>&1",
-        (%(grep "#{grep}") if grep)
+        (%(grep "#{grep}"#{" #{grep_options}" if grep_options}) if grep)
       ),
       host: host
   end

lib/kamal/commands/auditor.rb

@@ -9,7 +9,7 @@ class Kamal::Commands::Auditor < Kamal::Commands::Base
   # Runs remotely
   def record(line, **details)
     append \
-      [ :echo, audit_tags(**details).except(:version, :service_version).to_s, line ],
+      [ :echo, audit_tags(**details).except(:version, :service_version, :service).to_s, line ],
       audit_log_file
   end
 
@@ -21,7 +21,7 @@ class Kamal::Commands::Auditor < Kamal::Commands::Base
     def audit_log_file
       file = [ config.service, config.destination, "audit.log" ].compact.join("-")
 
-      "#{config.run_directory}/#{file}"
+      File.join(config.run_directory, file)
     end
 
     def audit_tags(**details)

lib/kamal/commands/base.rb

@@ -3,7 +3,6 @@ module Kamal::Commands
     delegate :sensitive, :argumentize, to: Kamal::Utils
 
     DOCKER_HEALTH_STATUS_FORMAT = "'{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'"
-    DOCKER_HEALTH_LOG_FORMAT    = "'{{json .State.Health}}'"
 
     attr_accessor :config
 
@@ -18,7 +17,7 @@ module Kamal::Commands
         elsif config.ssh.proxy && config.ssh.proxy.is_a?(Net::SSH::Proxy::Command)
           cmd << " -o ProxyCommand='#{config.ssh.proxy.command_line_template}'"
         end
-        cmd << " -t #{config.ssh.user}@#{host} -p #{config.ssh.port} '#{command.join(" ")}'"
+        cmd << " -t #{config.ssh.user}@#{host} -p #{config.ssh.port} '#{command.join(" ").gsub("'", "'\\\\''")}'"
       end
     end
 
@@ -71,13 +70,17 @@ module Kamal::Commands
       end
 
       def shell(command)
-        [ :sh, "-c", "'#{command.flatten.join(" ").gsub("'", "'\\''")}'" ]
+        [ :sh, "-c", "'#{command.flatten.join(" ").gsub("'", "'\\\\''")}'" ]
       end
 
       def docker(*args)
         args.compact.unshift :docker
       end
 
+      def git(*args, path: nil)
+        [ :git, *([ "-C", path ] if path), *args.compact ]
+      end
+
       def tags(**details)
         Kamal::Tags.from_config(config, **details)
       end

lib/kamal/commands/builder.rb

@@ -1,24 +1,32 @@
 require "active_support/core_ext/string/filters"
 
 class Kamal::Commands::Builder < Kamal::Commands::Base
-  delegate :create, :remove, :push, :clean, :pull, :info, :validate_image, to: :target
+  delegate :create, :remove, :push, :clean, :pull, :info, :context_hosts, :config_context_hosts, :validate_image,
+           :first_mirror, to: :target
+
+  include Clone
 
   def name
     target.class.to_s.remove("Kamal::Commands::Builder::").underscore.inquiry
   end
 
   def target
-    case
+    if config.builder.multiarch?
-    when !config.builder.multiarch? && !config.builder.cached?
+      if config.builder.remote?
-      native
+        if config.builder.local?
-    when !config.builder.multiarch? && config.builder.cached?
+          multiarch_remote
-      native_cached
+        else
-    when config.builder.local? && config.builder.remote?
+          native_remote
-      multiarch_remote
+        end
-    when config.builder.remote?
+      else
-      native_remote
+        multiarch
+      end
     else
-      multiarch
+      if config.builder.cached?
+        native_cached
+      else
+        native
+      end
     end
   end
 

lib/kamal/commands/builder/base.rb

@@ -2,8 +2,10 @@
 class Kamal::Commands::Builder::Base < Kamal::Commands::Base
   class BuilderError < StandardError; end
 
+  ENDPOINT_DOCKER_HOST_INSPECT = "'{{.Endpoints.docker.Host}}'"
+
   delegate :argumentize, to: Kamal::Utils
-  delegate :args, :secrets, :dockerfile, :local_arch, :local_host, :remote_arch, :remote_host, :cache_from, :cache_to, :ssh, to: :builder_config
+  delegate :args, :secrets, :dockerfile, :target, :local_arch, :local_host, :remote_arch, :remote_host, :cache_from, :cache_to, :ssh, to: :builder_config
 
   def clean
     docker :image, :rm, "--force", config.absolute_image
@@ -14,7 +16,7 @@ class Kamal::Commands::Builder::Base < Kamal::Commands::Base
   end
 
   def build_options
-    [ *build_tags, *build_cache, *build_labels, *build_args, *build_secrets, *build_dockerfile, *build_ssh ]
+    [ *build_tags, *build_cache, *build_labels, *build_args, *build_secrets, *build_dockerfile, *build_target, *build_ssh ]
   end
 
   def build_context
@@ -25,11 +27,22 @@ class Kamal::Commands::Builder::Base < Kamal::Commands::Base
     pipe \
       docker(:inspect, "-f", "'{{ .Config.Labels.service }}'", config.absolute_image),
       any(
-        [:grep, "-x", config.service],
+        [ :grep, "-x", config.service ],
         "(echo \"Image #{config.absolute_image} is missing the 'service' label\" && exit 1)"
       )
   end
 
+  def context_hosts
+    :true
+  end
+
+  def config_context_hosts
+    []
+  end
+
+  def first_mirror
+    docker(:info, "--format '{{index .RegistryConfig.Mirrors 0}}'")
+  end
 
   private
     def build_tags
@@ -38,8 +51,8 @@ class Kamal::Commands::Builder::Base < Kamal::Commands::Base
 
     def build_cache
       if cache_to && cache_from
-        ["--cache-to", cache_to,
+        [ "--cache-to", cache_to,
-          "--cache-from", cache_from]
+          "--cache-from", cache_from ]
       end
     end
 
@@ -63,6 +76,10 @@ class Kamal::Commands::Builder::Base < Kamal::Commands::Base
       end
     end
 
+    def build_target
+      argumentize "--target", target if target.present?
+    end
+
     def build_ssh
       argumentize "--ssh", ssh if ssh.present?
     end
@@ -70,4 +87,8 @@ class Kamal::Commands::Builder::Base < Kamal::Commands::Base
     def builder_config
       config.builder
     end
+
+    def context_host(builder_name)
+      docker :context, :inspect, builder_name, "--format", ENDPOINT_DOCKER_HOST_INSPECT
+    end
 end

lib/kamal/commands/builder/clone.rb

@@ -0,0 +1,29 @@
+module Kamal::Commands::Builder::Clone
+  extend ActiveSupport::Concern
+
+  included do
+    delegate :clone_directory, :build_directory, to: :"config.builder"
+  end
+
+  def clone
+    git :clone, Kamal::Git.root, "--recurse-submodules", path: clone_directory
+  end
+
+  def clone_reset_steps
+    [
+      git(:remote, "set-url", :origin, Kamal::Git.root, path: build_directory),
+      git(:fetch, :origin, path: build_directory),
+      git(:reset, "--hard", Kamal::Git.revision, path: build_directory),
+      git(:clean, "-fdx", path: build_directory),
+      git(:submodule, :update, "--init", path: build_directory)
+    ]
+  end
+
+  def clone_status
+    git :status, "--porcelain", path: build_directory
+  end
+
+  def clone_revision
+    git :"rev-parse", :HEAD, path: build_directory
+  end
+end

lib/kamal/commands/builder/multiarch.rb

@@ -7,6 +7,12 @@ class Kamal::Commands::Builder::Multiarch < Kamal::Commands::Builder::Base
     docker :buildx, :rm, builder_name
   end
 
+  def info
+    combine \
+      docker(:context, :ls),
+      docker(:buildx, :ls)
+  end
+
   def push
     docker :buildx, :build,
       "--push",
@@ -16,10 +22,8 @@ class Kamal::Commands::Builder::Multiarch < Kamal::Commands::Builder::Base
       build_context
   end
 
-  def info
+  def context_hosts
-    combine \
+    docker :buildx, :inspect, builder_name, "> /dev/null"
-      docker(:context, :ls),
-      docker(:buildx, :ls)
   end
 
   private

lib/kamal/commands/builder/multiarch/remote.rb

@@ -12,6 +12,16 @@ class Kamal::Commands::Builder::Multiarch::Remote < Kamal::Commands::Builder::Mu
       super
   end
 
+  def context_hosts
+    chain \
+      context_host(builder_name_with_arch(local_arch)),
+      context_host(builder_name_with_arch(remote_arch))
+  end
+
+  def config_context_hosts
+    [ local_host, remote_host ].compact
+  end
+
   private
     def builder_name
       super + "-remote"
@@ -48,4 +58,8 @@ class Kamal::Commands::Builder::Multiarch::Remote < Kamal::Commands::Builder::Mu
     def remove_context(arch)
       docker :context, :rm, builder_name_with_arch(arch)
     end
+
+    def platform_names
+      "linux/#{local_arch},linux/#{remote_arch}"
+    end
 end

lib/kamal/commands/builder/native.rb

@@ -7,14 +7,14 @@ class Kamal::Commands::Builder::Native < Kamal::Commands::Builder::Base
     # No-op on native without cache
   end
 
+  def info
+    # No-op on native
+  end
+
   def push
     combine \
       docker(:build, *build_options, build_context),
       docker(:push, config.absolute_image),
       docker(:push, config.latest_image)
   end
-
-  def info
-    # No-op on native
-  end
 end

lib/kamal/commands/builder/native/cached.rb

@@ -1,6 +1,6 @@
 class Kamal::Commands::Builder::Native::Cached < Kamal::Commands::Builder::Native
   def create
-    docker :buildx, :create, "--use", "--driver=docker-container"
+    docker :buildx, :create, "--name", builder_name, "--use", "--driver=docker-container"
   end
 
   def remove
@@ -13,4 +13,13 @@ class Kamal::Commands::Builder::Native::Cached < Kamal::Commands::Builder::Nativ
       *build_options,
       build_context
   end
+
+  def context_hosts
+    docker :buildx, :inspect, builder_name, "> /dev/null"
+  end
+
+  private
+    def builder_name
+      "kamal-#{config.service}-native-cached"
+    end
 end

lib/kamal/commands/builder/native/remote.rb

@@ -11,21 +11,29 @@ class Kamal::Commands::Builder::Native::Remote < Kamal::Commands::Builder::Nativ
       remove_buildx
   end
 
-  def push
-    docker :buildx, :build,
-      "--push",
-      "--platform", platform,
-      "--builder", builder_name,
-      *build_options,
-      build_context
-  end
-
   def info
     chain \
       docker(:context, :ls),
       docker(:buildx, :ls)
   end
 
+  def push
+    docker :buildx, :build,
+    "--push",
+    "--platform", platform,
+    "--builder", builder_name,
+    *build_options,
+    build_context
+  end
+
+  def context_hosts
+    context_host(builder_name_with_arch)
+  end
+
+  def config_context_hosts
+    [ remote_host ]
+  end
+
 
   private
     def builder_name

lib/kamal/commands/healthcheck.rb

@@ -1,60 +0,0 @@
-class Kamal::Commands::Healthcheck < Kamal::Commands::Base
-
-  def run
-    primary = config.role(config.primary_role)
-
-    docker :run,
-      "--detach",
-      "--name", container_name_with_version,
-      "--publish", "#{exposed_port}:#{config.healthcheck["port"]}",
-      "--label", "service=#{config.healthcheck_service}",
-      "-e", "KAMAL_CONTAINER_NAME=\"#{config.healthcheck_service}\"",
-      *primary.env_args,
-      *primary.health_check_args(cord: false),
-      *config.volume_args,
-      *primary.option_args,
-      config.absolute_image,
-      primary.cmd
-  end
-
-  def status
-    pipe container_id, xargs(docker(:inspect, "--format", DOCKER_HEALTH_STATUS_FORMAT))
-  end
-
-  def container_health_log
-    pipe container_id, xargs(docker(:inspect, "--format", DOCKER_HEALTH_LOG_FORMAT))
-  end
-
-  def logs
-    pipe container_id, xargs(docker(:logs, "--tail", log_lines, "2>&1"))
-  end
-
-  def stop
-    pipe container_id, xargs(docker(:stop))
-  end
-
-  def remove
-    pipe container_id, xargs(docker(:container, :rm))
-  end
-
-  private
-    def container_name_with_version
-      "#{config.healthcheck_service}-#{config.version}"
-    end
-
-    def container_id
-      container_id_for(container_name: container_name_with_version)
-    end
-
-    def health_url
-      "http://localhost:#{exposed_port}#{config.healthcheck["path"]}"
-    end
-
-    def exposed_port
-      config.healthcheck["exposed_port"]
-    end
-
-    def log_lines
-      config.healthcheck["log_lines"]
-    end
-end

lib/kamal/commands/hook.rb

@@ -9,6 +9,6 @@ class Kamal::Commands::Hook < Kamal::Commands::Base
 
   private
     def hook_file(hook)
-      "#{config.hooks_path}/#{hook}"
+      File.join(config.hooks_path, hook)
     end
 end

lib/kamal/commands/lock.rb

@@ -5,14 +5,14 @@ require "base64"
 class Kamal::Commands::Lock < Kamal::Commands::Base
   def acquire(message, version)
     combine \
-      [:mkdir, lock_dir],
+      [ :mkdir, lock_dir ],
       write_lock_details(message, version)
   end
 
   def release
     combine \
-      [:rm, lock_details_file],
+      [ :rm, lock_details_file ],
-      [:rm, "-r", lock_dir]
+      [ :rm, "-r", lock_dir ]
   end
 
   def status
@@ -21,31 +21,41 @@ class Kamal::Commands::Lock < Kamal::Commands::Base
       read_lock_details
   end
 
+  def ensure_locks_directory
+    [ :mkdir, "-p", locks_dir ]
+  end
+
   private
     def write_lock_details(message, version)
       write \
-        [:echo, "\"#{Base64.encode64(lock_details(message, version))}\""],
+        [ :echo, "\"#{Base64.encode64(lock_details(message, version))}\"" ],
         lock_details_file
     end
 
     def read_lock_details
       pipe \
-        [:cat, lock_details_file],
+        [ :cat, lock_details_file ],
-        [:base64, "-d"]
+        [ :base64, "-d" ]
     end
 
     def stat_lock_dir
       write \
-        [:stat, lock_dir],
+        [ :stat, lock_dir ],
         "/dev/null"
     end
 
