Merge branch 'main' into pr/223

* main:
  Simplify domain language to just "boot" and unscoped config keys
  Retain a fixed number of containers when pruning
  Don't assume rolling back in message
  Check all hosts before rolling back
  Ensure Traefik service name is consistent
  Extend traefik delay by 1 second
  Include traefik access logs
  Check if we are still getting a 404
  Also dump load balancer logs
  Dump traefik logs when app not booted
  Fix missing for apt-get
  Report on container health after failure
  Fix the integration test healthcheck
  Allow percentage-based rolling deployments
  Move `group_limit` & `group_wait` under `boot`
  Limit rolling deployment to boot operation
  Allow performing boot & start operations in groups

lib/mrsk/cli/app.rb

@@ -11,7 +11,7 @@ class Mrsk::Cli::App < Mrsk::Cli::Base
           execute *MRSK.app.tag_current_as_latest
         end
 
-        on(MRSK.hosts) do |host|
+        on(MRSK.hosts, **MRSK.boot_strategy) do |host|
           roles = MRSK.roles_on(host)
 
           roles.each do |role|

lib/mrsk/cli/main.rb

@@ -236,15 +236,24 @@ class Mrsk::Cli::Main < Mrsk::Cli::Base
   subcommand "lock", Mrsk::Cli::Lock
 
   private
-    def container_available?(version, host: MRSK.primary_host)
-      available = nil
-
-      on(host) do
-        first_role = MRSK.roles_on(host).first
-        available = capture_with_info(*MRSK.app(role: first_role).container_id_for_version(version)).present?
+    def container_available?(version)
+      begin
+        on(MRSK.hosts) do
+          MRSK.roles_on(host).each do |role|
+            container_id = capture_with_info(*MRSK.app(role: role).container_id_for_version(version))
+            raise "Container not found" unless container_id.present?
+          end
+        end
+      rescue SSHKit::Runner::ExecuteError => e
+        if e.message =~ /Container not found/
+          say "Error looking for container version #{version}: #{e.message}"
+          return false
+        else
+          raise
+        end
       end
 
-      available
+      true
     end
 
     def deploy_options

lib/mrsk/cli/prune.rb

@@ -7,7 +7,7 @@ class Mrsk::Cli::Prune < Mrsk::Cli::Base
     end
   end
 
-  desc "images", "Prune unused images older than 7 days"
+  desc "images", "Prune dangling images"
   def images
     with_lock do
       on(MRSK.hosts) do
@@ -17,7 +17,7 @@ class Mrsk::Cli::Prune < Mrsk::Cli::Base
     end
   end
 
-  desc "containers", "Prune stopped containers older than 3 days"
+  desc "containers", "Prune all stopped containers, except the last 5"
   def containers
     with_lock do
       on(MRSK.hosts) do

lib/mrsk/commander.rb

@@ -51,6 +51,14 @@ class Mrsk::Commander
     end
   end
 
+  def boot_strategy
+    if config.boot.limit.present?
+      { in: :groups, limit: config.boot.limit, wait: config.boot.wait }
+    else
+      {}
+    end
+  end
+
   def roles_on(host)
     roles.select { |role| role.hosts.include?(host.to_s) }.map(&:name)
   end

lib/mrsk/commands/prune.rb

@@ -2,11 +2,19 @@ require "active_support/duration"
 require "active_support/core_ext/numeric/time"
 
 class Mrsk::Commands::Prune < Mrsk::Commands::Base
-  def images(until_hours: 7.days.in_hours.to_i)
-    docker :image, :prune, "--all", "--force", "--filter", "label=service=#{config.service}", "--filter", "until=#{until_hours}h"
+  def images
+    docker :image, :prune, "--all", "--force", "--filter", "label=service=#{config.service}", "--filter", "dangling=true"
   end
 
-  def containers(until_hours: 3.days.in_hours.to_i)
-    docker :container, :prune, "--force", "--filter", "label=service=#{config.service}", "--filter", "until=#{until_hours}h"
+  def containers(keep_last: 5)
+    pipe \
+      docker(:ps, "-q", "-a", "--filter", "label=service=#{config.service}", *stopped_containers_filters),
+      "tail -n +#{keep_last + 1}",
+      "while read container_id; do docker rm $container_id; done"
   end
+
+  private
+    def stopped_containers_filters
+      [ "created", "exited", "dead" ].flat_map { |status| ["--filter", "status=#{status}"] }
+    end
 end

lib/mrsk/configuration.rb

@@ -87,6 +87,10 @@ class Mrsk::Configuration
     roles.select(&:running_traefik?).flat_map(&:hosts).uniq
   end
 
+  def boot
+    Mrsk::Configuration::Boot.new(config: self)
+  end
+
 
   def repository
     [ raw_config.registry["server"], image ].compact.join("/")

lib/mrsk/configuration/boot.rb

@@ -0,0 +1,20 @@
+class Mrsk::Configuration::Boot
+  def initialize(config:)
+    @options = config.raw_config.boot || {}
+    @host_count = config.all_hosts.count
+  end
+
+  def limit
+    limit = @options["limit"]
+
+    if limit.to_s.end_with?("%")
+      @host_count * limit.to_i / 100
+    else
+      limit
+    end
+  end
+
+  def wait
+    @options["wait"]
+  end
+end

lib/mrsk/configuration/role.rb

@@ -89,6 +89,9 @@ class Mrsk::Configuration::Role
     def traefik_labels
       if running_traefik?
         {
+          # Setting a service property ensures that the generated service name will be consistent between versions
+          "traefik.http.services.#{traefik_service}.loadbalancer.server.scheme" => "http",
+
           "traefik.http.routers.#{traefik_service}.rule" => "PathPrefix(`/`)",
           "traefik.http.middlewares.#{traefik_service}-retry.retry.attempts" => "5",
           "traefik.http.middlewares.#{traefik_service}-retry.retry.initialinterval" => "500ms",

lib/mrsk/utils/healthcheck_poller.rb

@@ -1,5 +1,5 @@
 class Mrsk::Utils::HealthcheckPoller
-  TRAEFIK_HEALTHY_DELAY = 1
+  TRAEFIK_HEALTHY_DELAY = 2
 
   class HealthcheckError < StandardError; end