+    def locks_dir
+      File.join(config.run_directory, "locks")
+    end
+
     def lock_dir
-      "#{config.run_directory}/lock-#{config.service}"
+      dir_name = [ config.service, config.destination ].compact.join("-")
+
+      File.join(locks_dir, dir_name)
     end
 
     def lock_details_file
-      [lock_dir, :details].join("/")
+      File.join(lock_dir, "details")
     end
 
     def lock_details(message, version)

lib/kamal/commands/prune.rb

@@ -26,7 +26,7 @@ class Kamal::Commands::Prune < Kamal::Commands::Base
 
   private
     def stopped_containers_filters
-      [ "created", "exited", "dead" ].flat_map { |status| ["--filter", "status=#{status}"] }
+      [ "created", "exited", "dead" ].flat_map { |status| [ "--filter", "status=#{status}" ] }
     end
 
     def active_image_list
@@ -43,4 +43,4 @@ class Kamal::Commands::Prune < Kamal::Commands::Base
     def healthcheck_service_filter
       [ "--filter", "label=service=#{config.healthcheck_service}" ]
     end
-  end
+end

lib/kamal/commands/registry.rb

@@ -3,21 +3,12 @@ class Kamal::Commands::Registry < Kamal::Commands::Base
 
   def login
     docker :login,
-      registry["server"],
+      registry.server,
-      "-u", sensitive(Kamal::Utils.escape_shell_value(lookup("username"))),
+      "-u", sensitive(Kamal::Utils.escape_shell_value(registry.username)),
-      "-p", sensitive(Kamal::Utils.escape_shell_value(lookup("password")))
+      "-p", sensitive(Kamal::Utils.escape_shell_value(registry.password))
   end
 
   def logout
-    docker :logout, registry["server"]
+    docker :logout, registry.server
   end
-
-  private
-    def lookup(key)
-      if registry[key].is_a?(Array)
-        ENV.fetch(registry[key].first).dup
-      else
-        registry[key]
-      end
-    end
 end

lib/kamal/commands/server.rb

@@ -1,5 +1,5 @@
 class Kamal::Commands::Server < Kamal::Commands::Base
   def ensure_run_directory
-    [:mkdir, "-p", config.run_directory]
+    [ :mkdir, "-p", config.run_directory ]
   end
 end

lib/kamal/commands/traefik.rb

@@ -1,19 +1,6 @@
 class Kamal::Commands::Traefik < Kamal::Commands::Base
   delegate :argumentize, :optionize, to: Kamal::Utils
-
+  delegate :port, :publish?, :labels, :env, :image, :options, :args, to: :"config.traefik"
-  DEFAULT_IMAGE = "traefik:v2.10"
-  CONTAINER_PORT = 80
-  DEFAULT_ARGS = {
-    'log.level' => 'DEBUG'
-  }
-  DEFAULT_LABELS = {
-    # These ensure we serve a 502 rather than a 404 if no containers are available
-    "traefik.http.routers.catchall.entryPoints" => "http",
-    "traefik.http.routers.catchall.rule" => "PathPrefix(`/`)",
-    "traefik.http.routers.catchall.service" => "unavailable",
-    "traefik.http.routers.catchall.priority" => 1,
-    "traefik.http.services.unavailable.loadbalancer.server.port" => "0"
-  }
 
   def run
     docker :run, "--name traefik",
@@ -46,16 +33,16 @@ class Kamal::Commands::Traefik < Kamal::Commands::Base
     docker :ps, "--filter", "name=^traefik$"
   end
 
-  def logs(since: nil, lines: nil, grep: nil)
+  def logs(since: nil, lines: nil, grep: nil, grep_options: nil)
     pipe \
       docker(:logs, "traefik", (" --since #{since}" if since), (" --tail #{lines}" if lines), "--timestamps", "2>&1"),
-      ("grep '#{grep}'" if grep)
+      ("grep '#{grep}'#{" #{grep_options}" if grep_options}" if grep)
   end
 
-  def follow_logs(host:, grep: nil)
+  def follow_logs(host:, grep: nil, grep_options: nil)
     run_over_ssh pipe(
       docker(:logs, "traefik", "--timestamps", "--tail", "10", "--follow", "2>&1"),
-      (%(grep "#{grep}") if grep)
+      (%(grep "#{grep}"#{" #{grep_options}" if grep_options}) if grep)
     ).join(" "), host: host
   end
 
@@ -67,29 +54,26 @@ class Kamal::Commands::Traefik < Kamal::Commands::Base
     docker :image, :prune, "--all", "--force", "--filter", "label=org.opencontainers.image.title=Traefik"
   end
 
-  def port
-    "#{host_port}:#{CONTAINER_PORT}"
-  end
-
-  def env_file
-    Kamal::EnvFile.new(config.traefik.fetch("env", {}))
-  end
-
-  def host_env_file_path
-    File.join host_env_directory, "traefik.env"
-  end
-
   def make_env_directory
-    make_directory(host_env_directory)
+    make_directory(env.secrets_directory)
   end
 
   def remove_env_file
-    [:rm, "-f", host_env_file_path]
+    [ :rm, "-f", env.secrets_file ]
+  end
+
+  def cleanup_kamal_proxy
+    chain \
+      docker(:container, :stop, "kamal-proxy"),
+      combine(
+        docker(:container, :prune, "--force", "--filter", "label=org.opencontainers.image.title=kamal-proxy"),
+        docker(:image, :prune, "--all", "--force", "--filter", "label=org.opencontainers.image.title=kamal-proxy")
+      )
   end
 
   private
     def publish_args
-      argumentize "--publish", port unless config.traefik["publish"] == false
+      argumentize "--publish", port if publish?
     end
 
     def label_args
@@ -97,34 +81,14 @@ class Kamal::Commands::Traefik < Kamal::Commands::Base
     end
 
     def env_args
-      argumentize "--env-file", host_env_file_path
+      env.args
-    end
-
-    def host_env_directory
-      File.join config.host_env_directory, "traefik"
-    end
-
-    def labels
-      DEFAULT_LABELS.merge(config.traefik["labels"] || {})
-    end
-
-    def image
-      config.traefik.fetch("image") { DEFAULT_IMAGE }
     end
 
     def docker_options_args
-      optionize(config.traefik["options"] || {})
+      optionize(options)
     end
 
     def cmd_option_args
-      if args = config.traefik["args"]
+      optionize args, with: "="
-        optionize DEFAULT_ARGS.merge(args), with: "="
-      else
-        optionize DEFAULT_ARGS, with: "="
-      end
-    end
-
-    def host_port
-      config.traefik["host_port"] || CONTAINER_PORT
     end
 end

lib/kamal/configuration.rb

@@ -1,15 +1,19 @@
 require "active_support/ordered_options"
 require "active_support/core_ext/string/inquiry"
 require "active_support/core_ext/module/delegation"
+require "active_support/core_ext/hash/keys"
 require "pathname"
 require "erb"
 require "net/ssh/proxy/jump"
 
 class Kamal::Configuration
-  delegate :service, :image, :servers, :env, :labels, :registry, :stop_wait_time, :hooks_path, :logging, to: :raw_config, allow_nil: true
+  delegate :service, :image, :labels, :stop_wait_time, :hooks_path, to: :raw_config, allow_nil: true
   delegate :argumentize, :optionize, to: Kamal::Utils
 
   attr_reader :destination, :raw_config
+  attr_reader :accessories, :boot, :builder, :env, :healthcheck, :logging, :traefik, :servers, :ssh, :sshkit, :registry
+
+  include Validation
 
   class << self
     def create_from(config_file:, destination: nil, version: nil)
@@ -42,7 +46,29 @@ class Kamal::Configuration
     @raw_config = ActiveSupport::InheritableOptions.new(raw_config)
     @destination = destination
     @declared_version = version
-    valid? if validate
+
+    validate! raw_config, example: validation_yml.symbolize_keys, context: "", with: Kamal::Configuration::Validator::Configuration
+
+    # Eager load config to validate it, these are first as they have dependencies later on
+    @servers = Servers.new(config: self)
+    @registry = Registry.new(config: self)
+
+    @accessories = @raw_config.accessories&.keys&.collect { |name| Accessory.new(name, config: self) } || []
+    @boot = Boot.new(config: self)
+    @builder = Builder.new(config: self)
+    @env = Env.new(config: @raw_config.env || {})
+
+    @healthcheck = Healthcheck.new(healthcheck_config: @raw_config.healthcheck)
+    @logging = Logging.new(logging_config: @raw_config.logging)
+    @traefik = Traefik.new(config: self)
+    @ssh = Ssh.new(config: self)
+    @sshkit = Sshkit.new(config: self)
+
+    ensure_destination_if_required
+    ensure_required_keys_present
+    ensure_valid_kamal_version
+    ensure_retain_containers_valid
+    ensure_valid_service_name
   end
 
 
@@ -71,24 +97,20 @@ class Kamal::Configuration
 
 
   def roles
-    @roles ||= role_names.collect { |role_name| Role.new(role_name, config: self) }
+    servers.roles
   end
 
   def role(name)
     roles.detect { |r| r.name == name.to_s }
   end
 
-  def accessories
-    @accessories ||= raw_config.accessories&.keys&.collect { |name| Kamal::Configuration::Accessory.new(name, config: self) } || []
-  end
-
   def accessory(name)
     accessories.detect { |a| a.name == name.to_s }
   end
 
 
   def all_hosts
-    roles.flat_map(&:hosts).uniq
+    (roles + accessories).flat_map(&:hosts).uniq
   end
 
   def primary_host
@@ -120,7 +142,7 @@ class Kamal::Configuration
   end
 
   def repository
-    [ raw_config.registry["server"], image ].compact.join("/")
+    [ registry.server, image ].compact.join("/")
   end
 
   def absolute_image
@@ -128,7 +150,11 @@ class Kamal::Configuration
   end
 
   def latest_image
-    "#{repository}:latest"
+    "#{repository}:#{latest_tag}"
+  end
+
+  def latest_tag
+    [ "latest", *destination ].join("-")
   end
 
   def service_with_version
@@ -153,40 +179,10 @@ class Kamal::Configuration
   end
 
   def logging_args
-    if logging.present?
+    logging.args
-      optionize({ "log-driver" => logging["driver"] }.compact) +
-        argumentize("--log-opt", logging["options"])
-    else
-      argumentize("--log-opt", { "max-size" => "10m" })
-    end
   end
 
 
-  def boot
-    Kamal::Configuration::Boot.new(config: self)
-  end
-
-  def builder
-    Kamal::Configuration::Builder.new(config: self)
-  end
-
-  def traefik
-    raw_config.traefik || {}
-  end
-
-  def ssh
-    Kamal::Configuration::Ssh.new(config: self)
-  end
-
-  def sshkit
-    Kamal::Configuration::Sshkit.new(config: self)
-  end
-
-
-  def healthcheck
-    { "path" => "/up", "port" => 3000, "max_attempts" => 7, "exposed_port" => 3999, "cord" => "/tmp/kamal-cord", "log_lines" => 50 }.merge(raw_config.healthcheck || {})
-  end
-
   def healthcheck_service
     [ "healthcheck", service, destination ].compact.join("-")
   end
@@ -216,19 +212,28 @@ class Kamal::Configuration
     raw_config.hooks_path || ".kamal/hooks"
   end
 
-  def host_env_directory
-    "#{run_directory}/env"
-  end
-
   def asset_path
     raw_config.asset_path
   end
 
 
-  def valid?
+  def host_env_directory
-    ensure_destination_if_required && ensure_required_keys_present && ensure_valid_kamal_version && ensure_retain_containers_valid && ensure_valid_service_name
+    File.join(run_directory, "env")
   end
 
+  def env_tags
+    @env_tags ||= if (tags = raw_config.env["tags"])
+      tags.collect { |name, config| Env::Tag.new(name, config: config) }
+    else
+      []
+    end
+  end
+
+  def env_tag(name)
+    env_tags.detect { |t| t.name == name.to_s }
+  end
+
+
   def to_h
     {
       roles: role_names,
@@ -244,11 +249,10 @@ class Kamal::Configuration
       builder: builder.to_h,
       accessories: raw_config.accessories,
       logging: logging_args,
-      healthcheck: healthcheck
+      healthcheck: healthcheck.to_h
     }.compact
   end
 
-
   private
     # Will raise ArgumentError if any required config keys are missing
     def ensure_destination_if_required
@@ -261,29 +265,21 @@ class Kamal::Configuration
 
     def ensure_required_keys_present
       %i[ service image registry servers ].each do |key|
-        raise ArgumentError, "Missing required configuration for #{key}" unless raw_config[key].present?
+        raise Kamal::ConfigurationError, "Missing required configuration for #{key}" unless raw_config[key].present?
       end
 
-      if raw_config.registry["username"].blank?
+      unless role(primary_role_name).present?
-        raise ArgumentError, "You must specify a username for the registry in config/deploy.yml"
+        raise Kamal::ConfigurationError, "The primary_role #{primary_role_name} isn't defined"
-      end
-
-      if raw_config.registry["password"].blank?
-        raise ArgumentError, "You must specify a password for the registry in config/deploy.yml (or set the ENV variable if that's used)"
-      end
-
-      unless role_names.include?(primary_role_name)
-        raise ArgumentError, "The primary_role #{primary_role_name} isn't defined"
       end
 
       if primary_role.hosts.empty?
-        raise ArgumentError, "No servers specified for the #{primary_role.name} primary_role"
+        raise Kamal::ConfigurationError, "No servers specified for the #{primary_role.name} primary_role"
       end
 
       unless allow_empty_roles?
         roles.each do |role|
           if role.hosts.empty?
-            raise ArgumentError, "No servers specified for the #{role.name} role. You can ignore this with allow_empty_roles: true"
+            raise Kamal::ConfigurationError, "No servers specified for the #{role.name} role. You can ignore this with allow_empty_roles: true"
           end
         end
       end
@@ -292,21 +288,21 @@ class Kamal::Configuration
     end
 
     def ensure_valid_service_name
-      raise ArgumentError, "Service name can only include alphanumeric characters, hyphens, and underscores" unless raw_config[:service] =~ /^[a-z0-9_-]+$/
+      raise Kamal::ConfigurationError, "Service name can only include alphanumeric characters, hyphens, and underscores" unless raw_config[:service] =~ /^[a-z0-9_-]+$/i
 
       true
     end
 
     def ensure_valid_kamal_version
       if minimum_version && Gem::Version.new(minimum_version) > Gem::Version.new(Kamal::VERSION)
-        raise ArgumentError, "Current version is #{Kamal::VERSION}, minimum required is #{minimum_version}"
+        raise Kamal::ConfigurationError, "Current version is #{Kamal::VERSION}, minimum required is #{minimum_version}"
       end
 
       true
     end
 
     def ensure_retain_containers_valid
-      raise ArgumentError, "Must retain at least 1 container" if retain_containers < 1
+      raise Kamal::ConfigurationError, "Must retain at least 1 container" if retain_containers < 1
 
       true
     end
@@ -319,7 +315,10 @@ class Kamal::Configuration
     def git_version
       @git_version ||=
         if Kamal::Git.used?
-          [ Kamal::Git.revision, Kamal::Git.uncommitted_changes.present? ? "_uncommitted_#{SecureRandom.hex(8)}" : "" ].join
+          if Kamal::Git.uncommitted_changes.present? && !builder.git_clone?
+            uncommitted_suffix = "_uncommitted_#{SecureRandom.hex(8)}"
+          end
+          [ Kamal::Git.revision, uncommitted_suffix ].compact.join
         else
           raise "Can't use commit hash as version, no git repository found in #{Dir.pwd}"
         end

lib/kamal/configuration/accessory.rb

@@ -1,30 +1,39 @@
 class Kamal::Configuration::Accessory
+  include Kamal::Configuration::Validation
+
   delegate :argumentize, :optionize, to: Kamal::Utils
 
-  attr_accessor :name, :specifics
+  attr_reader :name, :accessory_config, :env
 
   def initialize(name, config:)
-    @name, @config, @specifics = name.inquiry, config, config.raw_config["accessories"][name]
+    @name, @config, @accessory_config = name.inquiry, config, config.raw_config["accessories"][name]
+
+    validate! \
+      accessory_config,
+      example: validation_yml["accessories"]["mysql"],
+      context: "accessories/#{name}",
+      with: Kamal::Configuration::Validator::Accessory
+
+    @env = Kamal::Configuration::Env.new \
+      config: accessory_config.fetch("env", {}),
+      secrets_file: File.join(config.host_env_directory, "accessories", "#{service_name}.env"),
+      context: "accessories/#{name}/env"
   end
 
   def service_name
-    specifics["service"] || "#{config.service}-#{name}"
+    accessory_config["service"] || "#{config.service}-#{name}"
   end
 
   def image
-    specifics["image"]
+    accessory_config["image"]
   end
 
   def hosts
-    if (specifics.keys & ["host", "hosts", "roles"]).size != 1
-      raise ArgumentError, "Specify one of `host`, `hosts` or `roles` for accessory `#{name}`"
-    end
-
     hosts_from_host || hosts_from_hosts || hosts_from_roles
   end
 
   def port
-    if port = specifics["port"]&.to_s
+    if port = accessory_config["port"]&.to_s
       port.include?(":") ? port : "#{port}:#{port}"
     end
   end
@@ -34,42 +43,26 @@ class Kamal::Configuration::Accessory
   end
 
   def labels
-    default_labels.merge(specifics["labels"] || {})
+    default_labels.merge(accessory_config["labels"] || {})
   end
 
   def label_args
     argumentize "--label", labels
   end
 
-  def env
-    specifics["env"] || {}
-  end
-
-  def env_file
-    Kamal::EnvFile.new(env)
-  end
-
-  def host_env_directory
-    File.join config.host_env_directory, "accessories"
-  end
-
-  def host_env_file_path
-    File.join host_env_directory, "#{service_name}.env"
-  end
-
   def env_args
-    argumentize "--env-file", host_env_file_path
+    env.args
   end
 
   def files
-    specifics["files"]&.to_h do |local_to_remote_mapping|
+    accessory_config["files"]&.to_h do |local_to_remote_mapping|
       local_file, remote_file = local_to_remote_mapping.split(":")
       [ expand_local_file(local_file), expand_remote_file(remote_file) ]
     end || {}
   end
 
   def directories
-    specifics["directories"]&.to_h do |host_to_container_mapping|
+    accessory_config["directories"]&.to_h do |host_to_container_mapping|
       host_path, container_path = host_to_container_mapping.split(":")
       [ expand_host_path(host_path), container_path ]
     end || {}
@@ -84,7 +77,7 @@ class Kamal::Configuration::Accessory
   end
 
   def option_args
-    if args = specifics["options"]
+    if args = accessory_config["options"]
       optionize args
     else
       []
@@ -92,7 +85,7 @@ class Kamal::Configuration::Accessory
   end
 
   def cmd
-    specifics["cmd"]
+    accessory_config["cmd"]
   end
 
   private
@@ -111,10 +104,10 @@ class Kamal::Configuration::Accessory
     end
 
     def with_clear_env_loaded
-      (env["clear"] || env).each { |k, v| ENV[k] = v }
+      env.clear.each { |k, v| ENV[k] = v }
       yield
     ensure
-      (env["clear"] || env).each { |k, v| ENV.delete(k) }
+      env.clear.each { |k, v| ENV.delete(k) }
     end
 
     def read_dynamic_file(local_file)
@@ -126,25 +119,25 @@ class Kamal::Configuration::Accessory
     end
 
     def specific_volumes
-      specifics["volumes"] || []
+      accessory_config["volumes"] || []
     end
 
     def remote_files_as_volumes
-      specifics["files"]&.collect do |local_to_remote_mapping|
+      accessory_config["files"]&.collect do |local_to_remote_mapping|
         _, remote_file = local_to_remote_mapping.split(":")
         "#{service_data_directory + remote_file}:#{remote_file}"
       end || []
     end
 
     def remote_directories_as_volumes
-      specifics["directories"]&.collect do |host_to_container_mapping|
+      accessory_config["directories"]&.collect do |host_to_container_mapping|
         host_path, container_path = host_to_container_mapping.split(":")
         [ expand_host_path(host_path), container_path ].join(":")
       end || []
     end
 
     def expand_host_path(host_path)
-      absolute_path?(host_path) ? host_path : "#{service_data_directory}/#{host_path}"
+      absolute_path?(host_path) ? host_path : File.join(service_data_directory, host_path)
     end
 
     def absolute_path?(path)
@@ -156,30 +149,16 @@ class Kamal::Configuration::Accessory
     end
 
     def hosts_from_host
-      if specifics.key?("host")
+      [ accessory_config["host"] ] if accessory_config.key?("host")
-        host = specifics["host"]
-        if host
-          [host]
-        else
-          raise ArgumentError, "Missing host for accessory `#{name}`"
-        end
-      end
     end
 
     def hosts_from_hosts
-      if specifics.key?("hosts")
+      accessory_config["hosts"] if accessory_config.key?("hosts")
-        hosts = specifics["hosts"]
-        if hosts.is_a?(Array)
-          hosts
-        else
-          raise ArgumentError, "Hosts should be an Array for accessory `#{name}`"
-        end
-      end
     end
 
     def hosts_from_roles
-      if specifics.key?("roles")
+      if accessory_config.key?("roles")
-        specifics["roles"].flat_map { |role| config.role(role).hosts }
+        accessory_config["roles"].flat_map { |role| config.role(role).hosts }
       end
     end
 end

lib/kamal/configuration/boot.rb

@@ -1,20 +1,25 @@
 class Kamal::Configuration::Boot
+  include Kamal::Configuration::Validation
+
+  attr_reader :boot_config, :host_count
+
   def initialize(config:)
-    @options = config.raw_config.boot || {}
+    @boot_config = config.raw_config.boot || {}
     @host_count = config.all_hosts.count
+    validate! boot_config
   end
 
   def limit
-    limit = @options["limit"]
+    limit = boot_config["limit"]
 
     if limit.to_s.end_with?("%")
-      [@host_count * limit.to_i / 100, 1].max
+      [ host_count * limit.to_i / 100, 1 ].max
     else
       limit
     end
   end
 
   def wait
-    @options["wait"]
+    boot_config["wait"]
   end
 end

lib/kamal/configuration/builder.rb

@@ -1,67 +1,79 @@
 class Kamal::Configuration::Builder
-  def initialize(config:)
+  include Kamal::Configuration::Validation
-    @options = config.raw_config.builder || {}
-    @image = config.image
-    @server = config.registry["server"]
 
-    valid?
+  attr_reader :config, :builder_config
+  delegate :image, :service, to: :config
+  delegate :server, to: :"config.registry"
+
+  def initialize(config:)
+    @config = config
+    @builder_config = config.raw_config.builder || {}
+    @image = config.image
+    @server = config.registry.server
+    @service = config.service
+
+    validate! builder_config, with: Kamal::Configuration::Validator::Builder
   end
 
   def to_h
-    @options
+    builder_config
   end
 
   def multiarch?
-    @options["multiarch"] != false
+    builder_config["multiarch"] != false
   end
 
   def local?
-    !!@options["local"]
+    !!builder_config["local"]
   end
 
   def remote?
-    !!@options["remote"]
+    !!builder_config["remote"]
   end
 
   def cached?
-    !!@options["cache"]
+    !!builder_config["cache"]
   end
 
   def args
-    @options["args"] || {}
+    builder_config["args"] || {}
   end
 
   def secrets
-    @options["secrets"] || []
+    builder_config["secrets"] || []
   end
 
   def dockerfile
-    @options["dockerfile"] || "Dockerfile"
+    builder_config["dockerfile"] || "Dockerfile"
+  end
+
+  def target
+    builder_config["target"]
   end
 
   def context
-    @options["context"] || "."
+    builder_config["context"] || "."
   end
 
   def local_arch
-    @options["local"]["arch"] if local?
+    builder_config["local"]["arch"] if local?
   end
 
   def local_host
-    @options["local"]["host"] if local?
+    builder_config["local"]["host"] if local?
   end
 
   def remote_arch
-    @options["remote"]["arch"] if remote?
+    builder_config["remote"]["arch"] if remote?
   end
 
   def remote_host
-    @options["remote"]["host"] if remote?
+    builder_config["remote"]["host"] if remote?
   end
 
   def cache_from
     if cached?
-      case @options["cache"]["type"]
+      case builder_config["cache"]["type"]
       when "gha"
         cache_from_config_for_gha
       when "registry"
@@ -72,7 +84,7 @@ class Kamal::Configuration::Builder
 
   def cache_to
     if cached?
-      case @options["cache"]["type"]
+      case builder_config["cache"]["type"]
       when "gha"
         cache_to_config_for_gha
       when "registry"
@@ -82,22 +94,33 @@ class Kamal::Configuration::Builder
   end
 
   def ssh
-    @options["ssh"]
+    builder_config["ssh"]
+  end
+
+  def git_clone?
+    Kamal::Git.used? && builder_config["context"].nil?
+  end
+
+  def clone_directory
+    @clone_directory ||= File.join Dir.tmpdir, "kamal-clones", [ service, pwd_sha ].compact.join("-")
+  end
+
+  def build_directory
+    @build_directory ||=
+      if git_clone?
+        File.join clone_directory, repo_basename, repo_relative_pwd
+      else
+        "."
+      end
   end
 
   private
-    def valid?
-      if @options["cache"] && @options["cache"]["type"]
-        raise ArgumentError, "Invalid cache type: #{@options["cache"]["type"]}" unless ["gha", "registry"].include?(@options["cache"]["type"])
-      end
-    end
-
     def cache_image
-      @options["cache"]&.fetch("image", nil) || "#{@image}-build-cache"
+      builder_config["cache"]&.fetch("image", nil) || "#{image}-build-cache"
     end
 
     def cache_image_ref
-      [ @server, cache_image ].compact.join("/")
+      [ server, cache_image ].compact.join("/")
     end
 
     def cache_from_config_for_gha
@@ -109,10 +132,22 @@ class Kamal::Configuration::Builder
     end
 
     def cache_to_config_for_gha
-      [ "type=gha", @options["cache"]&.fetch("options", nil)].compact.join(",")
+      [ "type=gha", builder_config["cache"]&.fetch("options", nil) ].compact.join(",")
     end
 
     def cache_to_config_for_registry
-      [ "type=registry", @options["cache"]&.fetch("options", nil), "ref=#{cache_image_ref}" ].compact.join(",")
+      [ "type=registry", builder_config["cache"]&.fetch("options", nil), "ref=#{cache_image_ref}" ].compact.join(",")
+    end
+
+    def repo_basename
+      File.basename(Kamal::Git.root)
+    end
+
+    def repo_relative_pwd
+      Dir.pwd.delete_prefix(Kamal::Git.root)
+    end
+
+    def pwd_sha
+      Digest::SHA256.hexdigest(Dir.pwd)[0..12]
     end
 end

lib/kamal/configuration/docs/accessory.yml

@@ -0,0 +1,90 @@
+# Accessories
+#
+# Accessories can be booted on a single host, a list of hosts, or on specific roles.
+# The hosts do not need to be defined in the Kamal servers configuration.
+#
+# Accessories are managed separately from the main service - they are not updated
+# when you deploy and they do not have zero-downtime deployments.
+#
+# Run `kamal accessory boot <accessory>` to boot an accessory.
+# See `kamal accessory --help` for more information.
+
+# Configuring accessories
+#
+# First define the accessory in the `accessories`
+accessories:
+  mysql:
+
+    # Service name
+    #
+    # This is used in the service label and defaults to `<service>-<accessory>`
+    # where `<service>` is the main service name from the root configuration
+    service: mysql
+
+    # Image
+    #
+    # The Docker image to use, prefix with a registry if not using Docker hub
+    image: mysql:8.0
+
+    # Accessory hosts
+    #
+    # Specify one  of `host`, `hosts` or `roles`
+    host: mysql-db1
+    hosts:
+      - mysql-db1
+      - mysql-db2
+    roles:
+      - mysql
+
+    # Custom command
+    #
+    # You can set a custom command to run in the container, if you do not want to use the default
+    cmd: "bin/mysqld"
+
+    # Port mappings
+    #
+    # See https://docs.docker.com/network/, especially note the warning about the security
+    # implications of exposing ports publicly.
+    port: "127.0.0.1:3306:3306"
+
+    # Labels
+    labels:
+      app: myapp
+
+    # Options
+    # These are passed to the Docker run command in the form `--<name> <value>`
+    options:
+      restart: always
+      cpus: 2
+
+    # Environment variables
+    # See kamal docs env for more information
+    env:
+      ...
+
+    # Copying files
+    #
+    # You can specify files to mount into the container.
+    # The format is `local:remote` where `local` is the path to the file on the local machine
+    # and `remote` is the path to the file in the container.
+    #
+    # They will be uploaded from the local repo to the host and then mounted.
+    #
+    # ERB files will be evaluated before being copied.
+    files:
+      - config/my.cnf.erb:/etc/mysql/my.cnf
+      - config/myoptions.cnf:/etc/mysql/myoptions.cnf
+
+    # Directories
+    #
+    # You can specify directories to mount into the container. They will be created on the host
+    # before being mounted
+    directories:
+      - mysql-logs:/var/log/mysql
+
+    # Volumes
+    #
+    # Any other volumes to mount, in addition to the files and directories.
+    # They are not created or copied before mounting
+    volumes:
+      - /path/to/mysql-logs:/var/log/mysql

lib/kamal/configuration/docs/boot.yml

@@ -0,0 +1,19 @@
+# Booting
+#
+# When deploying to large numbers of hosts, you might prefer not to restart your services on every host at the same time.
+#
+# Kamal’s default is to boot new containers on all hosts in parallel. But you can control this with the boot configuration.
+
+# Fixed group sizes
+#
+# Here we boot 2 hosts at a time with a 10 second gap between each group.
+boot:
+  limit: 2
+  wait: 10
+
+# Percentage of hosts
+#
+# Here we boot 25% of the hosts at a time with a 2 second gap between each group.
+boot:
+  limit: 25%
+  wait: 2

lib/kamal/configuration/docs/builder.yml

@@ -0,0 +1,107 @@
+# Builder
+#
+# The builder configuration controls how the application is built with `docker build` or `docker buildx build`
+#
+# If no configuration is specified, Kamal will:
+# 1. Create a buildx context called `kamal-<service>-multiarch`
+# 2. Use `docker buildx build` to build a multiarch image for linux/amd64,linux/arm64 with that context
+#
+# See https://kamal-deploy.org/docs/configuration/builder-examples/ for more information
+
+# Builder options
+#
+# Options go under the builder key in the root configuration.
+builder:
+
+  # Multiarch
+  #
+  # Enables multiarch builds, defaults to `true`
+  multiarch: false
+
+  # Local configuration
+  #
+  # The build configuration for local builds, only used if multiarch is enabled (the default)
+  #
+  # If there is no remote configuration, by default we build for amd64 and arm64.
+  # If you only want to build for one architecture, you can specify it here.
+  # The docker socket is optional and uses the default docker host socket when not specified
+  local:
+    arch: amd64
+    host: /var/run/docker.sock
+
+  # Remote configuration
+  #
+  # The build configuration for remote builds, also only used if multiarch is enabled.
+  # The arch is required and can be either amd64 or arm64.
+  remote:
+    arch: arm64
+    host: ssh://docker@docker-builder
+
+  # Builder cache
+  #
+  # The type must be either 'gha' or 'registry'
+  #
+  # The image is only used for registry cache
+  cache:
+    type: registry
+    options: mode=max
+    image: kamal-app-build-cache
+
+  # Build context
+  #
+  # If this is not set, then a local git clone of the repo is used.
+  # This ensures a clean build with no uncommitted changes.
+  #
+  # To use the local checkout instead you can set the context to `.`, or a path to another directory.
+  context: .
+
+  # Dockerfile
+  #
+  # The Dockerfile to use for building, defaults to `Dockerfile`
+  dockerfile: Dockerfile.production
+
+  # Build target
+  #
+  # If not set, then the default target is used
+  target: production
+
+  # Build Arguments
+  #
+  # Any additional build arguments, passed to `docker build` with `--build-arg <key>=<value>`
+  args:
+    ENVIRONMENT: production
+
+  # Referencing build arguments
+  #
+  # ```shell
+  # ARG RUBY_VERSION
+  # FROM ruby:$RUBY_VERSION-slim as base
+  # ```
+
+  # Build secrets
+  #
+  # Values are read from the environment.
+  #
+  secrets:
+    - SECRET1
+    - SECRET2
+
+  # Referencing Build Secrets
+  #
+  # ```shell
+  # # Copy Gemfiles
+  # COPY Gemfile Gemfile.lock ./
+  #
+  # # Install dependencies, including private repositories via access token
+  # # Then remove bundle cache with exposed GITHUB_TOKEN)
+  # RUN --mount=type=secret,id=GITHUB_TOKEN \
+  #   BUNDLE_GITHUB__COM=x-access-token:$(cat /run/secrets/GITHUB_TOKEN) \
+  #   bundle install && \
+  #   rm -rf /usr/local/bundle/cache
+  # ```
+
+
+  # SSH
+  #
+  # SSH agent socket or keys to expose to the build
+  ssh: default=$SSH_AUTH_SOCK

lib/kamal/configuration/docs/configuration.yml

@@ -0,0 +1,168 @@
+# Kamal Configuration
+#
+# Configuration is read from the `config/deploy.yml`
+#
+
+# Destinations
+#
+# When running commands, you can specify a destination with the `-d` flag,
+# e.g. `kamal deploy -d staging`
+#
+# In this case the configuration will also be read from `config/deploy.staging.yml`
+# and merged with the base configuration.
+
+# Extensions
+#
+# Kamal will not accept unrecognized keys in the configuration file.
+#
+# However, you might want to declare a configuration block using YAML anchors
+# and aliases to avoid repetition.
+#
+# You can use prefix a configuration section with `x-` to indicate that it is an
+# extension. Kamal will ignore the extension and not raise an error.
+
+# The service name
+# This is a required value. It is used as the container name prefix.
+service: myapp
+
+# The Docker image name
+#
+# The image will be pushed to the configured registry.
+image: my-image
+
+# Labels
+#
+# Additional labels to add to the container
+labels:
+  my-label: my-value
+
+# Additional volumes to mount into the container
+volumes:
+  - /path/on/host:/path/in/container:ro
+
+# Registry
+#
+# The Docker registry configuration, see kamal docs registry
+registry:
+  ...
+
+# Servers
+#
+# The servers to deploy to, optionally with custom roles, see kamal docs servers
+servers:
+  ...
+
+# Environment variables
+#
+# See kamal docs env
+env:
+  ...
+
+# Asset Bridging
+#
+# Used for asset bridging across deployments, default to `nil`
+#
+# If there are changes to CSS or JS files, we may get requests
+# for the old versions on the new container and vice-versa.
+#
+# To avoid 404s we can specify an asset path.
+# Kamal will replace that path in the container with a mapped
+# volume containing both sets of files.
+# This requires that file names change when the contents change
+# (e.g. by including a hash of the contents in the name).
+
+# To configure this, set the path to the assets:
+asset_path: /path/to/assets
+
+# Path to hooks, defaults to `.kamal/hooks`
+# See https://kamal-deploy.org/docs/hooks for more information
+hooks_path: /user_home/kamal/hooks
+
+# Require destinations
+#
+# Whether deployments require a destination to be specified, defaults to `false`
+require_destination: true
+
+# The primary role
+#
+# This defaults to `web`, but if you have no web role, you can change this
+primary_role: workers
+
+# Allowing empty roles
+#
+# Whether roles with no servers are allowed. Defaults to `false`.
+allow_empty_roles: false
+
+# Stop wait time
+#
+# How long we wait for a container to stop before killing it, defaults to 30 seconds
+stop_wait_time: 60
+
+# Retain containers
+#
+# How many old containers and images we retain, defaults to 5
+retain_containers: 3
+
+# Minimum version
+#
+# The minimum version of Kamal required to deploy this configuration, defaults to nil
+minimum_version: 1.3.0
+
+# Readiness delay
+#
+# Seconds to wait for a container to boot after is running, default 7
+# This only applies to containers that do not specify a healthcheck
+readiness_delay: 4
+
+# Run directory
+#
+# Directory to store kamal runtime files in on the host, default `.kamal`
+run_directory: /etc/kamal
+
+# SSH options
+#
+# See kamal docs ssh
+ssh:
+  ...
+
+# Builder options
+#
+# See kamal docs builder
+builder:
+  ...
+
+# Accessories
+#
+# Additionals services to run in Docker, see kamal docs accessory
+accessories:
+  ...
+
+# Traefik
+#
+# The Traefik proxy is used for zero-downtime deployments, see kamal docs traefik
+traefik:
+  ...
+
+# SSHKit
+#
+# See kamal docs sshkit
+sshkit:
+  ...
+
+# Boot options
+#
+# See kamal docs boot
+boot:
+  ...
+
+# Healthcheck
+#
+# Configuring healthcheck commands, intervals and timeouts, see kamal docs healthcheck
+healthcheck:
+  ...
+
+# Logging
+#
+# Docker logging configuration, see kamal docs logging
+logging:
+  ...

lib/kamal/configuration/docs/env.yml

@@ -0,0 +1,72 @@
+# Environment variables
+#
+# Environment variables can be set directory in the Kamal configuration or
+# for loaded from a .env file, for secrets that should not be checked into Git.
+
+# Reading environment variables from the configuration
+#
+# Environment variables can be set directly in the configuration file.
+#
+# These are passed to the docker run command when deploying.
+env:
+  DATABASE_HOST: mysql-db1
+  DATABASE_PORT: 3306
+
+# Using .env file to load required environment variables
+#
+# Kamal uses dotenv to automatically load environment variables set in the .env file present
+# in the application root.
+#
+# This file can be used to set variables like KAMAL_REGISTRY_PASSWORD or database passwords.
+# But for this reason you must ensure that .env files are not checked into Git or included
+# in your Dockerfile! The format is just key-value like:
+# ```
+# KAMAL_REGISTRY_PASSWORD=pw
+# DB_PASSWORD=secret123
+# ```
+# See https://kamal-deploy.org/docs/commands/envify/ for how to use generated .env files.
+#
+# To pass the secrets you should list them under the `secret` key. When you do this the
+# other variables need to be moved under the `clear` key.
+#
+# Unlike clear values, secrets are not passed directly to the container,
+# but are stored in an env file on the host
+# The file is not updated when deploying, only when running `kamal envify` or `kamal env push`.
+env:
+  clear:
+    DB_USER: app
+  secret:
+    - DB_PASSWORD
+
+# Tags
+#
+# Tags are used to add extra env variables to specific hosts.
+# See kamal docs servers for how to tag hosts.
+#
+# Tags are only allowed in the top level env configuration (i.e not under a role specific env).
+#
+# The env variables can be specified with secret and clear values as explained above.
+env:
+  tags:
+    <tag1>:
+      MYSQL_USER: monitoring
+    <tag2>:
+      clear:
+        MYSQL_USER: readonly
+      secret:
+        - MYSQL_PASSWORD
+
+# Example configuration
+env:
+  clear:
+    MYSQL_USER: app
+  secret:
+    - MYSQL_PASSWORD
+  tags:
+    monitoring:
+      MYSQL_USER: monitoring
+    replica:
+      clear:
+        MYSQL_USER: readonly
+      secret:
+        - READONLY_PASSWORD

lib/kamal/configuration/docs/healthcheck.yml

@@ -0,0 +1,59 @@
+# Healthcheck configuration
+#
+# On roles that are running Traefik, Kamal will supply a default healthcheck to `docker run`.
+# For other roles, by default no healthcheck is supplied.
+#
+# If no healthcheck is supplied and the image does not define one, they we wait for the container
+# to reach a running state and then pause for the readiness delay.
+#
+# The default healthcheck is `curl -f http://localhost:<port>/<path>`, so it assumes that `curl`
+# is available within the container.
+
+# Healthcheck options
+#
+# These go under the `healthcheck` key in the root or role configuration.
+healthcheck:
+
+  # Command
+  #
+  # The command to run, defaults to `curl -f http://localhost:<port>/<path>` on roles running Traefik
+  cmd: "curl -f http://localhost"
+
+  # Interval
+  #
+  # The Docker healthcheck interval, defaults to `1s`
+  interval: 10s
+
+  # Max attempts
+  #
+  # The maximum number of times we poll the container to see if it is healthy, defaults to `7`
+  # Each check is separated by an increasing interval starting with 1 second.
+  max_attempts: 3
+
+  # Port
+  #
+  # The port to use in the healthcheck, defaults to `3000`
+  port: "80"
+
+  # Path
+  #
+  # The path to use in the healthcheck, defaults to `/up`
+  path: /health
+
+  # Cords for zero-downtime deployments
+  #
+  # The cord file is used for zero-downtime deployments. The healthcheck is augmented with a check
+  # for the existance of the file. This allows us to delete the file and force the container to
+  # become unhealthy, causing Traefik to stop routing traffic to it.
+  #
+  # Kamal mounts a volume at this location and creates the file before starting the container.
+  # You can set the value to `false` to disable the cord file, but this loses the zero-downtime
+  # guarantee.
+  #
+  # The default value is `/tmp/kamal-cord`
+  cord: /cord
+
+  # Log lines
+  #
+  # Number of lines to log from the container when the healthcheck fails, defaults to `50`
+  log_lines: 100

lib/kamal/configuration/docs/logging.yml

@@ -0,0 +1,21 @@
+# Custom logging configuration
+#
+# Set these to control the Docker logging driver and options.
+
+# Logging settings
+#
+# These go under the logging key in the configuration file.
+#
+# This can be specified in the root level or for a specific role.
+logging:
+
+  # Driver
+  #
+  # The logging driver to use, passed to Docker via `--log-driver`
+  driver: json-file
+
+  # Options
+  #
+  # Any logging options to pass to the driver, passed to Docker via `--log-opt`
+  options:
+    max-size: 100m

lib/kamal/configuration/docs/registry.yml

@@ -0,0 +1,49 @@
+# Registry
+#
+# The default registry is Docker Hub, but you can change it using registry/server:
+#
+# A reference to secret (in this case DOCKER_REGISTRY_TOKEN) will look up the secret
+# in the local environment.
+
+registry:
+  server: registry.digitalocean.com
+  username:
+    - DOCKER_REGISTRY_TOKEN
+  password:
+    - DOCKER_REGISTRY_TOKEN
+
+# Using AWS ECR as the container registry
+# You will need to have the aws CLI installed locally for this to work.
+# AWS ECR’s access token is only valid for 12hrs. In order to not have to manually regenerate the token every time, you can use ERB in the deploy.yml file to shell out to the aws cli command, and obtain the token:
+
+registry:
+  server: <your aws account id>.dkr.ecr.<your aws region id>.amazonaws.com
+  username: AWS
+  password: <%= %x(aws ecr get-login-password) %>
+
+# Using GCP Artifact Registry as the container registry
+# To sign into Artifact Registry, you would need to
+# [create a service account](https://cloud.google.com/iam/docs/service-accounts-create#creating)
+# and [set up roles and permissions](https://cloud.google.com/artifact-registry/docs/access-control#permissions).
+# Normally, assigning a roles/artifactregistry.writer role should be sufficient.
+#
+# Once the service account is ready, you need to generate and download a JSON key, base64 encode it and add to .env:
+#
+# ```shell
+# echo "KAMAL_REGISTRY_PASSWORD=$(base64 -i /path/to/key.json)" | tr -d "\\n"  >> .env
+# ```
+# Use the env variable as password along with _json_key_base64 as username.
+# Here’s the final configuration:
+
+registry:
+  server: <your registry region>-docker.pkg.dev
+  username: _json_key_base64
+  password:
+    - KAMAL_REGISTRY_PASSWORD
+
+# Validating the configuration
+#
+# You can validate the configuration by running:
+# ```shell
+# kamal registry login
+# ```

lib/kamal/configuration/docs/role.yml

@@ -0,0 +1,52 @@
+# Roles
+#
+# Roles are used to configure different types of servers in the deployment.
+# The most common use for this is to run a web servers and job servers.
+#
+# Kamal expects there to be a `web` role, unless you set a different `primary_role`
+# in the root configuration.
+
+# Role configuration
+#
+# Roles are specified under the servers key
+servers:
+
+  # Simple role configuration
+  #
+  #
+  # This can be a list of hosts, if you don't need custom configuration for the role.
+  #
+  # You can set tags on the hosts for custom env variables (see kamal docs env)
+  web:
+    - 172.1.0.1
+    - 172.1.0.2: experiment1
+    - 172.1.0.2: [ experiment1, experiment2 ]
+
+  # Custom role configuration
+  #
+  # When there are other options to set, the list of hosts goes under the `hosts` key
+  #
+  # By default only the primary role uses Traefik, but you can set `traefik` to change
+  # it.
+  #
+  # You can also set a custom cmd to run in the container, and overwrite other settings
+  # from the root configuration.
+  workers:
+    hosts:
+      - 172.1.0.3
+      - 172.1.0.4: experiment1
+    traefik: true
+    cmd: "bin/jobs"
+    options:
+      memory: 2g
+      cpus: 4
+    healthcheck:
+      ...
+    logging:
+      ...
+    labels:
+      my-label: workers
+    env:
+      ...
+    asset_path: /public
+

lib/kamal/configuration/docs/servers.yml

@@ -0,0 +1,27 @@
+# Servers
+#
+# Servers are split into different roles, with each role having its own configuration.
+#
+# For simpler deployments though where all servers are identical, you can just specify a list of servers
+# They will be implicitly assigned to the `web` role.
+servers:
+  - 172.0.0.1
+  - 172.0.0.2
+  - 172.0.0.3
+
+# Tagging servers
+#
+# Servers can be tagged, with the tags used to add custom env variables (see kamal docs env).
+servers:
+  - 172.0.0.1
+  - 172.0.0.2: experiments
+  - 172.0.0.3: [ experiments, three ]
+
+# Roles
+#
+# For more complex deployments (e.g. if you are running job hosts), you can specify roles, and configure each separately (see kamal docs role)
+servers:
+  web:
+    ...
+  workers:
+    ...

lib/kamal/configuration/docs/ssh.yml

@@ -0,0 +1,66 @@
+# SSH configuration
+#
+# Kamal uses SSH to connect run commands on your hosts.
+# By default it will attempt to connect to the root user on port 22
+#
+# If you are using non-root user, you may need to bootstrap your servers manually, before using them with Kamal. On Ubuntu, you’d do:
+#
+# ```shell
+# sudo apt update
+# sudo apt upgrade -y
+# sudo apt install -y docker.io curl git
+# sudo usermod -a -G docker app
+# ```
+
+
+# SSH options
+#
+# The options are specified under the ssh key in the configuration file.
+ssh:
+
+  # The SSH user
+  #
+  # Defaults to `root`
+  #
+  user: app
+
+  # The SSH port
+  #
+  # Defaults to 22
+  port: "2222"
+
+  # Proxy host
+  #
+  # Specified in the form <host> or <user>@<host>
+  proxy: root@proxy-host
+
+  # Proxy command
+  #
+  # A custom proxy command, required for older versions of SSH
+  proxy_command: "ssh -W %h:%p user@proxy"
+
+  # Log level
+  #
+  # Defaults to `fatal`. Set this to debug if you are having
+  # SSH connection issues.
+  log_level: debug
+
+  # Keys Only
+  #
+  # Set to true to use only private keys from keys and key_data parameters, 
+  # even if ssh-agent offers more identities. This option is intended for 
+  # situations where ssh-agent offers many different identites or you have 
+  # a need to overwrite all identites and force a single one.
+  keys_only: false
+
+  # Keys
+  #
+  # An array of file names of private keys to use for publickey
+  # and hostbased authentication
+  keys: [ "~/.ssh/id.pem" ]
+
+  # Key Data
+  #
+  # An array of strings, with each element of the array being
+  # a raw private key in PEM format.
+  key_data: [ "-----BEGIN OPENSSH PRIVATE KEY-----" ]

lib/kamal/configuration/docs/sshkit.yml

@@ -0,0 +1,23 @@
+# SSHKit
+#
+# [SSHKit](https://github.com/capistrano/sshkit) is the SSH toolkit used by Kamal.
+#
+# The default settings should be sufficient for most use cases, but
+# when connecting to a large number of hosts you may need to adjust
+
+# SSHKit options
+#
+# The options are specified under the sshkit key in the configuration file.
+sshkit:
+
+  # Max concurrent starts
+  #
+  # Creating SSH connections concurrently can be an issue when deploying to many servers.
+  # By default Kamal will limit concurrent connection starts to 30 at a time.
+  max_concurrent_starts: 10
+
+  # Pool idle timeout
+  #
+  # Kamal sets a long idle timeout of 900 seconds on connections to try to avoid
+  # re-connection storms after an idle period, like building an image or waiting for CI.
+  pool_idle_timeout: 300

lib/kamal/configuration/docs/traefik.yml

@@ -0,0 +1,62 @@
+# Traefik
+#
+# Traefik is a reverse proxy, used by Kamal for zero-downtime deployments.
+#
+# We start an instance on the hosts in it's own container.
+#
+# During a deployment:
+# 1. We start a new container which Traefik automatically detects due to the labels we have applied
+# 2. Traefik starts routing traffic to the new container
+# 3. We force the old container to fail it's healthcheck, causing Traefik to stop routing traffic to it
+# 4. We stop the old container
+
+# Traefik settings
+#
+# Traekik is configured in the root configuration under `traefik`.
+traefik:
+
+  # Image
+  #
+  # The Traefik image to use, defaults to `traefik:v2.10`
+  image: traefik:v2.9
+
+  # Host port
+  #
+  # The host port to publish the Traefik container on, defaults to `80`
+  host_port: "8080"
+
+  # Disabling publishing
+  #
+  # To avoid publishing the Traefik container, set this to `false`
+  publish: false
+
+  # Labels
+  #
+  # Additional labels to apply to the Traefik container
+  labels:
+    traefik.http.routers.catchall.entryPoints: http
+    traefik.http.routers.catchall.rule: PathPrefix(`/`)
+    traefik.http.routers.catchall.service: unavailable
+    traefik.http.routers.catchall.priority: "1"
+    traefik.http.services.unavailable.loadbalancer.server.port: "0"
+
+  # Arguments
+  #
+  # Additional arguments to pass to the Traefik container
+  args:
+    entryPoints.http.address: ":80"
+    entryPoints.http.forwardedHeaders.insecure: true
+    accesslog: true
+    accesslog.format: json
+
+  # Options
+  #
+  # Additional options to pass to `docker run`
+  options:
+    cpus: 2
+
+  # Environment variables
+  #
+  # See kamal docs env
+  env:
+    ...

lib/kamal/configuration/env.rb

@@ -0,0 +1,36 @@
+class Kamal::Configuration::Env
+  include Kamal::Configuration::Validation
+
+  attr_reader :secrets_keys, :clear, :secrets_file, :context
+  delegate :argumentize, to: Kamal::Utils
+
+  def initialize(config:, secrets_file: nil, context: "env")
+    @clear = config.fetch("clear", config.key?("secret") || config.key?("tags") ? {} : config)
+    @secrets_keys = config.fetch("secret", [])
+    @secrets_file = secrets_file
+    @context = context
+    validate! config, context: context, with: Kamal::Configuration::Validator::Env
+  end
+
+  def args
+    [ "--env-file", secrets_file, *argumentize("--env", clear) ]
+  end
+
+  def secrets_io
+    StringIO.new(Kamal::EnvFile.new(secrets).to_s)
+  end
+
+  def secrets
+    @secrets ||= secrets_keys.to_h { |key| [ key, ENV.fetch(key) ] }
+  end
+
+  def secrets_directory
+    File.dirname(secrets_file)
+  end
+
+  def merge(other)
+    self.class.new \
+      config: { "clear" => clear.merge(other.clear), "secret" => secrets_keys | other.secrets_keys },
+      secrets_file: secrets_file || other.secrets_file
+  end
+end

lib/kamal/configuration/env/tag.rb

@@ -0,0 +1,12 @@
+class Kamal::Configuration::Env::Tag
+  attr_reader :name, :config
+
+  def initialize(name, config:)
+    @name = name
+    @config = config
+  end
+
+  def env
+    Kamal::Configuration::Env.new(config: config)
+  end
+end

lib/kamal/configuration/healthcheck.rb

@@ -0,0 +1,63 @@
+class Kamal::Configuration::Healthcheck
+  include Kamal::Configuration::Validation
+
+  attr_reader :healthcheck_config
+
+  def initialize(healthcheck_config:, context: "healthcheck")
+    @healthcheck_config = healthcheck_config || {}
+    validate! @healthcheck_config, context: context
+  end
+
+  def merge(other)
+    self.class.new healthcheck_config: healthcheck_config.deep_merge(other.healthcheck_config)
+  end
+
+  def cmd
+    healthcheck_config.fetch("cmd", http_health_check)
+  end
+
+  def port
+    healthcheck_config.fetch("port", 3000)
+  end
+
+  def path
+    healthcheck_config.fetch("path", "/up")
+  end
+
+  def max_attempts
+    healthcheck_config.fetch("max_attempts", 7)
+  end
+
+  def interval
+    healthcheck_config.fetch("interval", "1s")
+  end
+
+  def cord
+    healthcheck_config.fetch("cord", "/tmp/kamal-cord")
+  end
+
+  def log_lines
+    healthcheck_config.fetch("log_lines", 50)
+  end
+
+  def set_port_or_path?
+    healthcheck_config["port"].present? || healthcheck_config["path"].present?
+  end
+
+  def to_h
+    {
+      "cmd" => cmd,
+      "interval" => interval,
+      "max_attempts" => max_attempts,
+      "port" => port,
+      "path" => path,
+      "cord" => cord,
+      "log_lines" => log_lines
+    }
+  end
+
+  private
+    def http_health_check
+      "curl -f #{URI.join("http://localhost:#{port}", path)} || exit 1" if path.present? || port.present?
+    end
+end

lib/kamal/configuration/logging.rb

@@ -0,0 +1,33 @@
+class Kamal::Configuration::Logging
+  delegate :optionize, :argumentize, to: Kamal::Utils
+
+  include Kamal::Configuration::Validation
+
+  attr_reader :logging_config
+
+  def initialize(logging_config:, context: "logging")
+    @logging_config = logging_config || {}
+    validate! @logging_config, context: context
+  end
+
+  def driver
+    logging_config["driver"]
+  end
+
+  def options
+    logging_config.fetch("options", {})
+  end
+
+  def merge(other)
+    self.class.new logging_config: logging_config.deep_merge(other.logging_config)
+  end
+
+  def args
+    if driver.present? || options.present?
+      optionize({ "log-driver" => driver }.compact) +
+        argumentize("--log-opt", options)
+    else
+      argumentize("--log-opt", { "max-size" => "10m" })
+    end
+  end
+end

lib/kamal/configuration/registry.rb

@@ -0,0 +1,31 @@
+class Kamal::Configuration::Registry
+  include Kamal::Configuration::Validation
+
+  attr_reader :registry_config
+
+  def initialize(config:)
+    @registry_config = config.raw_config.registry || {}
+    validate! registry_config, with: Kamal::Configuration::Validator::Registry
+  end
+
+  def server
+    registry_config["server"]
+  end
+
+  def username
+    lookup("username")
+  end
+
+  def password
+    lookup("password")
+  end
+
+  private
+    def lookup(key)
+      if registry_config[key].is_a?(Array)
+        ENV.fetch(registry_config[key].first).dup
+      else
+        registry_config[key]
+      end
+    end
+end

lib/kamal/configuration/role.rb

@@ -1,12 +1,33 @@
 class Kamal::Configuration::Role
+  include Kamal::Configuration::Validation
+
   CORD_FILE = "cord"
   delegate :argumentize, :optionize, to: Kamal::Utils
 
-  attr_accessor :name
+  attr_reader :name, :config, :specialized_env, :specialized_logging, :specialized_healthcheck
+
   alias to_s name
 
   def initialize(name, config:)
     @name, @config = name.inquiry, config
+    validate! \
+      specializations,
+      example: validation_yml["servers"]["workers"],
+      context: "servers/#{name}",
+      with: Kamal::Configuration::Validator::Role
+
+    @specialized_env = Kamal::Configuration::Env.new \
+      config: specializations.fetch("env", {}),
+      secrets_file: File.join(config.host_env_directory, "roles", "#{container_prefix}.env"),
+      context: "servers/#{name}/env"
+
+    @specialized_logging = Kamal::Configuration::Logging.new \
+      logging_config: specializations.fetch("logging", {}),
+      context: "servers/#{name}/logging"
+
+    @specialized_healthcheck = Kamal::Configuration::Healthcheck.new \
+      healthcheck_config: specializations.fetch("healthcheck", {}),
+      context: "servers/#{name}/healthcheck"
   end
 
   def primary_host
@@ -14,7 +35,11 @@ class Kamal::Configuration::Role
   end
 
   def hosts
-    @hosts ||= extract_hosts_from_config
+    tagged_hosts.keys
+  end
+
+  def env_tags(host)
+    tagged_hosts.fetch(host).collect { |tag| config.env_tag(tag) }
   end
 
   def cmd
@@ -38,40 +63,21 @@ class Kamal::Configuration::Role
   end
 
   def logging_args
-    args = config.logging || {}
+    logging.args
-    args.deep_merge!(specializations["logging"]) if specializations["logging"].present?
+  end
 
-    if args.any?
+  def logging
-      optionize({ "log-driver" => args["driver"] }.compact) +
+    @logging ||= config.logging.merge(specialized_logging)
-        argumentize("--log-opt", args["options"])
-    else
-      config.logging_args
-    end
   end
 
 
-  def env
+  def env(host)
-    if config.env && config.env["secret"]
+    @envs ||= {}
-      merged_env_with_secrets
+    @envs[host] ||= [ config.env, specialized_env, *env_tags(host).map(&:env) ].reduce(:merge)
-    else
-      merged_env
-    end
   end
 
-  def env_file
+  def env_args(host)
-    Kamal::EnvFile.new(env)
+    env(host).args
-  end
-
-  def host_env_directory
-    File.join config.host_env_directory, "roles"
-  end
-
-  def host_env_file_path
-    File.join host_env_directory, "#{[config.service, name, config.destination].compact.join("-")}.env"
-  end
-
-  def env_args
-    argumentize "--env-file", host_env_file_path
   end
 
   def asset_volume_args
@@ -80,28 +86,29 @@ class Kamal::Configuration::Role
 
 
   def health_check_args(cord: true)
-    if health_check_cmd.present?
+    if running_traefik? || healthcheck.set_port_or_path?
       if cord && uses_cord?
-        optionize({ "health-cmd" => health_check_cmd_with_cord, "health-interval" => health_check_interval })
+        optionize({ "health-cmd" => health_check_cmd_with_cord, "health-interval" => healthcheck.interval })
           .concat(cord_volume.docker_args)
       else
-        optionize({ "health-cmd" => health_check_cmd, "health-interval" => health_check_interval })
+        optionize({ "health-cmd" => healthcheck.cmd, "health-interval" => healthcheck.interval })
       end
     else
       []
     end
   end
 
-  def health_check_cmd
+  def healthcheck
-    health_check_options["cmd"] || http_health_check(port: health_check_options["port"], path: health_check_options["path"])
+    @healthcheck ||=
+      if running_traefik?
+        config.healthcheck.merge(specialized_healthcheck)
+      else
+        specialized_healthcheck
+      end
   end
 
   def health_check_cmd_with_cord
-    "(#{health_check_cmd}) && (stat #{cord_container_file} > /dev/null || exit 1)"
+    "(#{healthcheck.cmd}) && (stat #{cord_container_file} > /dev/null || exit 1)"
-  end
-
-  def health_check_interval
-    health_check_options["interval"] || "1s"
   end
 
 
@@ -119,17 +126,17 @@ class Kamal::Configuration::Role
 
 
   def uses_cord?
-    running_traefik? && cord_volume && health_check_cmd.present?
+    running_traefik? && cord_volume && healthcheck.cmd.present?
   end
 
   def cord_host_directory
-    File.join config.run_directory_as_docker_volume, "cords", [container_prefix, config.run_id].join("-")
+    File.join config.run_directory_as_docker_volume, "cords", [ container_prefix, config.run_id ].join("-")
   end
 
   def cord_volume
-    if (cord = health_check_options["cord"])
+    if (cord = healthcheck.cord)
       @cord_volume ||= Kamal::Configuration::Volume.new \
-        host_path: File.join(config.run_directory, "cords", [container_prefix, config.run_id].join("-")),
+        host_path: File.join(config.run_directory, "cords", [ container_prefix, config.run_id ].join("-")),
         container_path: cord
     end
   end
@@ -180,22 +187,37 @@ class Kamal::Configuration::Role
   end
 
   private
-    attr_accessor :config
+    def tagged_hosts
+      {}.tap do |tagged_hosts|
+        extract_hosts_from_config.map do |host_config|
+          if host_config.is_a?(Hash)
+            host, tags = host_config.first
+            tagged_hosts[host] = Array(tags)
+          elsif host_config.is_a?(String)
+            tagged_hosts[host_config] = []
+          end
+        end
+      end
+    end
 
     def extract_hosts_from_config
-      if config.servers.is_a?(Array)
+      if config.raw_config.servers.is_a?(Array)
-        config.servers
+        config.raw_config.servers
       else
-        servers = config.servers[name]
+        servers = config.raw_config.servers[name]
         servers.is_a?(Array) ? servers : Array(servers["hosts"])
       end
     end
 
     def default_labels
-      if config.destination
+      { "service" => config.service, "role" => name, "destination" => config.destination }
-        { "service" => config.service, "role" => name, "destination" => config.destination }
+    end
+
+    def specializations
+      if config.raw_config.servers.is_a?(Array) || config.raw_config.servers[name].is_a?(Array)
+        {}
       else
-        { "service" => config.service, "role" => name }
+        config.raw_config.servers[name]
       end
     end
 
@@ -217,7 +239,7 @@ class Kamal::Configuration::Role
     end
 
     def traefik_service
-      [ config.service, name, config.destination ].compact.join("-")
+      container_prefix
     end
 
     def custom_labels
@@ -226,45 +248,4 @@ class Kamal::Configuration::Role
         labels.merge!(specializations["labels"]) if specializations["labels"].present?
       end
     end
-
-    def specializations
-      if config.servers.is_a?(Array) || config.servers[name].is_a?(Array)
-        { }
-      else
-        config.servers[name].except("hosts")
-      end
-    end
-
-    def specialized_env
-      specializations["env"] || {}
-    end
-
-    def merged_env
-      config.env&.merge(specialized_env) || {}
-    end
-
-    # Secrets are stored in an array, which won't merge by default, so have to do it by hand.
-    def merged_env_with_secrets
-      merged_env.tap do |new_env|
-        new_env["secret"] = Array(config.env["secret"]) + Array(specialized_env["secret"])
-
-        # If there's no secret/clear split, everything is clear
-        clear_app_env  = config.env["secret"] ? Array(config.env["clear"]) : Array(config.env["clear"] || config.env)
-        clear_role_env = specialized_env["secret"] ? Array(specialized_env["clear"]) : Array(specialized_env["clear"] || specialized_env)
-
-        new_env["clear"] = clear_app_env.to_h.merge(clear_role_env.to_h)
-      end
-    end
-
-    def http_health_check(port:, path:)
-      "curl -f #{URI.join("http://localhost:#{port}", path)} || exit 1" if path.present? || port.present?
-    end
-
-    def health_check_options
-      @health_check_options ||= begin
-        options = specializations["healthcheck"] || {}
-        options = config.healthcheck.merge(options) if running_traefik?
-        options
-      end
-    end
 end

lib/kamal/configuration/servers.rb

@@ -0,0 +1,18 @@
+class Kamal::Configuration::Servers
+  include Kamal::Configuration::Validation
+
+  attr_reader :config, :servers_config, :roles
+
+  def initialize(config:)
+    @config = config
+    @servers_config = config.raw_config.servers
+    validate! servers_config, with: Kamal::Configuration::Validator::Servers
+
+    @roles = role_names.map { |role_name| Kamal::Configuration::Role.new role_name, config: config }
+  end
+
+  private
+    def role_names
+      servers_config.is_a?(Array) ? [ "web" ] : servers_config.keys.sort
+    end
+end

lib/kamal/configuration/ssh.rb

@@ -1,28 +1,45 @@
 class Kamal::Configuration::Ssh
   LOGGER = ::Logger.new(STDERR)
 
+  include Kamal::Configuration::Validation
+
+  attr_reader :ssh_config
+
   def initialize(config:)
-    @config = config.raw_config.ssh || {}
+    @ssh_config = config.raw_config.ssh || {}
+    validate! ssh_config
   end
 
   def user
-    config.fetch("user", "root")
+    ssh_config.fetch("user", "root")
   end
 
   def port
-    config.fetch("port", 22)
+    ssh_config.fetch("port", 22)
   end
 
   def proxy
-    if (proxy = config["proxy"])
+    if (proxy = ssh_config["proxy"])
       Net::SSH::Proxy::Jump.new(proxy.include?("@") ? proxy : "root@#{proxy}")
-    elsif (proxy_command = config["proxy_command"])
+    elsif (proxy_command = ssh_config["proxy_command"])
       Net::SSH::Proxy::Command.new(proxy_command)
     end
   end
 
+  def keys_only
+    ssh_config["keys_only"]
+  end
+
+  def keys
+    ssh_config["keys"]
+  end
+
+  def key_data
+    ssh_config["key_data"]
+  end
+
   def options
-    { user: user, port: port, proxy: proxy, logger: logger, keepalive: true, keepalive_interval: 30 }.compact
+    { user: user, port: port, proxy: proxy, logger: logger, keepalive: true, keepalive_interval: 30, keys_only: keys_only, keys: keys, key_data: key_data }.compact
   end
 
   def to_h
@@ -30,13 +47,11 @@ class Kamal::Configuration::Ssh
   end
 
   private
-    attr_accessor :config
-
     def logger
       LOGGER.tap { |logger| logger.level = log_level }
     end
 
     def log_level
-      config.fetch("log_level", :fatal)
+      ssh_config.fetch("log_level", :fatal)
     end
 end

lib/kamal/configuration/sshkit.rb

@@ -1,20 +1,22 @@
 class Kamal::Configuration::Sshkit
+  include Kamal::Configuration::Validation
+
+  attr_reader :sshkit_config
+
   def initialize(config:)
-    @options = config.raw_config.sshkit || {}
+    @sshkit_config = config.raw_config.sshkit || {}
+    validate! sshkit_config
   end
 
   def max_concurrent_starts
-    options.fetch("max_concurrent_starts", 30)
+    sshkit_config.fetch("max_concurrent_starts", 30)
   end
 
   def pool_idle_timeout
-    options.fetch("pool_idle_timeout", 900)
+    sshkit_config.fetch("pool_idle_timeout", 900)
   end
 
   def to_h
-    options
+    sshkit_config
   end
-
-  private
-    attr_accessor :options
 end

lib/kamal/configuration/traefik.rb

@@ -0,0 +1,60 @@
+class Kamal::Configuration::Traefik
+  DEFAULT_IMAGE = "traefik:v2.10"
+  CONTAINER_PORT = 80
+  DEFAULT_ARGS = {
+    "log.level" => "DEBUG"
+  }
+  DEFAULT_LABELS = {
+    # These ensure we serve a 502 rather than a 404 if no containers are available
+    "traefik.http.routers.catchall.entryPoints" => "http",
+    "traefik.http.routers.catchall.rule" => "PathPrefix(`/`)",
+    "traefik.http.routers.catchall.service" => "unavailable",
+    "traefik.http.routers.catchall.priority" => 1,
+    "traefik.http.services.unavailable.loadbalancer.server.port" => "0"
+  }
+
+  include Kamal::Configuration::Validation
+
+  attr_reader :config, :traefik_config
+
+  def initialize(config:)
+    @config = config
+    @traefik_config = config.raw_config.traefik || {}
+    validate! traefik_config
+  end
+
+  def publish?
+    traefik_config["publish"] != false
+  end
+
+  def labels
+    DEFAULT_LABELS.merge(traefik_config["labels"] || {})
+  end
+
+  def env
+    Kamal::Configuration::Env.new \
+      config: traefik_config.fetch("env", {}),
+      secrets_file: File.join(config.host_env_directory, "traefik", "traefik.env"),
+      context: "traefik/env"
+  end
+
+  def host_port
+    traefik_config.fetch("host_port", CONTAINER_PORT)
+  end
+
+  def options
+    traefik_config.fetch("options", {})
+  end
+
+  def port
+    "#{host_port}:#{CONTAINER_PORT}"
+  end
+
+  def args
+    DEFAULT_ARGS.merge(traefik_config.fetch("args", {}))
+  end
+
+  def image
+    traefik_config.fetch("image", DEFAULT_IMAGE)
+  end
+end

lib/kamal/configuration/validation.rb

@@ -0,0 +1,27 @@
+require "yaml"
+require "active_support/inflector"
+
+module Kamal::Configuration::Validation
+  extend ActiveSupport::Concern
+
+  class_methods do
+    def validation_doc
+      @validation_doc ||= File.read(File.join(File.dirname(__FILE__), "docs", "#{validation_config_key}.yml"))
+    end
+
+    def validation_config_key
+      @validation_config_key ||= name.demodulize.underscore
+    end
+  end
+
+  def validate!(config, example: nil, context: nil, with: Kamal::Configuration::Validator)
+    context ||= self.class.validation_config_key
+    example ||= validation_yml[self.class.validation_config_key]
+
+    with.new(config, example: example, context: context).validate!
+  end
+
+  def validation_yml
+    @validation_yml ||= YAML.load(self.class.validation_doc)
+  end
+end

lib/kamal/configuration/validator.rb

@@ -0,0 +1,153 @@
+class Kamal::Configuration::Validator
+  attr_reader :config, :example, :context
+
+  def initialize(config, example:, context:)
+    @config = config
+    @example = example
+    @context = context
+  end
+
+  def validate!
+    validate_against_example! config, example
+  end
+
+  private
+    def validate_against_example!(validation_config, example)
+      validate_type! validation_config, Hash
+
+      check_unknown_keys! validation_config, example
+
+      validation_config.each do |key, value|
+        next if extension?(key)
+        with_context(key) do
+          example_value = example[key]
+
+          if example_value == "..."
+            validate_type! value, *(Array if key == :servers), Hash
+          elsif key == "hosts"
+            validate_servers! value
+          elsif example_value.is_a?(Array)
+            validate_array_of! value, example_value.first.class
+          elsif example_value.is_a?(Hash)
+            case key.to_s
+            when "options", "args"
+              validate_type! value, Hash
+            when "labels"
+              validate_hash_of! value, example_value.first[1].class
+            else
+              validate_against_example! value, example_value
+            end
+          else
+            validate_type! value, example_value.class
+          end
+        end
+      end
+    end
+
+
+    def valid_type?(value, type)
+      value.is_a?(type) ||
+        (type == String && stringish?(value)) ||
+        (boolean?(type) && boolean?(value.class))
+    end
+
+    def type_description(type)
+      if type == Integer || type == Array
+        "an #{type.name.downcase}"
+      elsif type == TrueClass || type == FalseClass
+        "a boolean"
+      else
+        "a #{type.name.downcase}"
+      end
+    end
+
+    def boolean?(type)
+      type == TrueClass || type == FalseClass
+    end
+
+    def stringish?(value)
+      value.is_a?(String) || value.is_a?(Symbol) || value.is_a?(Numeric) || value.is_a?(TrueClass) || value.is_a?(FalseClass)
+    end
+
+    def validate_array_of!(array, type)
+      validate_type! array, Array
+
+      array.each_with_index do |value, index|
+        with_context(index) do
+          validate_type! value, type
+        end
+      end
+    end
+
+    def validate_hash_of!(hash, type)
+      validate_type! hash, Hash
+
+      hash.each do |key, value|
+        with_context(key) do
+          validate_type! value, type
+        end
+      end
+    end
+
+    def validate_servers!(servers)
+      validate_type! servers, Array
+
+      servers.each_with_index do |server, index|
+        with_context(index) do
+          validate_type! server, String, Hash
+
+          if server.is_a?(Hash)
+            error "multiple hosts found" unless server.size == 1
+            host, tags = server.first
+
+            with_context(host) do
+              validate_type! tags, String, Array
+              validate_array_of! tags, String if tags.is_a?(Array)
+            end
+          end
+        end
+      end
+    end
+
+    def validate_type!(value, *types)
+      type_error(*types) unless types.any? { |type| valid_type?(value, type) }
+    end
+
+    def error(message)
+      raise Kamal::ConfigurationError, "#{error_context}#{message}"
+    end
+
+    def type_error(*expected_types)
+      error "should be #{expected_types.map { |type| type_description(type) }.join(" or ")}"
+    end
+
+    def unknown_keys_error(unknown_keys)
+      error "unknown #{"key".pluralize(unknown_keys.count)}: #{unknown_keys.join(", ")}"
+    end
+
+    def error_context
+      "#{context}: " if context.present?
+    end
+
+    def with_context(context)
+      old_context = @context
+      @context = [ @context, context ].select(&:present?).join("/")
+      yield
+    ensure
+      @context = old_context
+    end
+
+    def allow_extensions?
+      false
+    end
+
+    def extension?(key)
+      key.to_s.start_with?("x-")
+    end
+
+    def check_unknown_keys!(config, example)
+      unknown_keys = config.keys - example.keys
+      unknown_keys.reject! { |key| extension?(key) } if allow_extensions?
+      unknown_keys_error unknown_keys if unknown_keys.present?
+    end
+end

lib/kamal/configuration/validator/accessory.rb

@@ -0,0 +1,9 @@
+class Kamal::Configuration::Validator::Accessory < Kamal::Configuration::Validator
+  def validate!
+    super
+
+    if (config.keys & [ "host", "hosts", "roles" ]).size != 1
+      error "specify one of `host`, `hosts` or `roles`"
+    end
+  end
+end

lib/kamal/configuration/validator/builder.rb

@@ -0,0 +1,9 @@
+class Kamal::Configuration::Validator::Builder < Kamal::Configuration::Validator
+  def validate!
+    super
+
+    if config["cache"] && config["cache"]["type"]
+      error "Invalid cache type: #{config["cache"]["type"]}" unless [ "gha", "registry" ].include?(config["cache"]["type"])
+    end
+  end
+end

lib/kamal/configuration/validator/configuration.rb

@@ -0,0 +1,6 @@
+class Kamal::Configuration::Validator::Configuration < Kamal::Configuration::Validator
+  private
+    def allow_extensions?
+      true
+    end
+end

lib/kamal/configuration/validator/env.rb

@@ -0,0 +1,54 @@
+class Kamal::Configuration::Validator::Env < Kamal::Configuration::Validator
+  SPECIAL_KEYS = [ "clear", "secret", "tags" ]
+
+  def validate!
+    if known_keys.any?
+      validate_complex_env!
+    else
+      validate_simple_env!
+    end
+  end
+
+  private
+    def validate_simple_env!
+      validate_hash_of!(config, String)
+    end
+
+    def validate_complex_env!
+      unknown_keys_error unknown_keys if unknown_keys.any?
+
+      with_context("clear") { validate_hash_of!(config["clear"], String) } if config.key?("clear")
+      with_context("secret") { validate_array_of!(config["secret"], String) } if config.key?("secret")
+      validate_tags! if config.key?("tags")
+    end
+
+    def known_keys
+      @known_keys ||= config.keys & SPECIAL_KEYS
+    end
+
+    def unknown_keys
+      @unknown_keys ||= config.keys - SPECIAL_KEYS
+    end
+
+    def validate_tags!
+      if context == "env"
+        with_context("tags") do
+          validate_type! config["tags"], Hash
+
+          config["tags"].each do |tag, value|
+            with_context(tag) do
+              validate_type! value, Hash
+
+              Kamal::Configuration::Validator::Env.new(
+                value,
+                example: example["tags"].values[1],
+                context: context
+              ).validate!
+            end
+          end
+        end
+      else
+        error "tags are only allowed in the root env"
+      end
+    end
+end

lib/kamal/configuration/validator/registry.rb

@@ -0,0 +1,25 @@
+class Kamal::Configuration::Validator::Registry < Kamal::Configuration::Validator
+  STRING_OR_ONE_ITEM_ARRAY_KEYS = [ "username", "password" ]
+
+  def validate!
+    validate_against_example! \
+      config.except(*STRING_OR_ONE_ITEM_ARRAY_KEYS),
+      example.except(*STRING_OR_ONE_ITEM_ARRAY_KEYS)
+
+    validate_string_or_one_item_array! "username"
+    validate_string_or_one_item_array! "password"
+  end
+
+  private
+    def validate_string_or_one_item_array!(key)
+      with_context(key) do
+        value = config[key]
+
+        error "is required" unless value.present?
+
+        unless value.is_a?(String) || (value.is_a?(Array) && value.size == 1 && value.first.is_a?(String))
+          error "should be a string or an array with one string (for secret lookup)"
+        end
+      end
+    end
+end

lib/kamal/configuration/validator/role.rb

@@ -0,0 +1,11 @@
+class Kamal::Configuration::Validator::Role < Kamal::Configuration::Validator
+  def validate!
+    validate_type! config, Array, Hash
+
+    if config.is_a?(Array)
+      validate_servers! "servers", config
+    else
+      super
+    end
+  end
+end

lib/kamal/configuration/validator/servers.rb

@@ -0,0 +1,7 @@
+class Kamal::Configuration::Validator::Servers < Kamal::Configuration::Validator
+  def validate!
+    validate_type! config, Array, Hash
+
+    validate_servers! config if config.is_a?(Array)
+  end
+end

lib/kamal/env_file.rb

@@ -3,21 +3,11 @@ class Kamal::EnvFile
   def initialize(env)
     @env = env
   end
-  
+
   def to_s
     env_file = StringIO.new.tap do |contents|
-      if (secrets = @env["secret"]).present?
+      @env.each do |key, value|
-        @env.fetch("secret", @env)&.each do |key|
+        contents << docker_env_file_line(key, value)
-          contents << docker_env_file_line(key, ENV.fetch(key))
-        end
-
-        @env["clear"]&.each do |key, value|
-          contents << docker_env_file_line(key, value)
-        end
-      else
-        @env.fetch("clear", @env)&.each do |key, value|
-          contents << docker_env_file_line(key, value)
-        end
       end
     end.string
 
@@ -26,14 +16,21 @@ class Kamal::EnvFile
   end
 
   alias to_str to_s
-  
+
   private
     def docker_env_file_line(key, value)
-      "#{key.to_s}=#{escape_docker_env_file_value(value)}\n"
+      "#{key}=#{escape_docker_env_file_value(value)}\n"
     end
 
     # Escape a value to make it safe to dump in a docker file.
     def escape_docker_env_file_value(value)
+      # keep non-ascii(UTF-8) characters as it is
+      value.to_s.scan(/[\x00-\x7F]+|[^\x00-\x7F]+/).map do |part|
+        part.ascii_only? ? escape_docker_env_file_ascii_value(part) : part
+      end.join
+    end
+
+    def escape_docker_env_file_ascii_value(value)
       # Doublequotes are treated literally in docker env files
       # so remove leading and trailing ones and unescape any others
       value.to_s.dump[1..-2].gsub(/\\"/, "\"")

lib/kamal/git.rb

@@ -9,6 +9,10 @@ module Kamal::Git
     `git config user.name`.strip
   end
 
+  def email
+    `git config user.email`.strip
+  end
+
   def revision
     `git rev-parse HEAD`.strip
   end
@@ -16,4 +20,8 @@ module Kamal::Git
   def uncommitted_changes
     `git status --porcelain`.strip
   end
+
+  def root
+    `git rev-parse --show-toplevel`.strip
+  end
 end

lib/kamal/sshkit_with_ext.rb

@@ -103,3 +103,39 @@ class SSHKit::Backend::Netssh
 
   prepend LimitConcurrentStartsInstance
 end
+
+class SSHKit::Runner::Parallel
+  # SSHKit joins the threads in sequence and fails on the first error it encounters, which means that we wait threads
+  # before the first failure to complete but not for ones after.
+  #
+  # We'll patch it to wait for them all to complete, and to record all the threads that errored so we can see when a
+  # problem occurs on multiple hosts.
+  module CompleteAll
+    def execute
+      threads = hosts.map do |host|
+        Thread.new(host) do |h|
+          backend(h, &block).run
+        rescue ::StandardError => e
+          e2 = SSHKit::Runner::ExecuteError.new e
+          raise e2, "Exception while executing #{host.user ? "as #{host.user}@" : "on host "}#{host}: #{e.message}"
+        end
+      end
+
+      exceptions = []
+      threads.each do |t|
+        begin
+          t.join
+        rescue SSHKit::Runner::ExecuteError => e
+          exceptions << e
+        end
+      end
+      if exceptions.one?
+        raise exceptions.first
+      elsif exceptions.many?
+        raise exceptions.first, [ "Exceptions on #{exceptions.count} hosts:", exceptions.map(&:message) ].join("\n")
+      end
+    end
+  end
+
+  prepend CompleteAll
+end

lib/kamal/tags.rb

@@ -10,10 +10,11 @@ class Kamal::Tags
 
     def default_tags(config)
       { recorded_at: Time.now.utc.iso8601,
-        performer: `whoami`.chomp,
+        performer: Kamal::Git.email.presence || `whoami`.chomp,
         destination: config.destination,
         version: config.version,
-        service_version: service_version(config) }
+        service_version: service_version(config),
+        service: config.service }
     end
 
     def service_version(config)

lib/kamal/utils.rb

@@ -9,7 +9,7 @@ module Kamal::Utils
       if value.present?
         attr = "#{key}=#{escape_shell_value(value)}"
         attr = self.sensitive(attr, redaction: "#{key}=[REDACTED]") if sensitive
-        [ argument, attr]
+        [ argument, attr ]
       else
         [ argument, key ]
       end
@@ -29,7 +29,7 @@ module Kamal::Utils
 
   # Flattens a one-to-many structure into an array of two-element arrays each containing a key-value pair
   def flatten_args(args)
-    args.flat_map { |key, value| value.try(:map) { |entry| [key, entry] } || [ [ key, value ] ] }
+    args.flat_map { |key, value| value.try(:map) { |entry| [ key, entry ] } || [ [ key, value ] ] }
   end
 
   # Marks sensitive values for redaction in logs and human-visible output.
@@ -66,12 +66,15 @@ module Kamal::Utils
     Array(filters).select do |filter|
       matches += Array(items).select do |item|
         # Only allow * for a wildcard
-        pattern = Regexp.escape(filter).gsub('\*', '.*')
         # items are roles or hosts
-        (item.respond_to?(:name) ? item.name : item).match(/^#{pattern}$/)
+        File.fnmatch(filter, item.to_s, File::FNM_EXTGLOB)
       end
     end
 
-    matches
+    matches.uniq
+  end
+
+  def stable_sort!(elements, &block)
+    elements.sort_by!.with_index { |element, index| [ block.call(element), index ] }
   end
 end

lib/kamal/version.rb

@@ -1,3 +1,3 @@
 module Kamal
-  VERSION = "1.4.0"
+  VERSION = "1.9.0"
 end

test/cli/accessory_test.rb

@@ -7,7 +7,7 @@ class CliAccessoryTest < CliTestCase
 
     run_command("boot", "mysql").tap do |output|
       assert_match /docker login.*on 1.1.1.3/, output
-      assert_match "docker run --name app-mysql --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 3306:3306 --env-file .kamal/env/accessories/app-mysql.env --volume $PWD/app-mysql/etc/mysql/my.cnf:/etc/mysql/my.cnf --volume $PWD/app-mysql/data:/var/lib/mysql --label service=\"app-mysql\" mysql:5.7 on 1.1.1.3", output
+      assert_match "docker run --name app-mysql --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 3306:3306 --env-file .kamal/env/accessories/app-mysql.env --env MYSQL_ROOT_HOST=\"%\" --volume $PWD/app-mysql/etc/mysql/my.cnf:/etc/mysql/my.cnf --volume $PWD/app-mysql/data:/var/lib/mysql --label service=\"app-mysql\" mysql:5.7 on 1.1.1.3", output
     end
   end
 
@@ -21,7 +21,7 @@ class CliAccessoryTest < CliTestCase
       assert_match /docker login.*on 1.1.1.3/, output
       assert_match /docker login.*on 1.1.1.1/, output
       assert_match /docker login.*on 1.1.1.2/, output
-      assert_match "docker run --name app-mysql --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 3306:3306 --env-file .kamal/env/accessories/app-mysql.env --volume $PWD/app-mysql/etc/mysql/my.cnf:/etc/mysql/my.cnf --volume $PWD/app-mysql/data:/var/lib/mysql --label service=\"app-mysql\" mysql:5.7 on 1.1.1.3", output
+      assert_match "docker run --name app-mysql --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 3306:3306 --env-file .kamal/env/accessories/app-mysql.env --env MYSQL_ROOT_HOST=\"%\" --volume $PWD/app-mysql/etc/mysql/my.cnf:/etc/mysql/my.cnf --volume $PWD/app-mysql/data:/var/lib/mysql --label service=\"app-mysql\" mysql:5.7 on 1.1.1.3", output
       assert_match "docker run --name app-redis --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 6379:6379 --env-file .kamal/env/accessories/app-redis.env --volume $PWD/app-redis/data:/data --label service=\"app-redis\" redis:latest on 1.1.1.1", output
       assert_match "docker run --name app-redis --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 6379:6379 --env-file .kamal/env/accessories/app-redis.env --volume $PWD/app-redis/data:/data --label service=\"app-redis\" redis:latest on 1.1.1.2", output
     end
@@ -76,11 +76,20 @@ class CliAccessoryTest < CliTestCase
   end
 
   test "details" do
-    assert_match "docker ps --filter label=service=app-mysql", run_command("details", "mysql")
+    run_command("details", "mysql").tap do |output|
+      assert_match "docker ps --filter label=service=app-mysql", output
+      assert_match "Accessory mysql Host: 1.1.1.3", output
+    end
+  end
+
+  test "details with non-existent accessory" do
+    assert_equal "No accessory by the name of 'hello' (options: mysql and redis)", stderred { run_command("details", "hello") }
   end
 
   test "details with all" do
     run_command("details", "all").tap do |output|
+      assert_match "Accessory mysql Host: 1.1.1.3", output
+      assert_match "Accessory redis Host: 1.1.1.2", output
       assert_match "docker ps --filter label=service=app-mysql", output
       assert_match "docker ps --filter label=service=app-redis", output
     end
@@ -107,6 +116,20 @@ class CliAccessoryTest < CliTestCase
     assert_match "docker logs app-mysql  --tail 100 --timestamps 2>&1", run_command("logs", "mysql")
   end
 
+  test "logs with grep" do
+    SSHKit::Backend::Abstract.any_instance.stubs(:exec)
+      .with("ssh -t root@1.1.1.3 'docker logs app-mysql --timestamps 2>&1 | grep \'hey\''")
+
+    assert_match "docker logs app-mysql --timestamps 2>&1 | grep 'hey'", run_command("logs", "mysql", "--grep", "hey")
+  end
+
+  test "logs with grep and grep options" do
+    SSHKit::Backend::Abstract.any_instance.stubs(:exec)
+      .with("ssh -t root@1.1.1.3 'docker logs app-mysql --timestamps 2>&1 | grep \'hey\' -C 2'")
+
+    assert_match "docker logs app-mysql --timestamps 2>&1 | grep 'hey' -C 2", run_command("logs", "mysql", "--grep", "hey", "--grep-options", "-C 2")
+  end
+
   test "logs with follow" do
     SSHKit::Backend::Abstract.any_instance.stubs(:exec)
       .with("ssh -t root@1.1.1.3 -p 22 'docker logs app-mysql --timestamps --tail 10 --follow 2>&1'")
@@ -114,6 +137,20 @@ class CliAccessoryTest < CliTestCase
     assert_match "docker logs app-mysql --timestamps --tail 10 --follow 2>&1", run_command("logs", "mysql", "--follow")
   end
 
+  test "logs with follow and grep" do
+    SSHKit::Backend::Abstract.any_instance.stubs(:exec)
+      .with("ssh -t root@1.1.1.3 -p 22 'docker logs app-mysql --timestamps --tail 10 --follow 2>&1 | grep \"hey\"'")
+
+    assert_match "docker logs app-mysql --timestamps --tail 10 --follow 2>&1 | grep \"hey\"", run_command("logs", "mysql", "--follow", "--grep", "hey")
+  end
+
+  test "logs with follow, grep, and grep options" do
+    SSHKit::Backend::Abstract.any_instance.stubs(:exec)
+      .with("ssh -t root@1.1.1.3 -p 22 'docker logs app-mysql --timestamps --tail 10 --follow 2>&1 | grep \"hey\" -C 2'")
+
+    assert_match "docker logs app-mysql --timestamps --tail 10 --follow 2>&1 | grep \"hey\" -C 2", run_command("logs", "mysql", "--follow", "--grep", "hey", "--grep-options", "-C 2")
+  end
+
   test "remove with confirmation" do
     Kamal::Cli::Accessory.any_instance.expects(:stop).with("mysql")
     Kamal::Cli::Accessory.any_instance.expects(:remove_container).with("mysql")
@@ -154,9 +191,9 @@ class CliAccessoryTest < CliTestCase
 
     run_command("boot", "redis", "--hosts", "1.1.1.1").tap do |output|
       assert_match /docker login.*on 1.1.1.1/, output
-      refute_match /docker login.*on 1.1.1.2/, output
+      assert_no_match /docker login.*on 1.1.1.2/, output
       assert_match "docker run --name app-redis --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 6379:6379 --env-file .kamal/env/accessories/app-redis.env --volume $PWD/app-redis/data:/data --label service=\"app-redis\" redis:latest on 1.1.1.1", output
-      refute_match "docker run --name app-redis --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 6379:6379 --env-file .kamal/env/accessories/app-redis.env --volume $PWD/app-redis/data:/data --label service=\"app-redis\" redis:latest on 1.1.1.2", output
+      assert_no_match "docker run --name app-redis --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 6379:6379 --env-file .kamal/env/accessories/app-redis.env --volume $PWD/app-redis/data:/data --label service=\"app-redis\" redis:latest on 1.1.1.2", output
     end
   end
 
@@ -166,14 +203,32 @@ class CliAccessoryTest < CliTestCase
 
     run_command("boot", "redis", "--hosts", "1.1.1.1,1.1.1.3").tap do |output|
       assert_match /docker login.*on 1.1.1.1/, output
-      refute_match /docker login.*on 1.1.1.3/, output
+      assert_no_match /docker login.*on 1.1.1.3/, output
       assert_match "docker run --name app-redis --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 6379:6379 --env-file .kamal/env/accessories/app-redis.env --volume $PWD/app-redis/data:/data --label service=\"app-redis\" redis:latest on 1.1.1.1", output
-      refute_match "docker run --name app-redis --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 6379:6379 --env-file .kamal/env/accessories/app-redis.env --volume $PWD/app-redis/data:/data --label service=\"app-redis\" redis:latest on 1.1.1.3", output
+      assert_no_match "docker run --name app-redis --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 6379:6379 --env-file .kamal/env/accessories/app-redis.env --volume $PWD/app-redis/data:/data --label service=\"app-redis\" redis:latest on 1.1.1.3", output
+    end
+  end
+
+  test "downgrade" do
+    run_command("downgrade", "-y", "all").tap do |output|
+      assert_match "Downgrading all accessories on 1.1.1.3,1.1.1.1,1.1.1.2...", output
+      assert_match "docker container stop app-mysql on 1.1.1.3", output
+      assert_match "docker run --name app-mysql --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 3306:3306 --env-file .kamal/env/accessories/app-mysql.env --env MYSQL_ROOT_HOST=\"%\" --volume $PWD/app-mysql/etc/mysql/my.cnf:/etc/mysql/my.cnf --volume $PWD/app-mysql/data:/var/lib/mysql --label service=\"app-mysql\" mysql:5.7 on 1.1.1.3", output
+      assert_match "Downgraded all accessories on 1.1.1.3,1.1.1.1,1.1.1.2", output
+    end
+  end
+
+  test "downgrade rolling" do
+    run_command("downgrade", "--rolling", "-y", "all").tap do |output|
+      assert_match "Downgrading all accessories on 1.1.1.3...", output
+      assert_match "docker container stop app-mysql on 1.1.1.3", output
+      assert_match "docker run --name app-mysql --detach --restart unless-stopped --log-opt max-size=\"10m\" --publish 3306:3306 --env-file .kamal/env/accessories/app-mysql.env --env MYSQL_ROOT_HOST=\"%\" --volume $PWD/app-mysql/etc/mysql/my.cnf:/etc/mysql/my.cnf --volume $PWD/app-mysql/data:/var/lib/mysql --label service=\"app-mysql\" mysql:5.7 on 1.1.1.3", output
+      assert_match "Downgraded all accessories on 1.1.1.3", output
     end
   end
 
   private
     def run_command(*command)
-      stdouted { Kamal::Cli::Accessory.start([*command, "-c", "test/fixtures/deploy_with_accessories.yml"]) }
+      stdouted { Kamal::Cli::Accessory.start([ *command, "-c", "test/fixtures/deploy_with_accessories.yml" ]) }
     end
 end

test/cli/app_test.rb

@@ -23,11 +23,11 @@ class CliAppTest < CliTestCase
       .returns("running") # health check
 
     SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
-      .with(:docker, :ps, "--filter", "label=service=app", "--filter", "label=role=web", "--filter", "status=running", "--filter", "status=restarting", "--latest", "--format", "\"{{.Names}}\"", "|", "while read line; do echo ${line#app-web-}; done", raise_on_non_zero_exit: false)
+      .with(:sh, "-c", "'docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting'", "|", :head, "-1", "|", "while read line; do echo ${line#app-web-}; done", raise_on_non_zero_exit: false)
       .returns("123") # old version
 
     SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
-      .with(:docker, :inspect, "-f '{{ range .Mounts }}{{printf \"%s %s\\n\" .Source .Destination}}{{ end }}'", "app-web-123", "|", :awk, "'$2 == \"/tmp/kamal-cord\" {print $1}'", :raise_on_non_zero_exit => false)
+      .with(:docker, :inspect, "-f '{{ range .Mounts }}{{printf \"%s %s\\n\" .Source .Destination}}{{ end }}'", "app-web-123", "|", :awk, "'$2 == \"/tmp/kamal-cord\" {print $1}'", raise_on_non_zero_exit: false)
       .returns("cordfile") # old version
 
     SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
@@ -45,7 +45,7 @@ class CliAppTest < CliTestCase
   end
 
   test "boot uses group strategy when specified" do
-    Kamal::Cli::App.any_instance.stubs(:on).with("1.1.1.1").twice # acquire & release lock
+    Kamal::Cli::App.any_instance.stubs(:on).with("1.1.1.1").times(3) # ensure locks dir, acquire & release lock
     Kamal::Cli::App.any_instance.stubs(:on).with([ "1.1.1.1" ]) # tag container
 
     # Strategy is used when booting the containers
@@ -54,14 +54,14 @@ class CliAppTest < CliTestCase
     run_command("boot", config: :with_boot_strategy)
   end
 
-  test "boot errors leave lock in place" do
+  test "boot errors don't leave lock in place" do
     Kamal::Cli::App.any_instance.expects(:using_version).raises(RuntimeError)
 
-    assert !KAMAL.holding_lock?
+    assert_not KAMAL.holding_lock?
     assert_raises(RuntimeError) do
       stderred { run_command("boot") }
     end
-    assert KAMAL.holding_lock?
+    assert_not KAMAL.holding_lock?
   end
 
   test "boot with assets" do
@@ -75,11 +75,11 @@ class CliAppTest < CliTestCase
       .returns("running") # health check
 
     SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
-      .with(:docker, :ps, "--filter", "label=service=app", "--filter", "label=role=web", "--filter", "status=running", "--filter", "status=restarting", "--latest", "--format", "\"{{.Names}}\"", "|", "while read line; do echo ${line#app-web-}; done", raise_on_non_zero_exit: false)
+      .with(:sh, "-c", "'docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting'", "|", :head, "-1", "|", "while read line; do echo ${line#app-web-}; done", raise_on_non_zero_exit: false)
       .returns("123").twice # old version
 
     SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
-      .with(:docker, :inspect, "-f '{{ range .Mounts }}{{printf \"%s %s\\n\" .Source .Destination}}{{ end }}'", "app-web-123", "|", :awk, "'$2 == \"/tmp/kamal-cord\" {print $1}'", :raise_on_non_zero_exit => false)
+      .with(:docker, :inspect, "-f '{{ range .Mounts }}{{printf \"%s %s\\n\" .Source .Destination}}{{ end }}'", "app-web-123", "|", :awk, "'$2 == \"/tmp/kamal-cord\" {print $1}'", raise_on_non_zero_exit: false)
       .returns("") # old version
 
     run_command("boot", config: :with_assets).tap do |output|
@@ -92,6 +92,82 @@ class CliAppTest < CliTestCase
     end
   end
 
+  test "boot with host tags" do
+    Object.any_instance.stubs(:sleep)
+
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+    .with(:docker, :container, :ls, "--all", "--filter", "name=^app-web-latest$", "--quiet", raise_on_non_zero_exit: false)
+    .returns("12345678") # running version
+
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:docker, :container, :ls, "--all", "--filter", "name=^app-web-latest$", "--quiet", "|", :xargs, :docker, :inspect, "--format", "'{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'")
+      .returns("running") # health check
+
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:sh, "-c", "'docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting'", "|", :head, "-1", "|", "while read line; do echo ${line#app-web-}; done", raise_on_non_zero_exit: false)
+      .returns("123") # old version
+
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:docker, :inspect, "-f '{{ range .Mounts }}{{printf \"%s %s\\n\" .Source .Destination}}{{ end }}'", "app-web-123", "|", :awk, "'$2 == \"/tmp/kamal-cord\" {print $1}'", raise_on_non_zero_exit: false)
+      .returns("") # old version
+
+    run_command("boot", config: :with_env_tags).tap do |output|
+      assert_match "docker tag dhh/app:latest dhh/app:latest", output
+      assert_match %r{docker run --detach --restart unless-stopped --name app-web-latest --hostname 1.1.1.1-[0-9a-f]{12} -e KAMAL_CONTAINER_NAME="app-web-latest" -e KAMAL_VERSION="latest" --env-file .kamal/env/roles/app-web.env --env TEST="root" --env EXPERIMENT="disabled" --env SITE="site1"}, output
+      assert_match "docker container ls --all --filter name=^app-web-123$ --quiet | xargs docker stop", output
+    end
+  end
+
+  test "boot with web barrier opened" do
+    Object.any_instance.stubs(:sleep)
+
+    SSHKit::Backend::Abstract.any_instance.stubs(:capture_with_info).returns("123") # old version
+
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:docker, :container, :ls, "--all", "--filter", "name=^app-web-latest$", "--quiet", "|", :xargs, :docker, :inspect, "--format", "'{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'")
+      .returns("running").at_least_once # web health check passing
+
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:docker, :container, :ls, "--all", "--filter", "name=^app-web-123$", "--quiet", "|", :xargs, :docker, :inspect, "--format", "'{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'")
+      .returns("unhealthy").at_least_once # web health check failing
+
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:docker, :container, :ls, "--all", "--filter", "name=^app-workers-latest$", "--quiet", "|", :xargs, :docker, :inspect, "--format", "'{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'")
+      .returns("running").at_least_once # workers health check
+
+    run_command("boot", config: :with_roles, host: nil).tap do |output|
+      assert_match "Waiting for the first healthy web container before booting workers on 1.1.1.3...", output
+      assert_match "Waiting for the first healthy web container before booting workers on 1.1.1.4...", output
+      assert_match "First web container is healthy, booting workers on 1.1.1.3", output
+      assert_match "First web container is healthy, booting workers on 1.1.1.4", output
+  end
+  end
+
+  test "boot with web barrier closed" do
+    Thread.report_on_exception = false
+
+    Object.any_instance.stubs(:sleep)
+
+    SSHKit::Backend::Abstract.any_instance.stubs(:capture_with_info).returns("123") # old version
+
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:docker, :container, :ls, "--all", "--filter", "name=^app-web-latest$", "--quiet", "|", :xargs, :docker, :inspect, "--format", "'{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'")
+      .returns("unhealthy").at_least_once # web health check failing
+
+    stderred do
+      run_command("boot", config: :with_roles, host: nil, allow_execute_error: true).tap do |output|
+        assert_match "Waiting for the first healthy web container before booting workers on 1.1.1.3...", output
+        assert_match "Waiting for the first healthy web container before booting workers on 1.1.1.4...", output
+        assert_match "First web container is unhealthy, not booting workers on 1.1.1.3", output
+        assert_match "First web container is unhealthy, not booting workers on 1.1.1.4", output
+        assert_match "Running docker container ls --all --filter name=^app-web-latest$ --quiet | xargs docker stop on 1.1.1.1", output
+        assert_match "Running docker container ls --all --filter name=^app-web-latest$ --quiet | xargs docker stop on 1.1.1.2", output
+      end
+    end
+  ensure
+    Thread.report_on_exception = true
+  end
+
   test "start" do
     run_command("start").tap do |output|
       assert_match "docker start app-web-999", output
@@ -100,14 +176,18 @@ class CliAppTest < CliTestCase
 
   test "stop" do
     run_command("stop").tap do |output|
-      assert_match "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker stop", output
+      assert_match "sh -c 'docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting' | head -1 | xargs docker stop", output
     end
   end
 
   test "stale_containers" do
     SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
       .with(:docker, :ps, "--filter", "label=service=app", "--filter", "label=role=web", "--format", "\"{{.Names}}\"", "|", "while read line; do echo ${line#app-web-}; done", raise_on_non_zero_exit: false)
-      .returns("12345678\n87654321")
+      .returns("12345678\n87654321\n")
+
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:sh, "-c", "'docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting'", "|", :head, "-1", "|", "while read line; do echo ${line#app-web-}; done", raise_on_non_zero_exit: false)
+      .returns("12345678\n")
 
     run_command("stale_containers").tap do |output|
       assert_match /Detected stale container for role web with version 87654321/, output
@@ -117,7 +197,11 @@ class CliAppTest < CliTestCase
   test "stop stale_containers" do
     SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
       .with(:docker, :ps, "--filter", "label=service=app", "--filter", "label=role=web", "--format", "\"{{.Names}}\"", "|", "while read line; do echo ${line#app-web-}; done", raise_on_non_zero_exit: false)
-      .returns("12345678\n87654321")
+      .returns("12345678\n87654321\n")
+
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:sh, "-c", "'docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting'", "|", :head, "-1", "|", "while read line; do echo ${line#app-web-}; done", raise_on_non_zero_exit: false)
+      .returns("12345678\n")
 
     run_command("stale_containers", "--stop").tap do |output|
       assert_match /Stopping stale container for role web with version 87654321/, output
@@ -133,7 +217,7 @@ class CliAppTest < CliTestCase
 
   test "remove" do
     run_command("remove").tap do |output|
-      assert_match /#{Regexp.escape("docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker stop")}/, output
+      assert_match /#{Regexp.escape("sh -c 'docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting' | head -1 | xargs docker stop")}/, output
       assert_match /#{Regexp.escape("docker container prune --force --filter label=service=app")}/, output
       assert_match /#{Regexp.escape("docker image prune --all --force --filter label=service=app")}/, output
     end
@@ -165,7 +249,7 @@ class CliAppTest < CliTestCase
 
   test "exec with reuse" do
     run_command("exec", "--reuse", "ruby -v").tap do |output|
-      assert_match "docker ps --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest --format \"{{.Names}}\" | while read line; do echo ${line#app-web-}; done", output # Get current version
+      assert_match "sh -c 'docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting' | head -1 | while read line; do echo ${line#app-web-}; done", output # Get current version
       assert_match "docker exec app-web-999 ruby -v", output
     end
   end
@@ -184,7 +268,7 @@ class CliAppTest < CliTestCase
       .with("ssh -t root@1.1.1.1 -p 22 'docker exec -it app-web-999 ruby -v'")
     run_command("exec", "-i", "--reuse", "ruby -v").tap do |output|
       assert_match "Get current version of running container...", output
-      assert_match "Running docker ps --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest --format \"{{.Names}}\" | while read line; do echo ${line#app-web-}; done on 1.1.1.1", output
+      assert_match "Running /usr/bin/env sh -c 'docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting' | head -1 | while read line; do echo ${line#app-web-}; done on 1.1.1.1", output
       assert_match "Launching interactive command with version 999 via SSH from existing container on 1.1.1.1...", output
     end
   end
@@ -203,34 +287,76 @@ class CliAppTest < CliTestCase
 
   test "logs" do
     SSHKit::Backend::Abstract.any_instance.stubs(:exec)
-      .with("ssh -t root@1.1.1.1 'docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest| xargs docker logs --timestamps --tail 10 2>&1'")
+      .with("ssh -t root@1.1.1.1 'sh -c 'docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting' | head -1| xargs docker logs --timestamps --tail 10 2>&1'")
 
-    assert_match "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs --tail 100 2>&1", run_command("logs")
+    assert_match "sh -c 'docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting' | head -1 | xargs docker logs --tail 100 2>&1", run_command("logs")
+
+    assert_match "sh -c 'docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting' | head -1 | xargs docker logs 2>&1 | grep 'hey'", run_command("logs", "--grep", "hey")
+
+    assert_match "sh -c 'docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting' | head -1 | xargs docker logs 2>&1 | grep 'hey' -C 2", run_command("logs", "--grep", "hey", "--grep-options", "-C 2")
   end
 
   test "logs with follow" do
     SSHKit::Backend::Abstract.any_instance.stubs(:exec)
-      .with("ssh -t root@1.1.1.1 -p 22 'docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs --timestamps --tail 10 --follow 2>&1'")
+      .with("ssh -t root@1.1.1.1 -p 22 'sh -c '\\''docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''\\'\\'''\\''{{.ID}}'\\''\\'\\'''\\'') ; docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting'\\'' | head -1 | xargs docker logs --timestamps --tail 10 --follow 2>&1'")
 
-    assert_match "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs --timestamps --tail 10 --follow 2>&1", run_command("logs", "--follow")
+    assert_match "sh -c '\\''docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''\\'\\'''\\''{{.ID}}'\\''\\'\\'''\\'') ; docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting'\\'' | head -1 | xargs docker logs --timestamps --tail 10 --follow 2>&1", run_command("logs", "--follow")
+  end
+
+  test "logs with follow and grep" do
+    SSHKit::Backend::Abstract.any_instance.stubs(:exec)
+      .with("ssh -t root@1.1.1.1 -p 22 'sh -c '\\''docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''\\'\\'''\\''{{.ID}}'\\''\\'\\'''\\'') ; docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting'\\'' | head -1 | xargs docker logs --timestamps --follow 2>&1 | grep \"hey\"'")
+
+    assert_match "sh -c '\\''docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''\\'\\'''\\''{{.ID}}'\\''\\'\\'''\\'') ; docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting'\\'' | head -1 | xargs docker logs --timestamps --follow 2>&1 | grep \"hey\"", run_command("logs", "--follow", "--grep", "hey")
+  end
+
+  test "logs with follow, grep and grep options" do
+    SSHKit::Backend::Abstract.any_instance.stubs(:exec)
+      .with("ssh -t root@1.1.1.1 -p 22 'sh -c '\\''docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''\\'\\'''\\''{{.ID}}'\\''\\'\\'''\\'') ; docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting'\\'' | head -1 | xargs docker logs --timestamps --follow 2>&1 | grep \"hey\" -C 2'")
+
+    assert_match "sh -c '\\''docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''\\'\\'''\\''{{.ID}}'\\''\\'\\'''\\'') ; docker ps --latest --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting'\\'' | head -1 | xargs docker logs --timestamps --follow 2>&1 | grep \"hey\" -C 2", run_command("logs", "--follow", "--grep", "hey", "--grep-options", "-C 2")
   end
 
   test "version" do
     run_command("version").tap do |output|
-      assert_match "docker ps --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest --format \"{{.Names}}\" | while read line; do echo ${line#app-web-}; done", output
+      assert_match "sh -c 'docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting' | head -1 | while read line; do echo ${line#app-web-}; done", output
     end
   end
 
 
   test "version through main" do
-    stdouted { Kamal::Cli::Main.start(["app", "version", "-c", "test/fixtures/deploy_with_accessories.yml", "--hosts", "1.1.1.1"]) }.tap do |output|
+    stdouted { Kamal::Cli::Main.start([ "app", "version", "-c", "test/fixtures/deploy_with_accessories.yml", "--hosts", "1.1.1.1" ]) }.tap do |output|
-      assert_match "docker ps --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest --format \"{{.Names}}\" | while read line; do echo ${line#app-web-}; done", output
+      assert_match "sh -c 'docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --filter ancestor=$(docker image ls --filter reference=dhh/app:latest --format '\\''{{.ID}}'\\'') ; docker ps --latest --format '\\''{{.Names}}'\\'' --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting' | head -1 | while read line; do echo ${line#app-web-}; done", output
+    end
+  end
+
+  test "long hostname" do
+    stub_running
+
+    hostname = "this-hostname-is-really-unacceptably-long-to-be-honest.example.com"
+
+    stdouted { Kamal::Cli::App.start([ "boot", "-c", "test/fixtures/deploy_with_uncommon_hostnames.yml", "--hosts", hostname ]) }.tap do |output|
+      assert_match /docker run --detach --restart unless-stopped --name app-web-latest --hostname this-hostname-is-really-unacceptably-long-to-be-hon-[0-9a-f]{12} /, output
+    end
+  end
+
+  test "hostname is trimmed if will end with a period" do
+    stub_running
+
+    hostname = "this-hostname-with-random-part-is-too-long.example.com"
+
+    stdouted { Kamal::Cli::App.start([ "boot", "-c", "test/fixtures/deploy_with_uncommon_hostnames.yml", "--hosts", hostname ]) }.tap do |output|
+      assert_match /docker run --detach --restart unless-stopped --name app-web-latest --hostname this-hostname-with-random-part-is-too-long.example-[0-9a-f]{12} /, output
     end
   end
 
   private
-    def run_command(*command, config: :with_accessories)
+    def run_command(*command, config: :with_accessories, host: "1.1.1.1", allow_execute_error: false)
-      stdouted { Kamal::Cli::App.start([*command, "-c", "test/fixtures/deploy_#{config}.yml", "--hosts", "1.1.1.1"]) }
+      stdouted do
+        Kamal::Cli::App.start([ *command, "-c", "test/fixtures/deploy_#{config}.yml", *([ "--hosts", host ] if host) ])
+      rescue SSHKit::Runner::ExecuteError => e
+        raise e unless allow_execute_error
+      end
     end
 
     def stub_running