web47core 3.2.3.40 → 3.2.3.42

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1c3106bd513997d40d9af68ad8640578c820d28031c045019123c33cff38984f
-  data.tar.gz: d8d6cb5a0f0c0ce213abd8ca89948dbc242bdd694521d822991b959f446b8b16
+  metadata.gz: c408e11bf933327bb1c63ef9931033cb2b831ad86b3bfbf73a9e8b5c53eb89dd
+  data.tar.gz: d3cd71bfc1661417a8afc7c0897c4330fe9e32bc5cb8a80dbfee8cca7309de90
 SHA512:
-  metadata.gz: 5540f6e7610972cd51e79a377c111e46cce0bcf354fbe5dea7e0bfc2cb39ad889f77be55a131c04af555982f7989e3a1c60abce03a91a3c505aadab73be631aa
-  data.tar.gz: 89b9d953406a602fae976918bfd0244daf9242f1e6f11476bf97a413b3cab8829dd6d9c7d6392cdb7017613192096be98771a715b32f7a5ce5f4a36c649e4592
+  metadata.gz: edfc1faa44990b256df9722a90da6d97d3857a09dbdbbc985a9d0b3395c907414a4bd39c502f1ea71e229ae008530ea8a41c1cf6d03bed95ba485a2a4584e20c
+  data.tar.gz: '08816099746e7f020601ac51972d90a8a72b680c11bb14968c7acd3ef281defcb48646b172742c216ad40591b4ca7a045b6e66976a1424eda7216cff87f58eae'

data/app/controllers/status_controller.rb

@@ -13,7 +13,8 @@ class StatusController < ActionController::Base
                    redis: redis_status,
                    jobs: delayed_jobs_status,
                    cron_job_servers: cron_job_servers_status,
-                   workers: workers_status }
+                   workers: workers_status,
+                   version: version_status }
 
     respond_to do |format|
       format.html { render :index, locals: { components: ui_decorators(components) }, layout: false }
@@ -105,32 +106,27 @@ class StatusController < ActionController::Base
   end
 
   #
-  # Cron job primary status
+  # Cron job primary status. Up or down only. Host, pid, and check-in time stay on the admin page.
   #
   def cron_job_servers_status
     server = cron_job_server
-    raise 'No primary server' if server.blank?
+    raise 'Down' if server.blank? || server.dead?
 
-    server_info = "#{server.host_name}(#{server.pid}), last check in #{server.last_check_in_at}"
-    raise "Primary Server is DEAD: #{server_info}" if server.dead?
-
-    report_success "Primary Server is alive: #{server_info}",
-                   primary_count: 1,
-                   total_count: Cron::Server.count
+    report_success 'Up', primary_count: 1, total_count: Cron::Server.count
   rescue StandardError => error
     report_error(error, primary_count: 0, total_count: Cron::Server.count)
   end
 
   #
-  # Desired and current worker status for servers
+  # Desired and current worker counts, last written by AutoScaler.
   #
   def workers_status
-    server = cron_job_server
-    raise 'No primary server' if server.blank? || server.dead?
-
-    report_success "Workers #{server.desired_server_count}/ #{server.current_server_count}",
-                   desired_count: server.desired_server_count,
-                   current_count: server.current_server_count
+    config = SystemConfiguration.configuration
+    desired = config.aws_desired_worker_count.to_i
+    current = config.aws_current_worker_count.to_i
+    report_success "Workers #{desired}/ #{current}",
+                   desired_count: desired,
+                   current_count: current
   rescue StandardError => error
     report_error(error, desired_count: 0, current_count: 0)
   end
@@ -138,4 +134,22 @@ class StatusController < ActionController::Base
   def cron_job_server
     @cron_job_server ||= Cron::Server.primary_server
   end
+
+  #
+  # Deployed image identity. GIT_SHA and BUILT_AT are injected at image build time.
+  # Missing values stay successful so an older task does not fail the health check.
+  #
+  def version_status
+    built_at = ENV['BUILT_AT'].presence
+    git_sha = ENV['GIT_SHA'].presence
+    report_success(version_message(built_at, git_sha), built_at: built_at, git_sha: git_sha)
+  end
+
+  def version_message(built_at, git_sha)
+    return 'Version not set' if built_at.blank? && git_sha.blank?
+    return built_at if git_sha.blank?
+    return git_sha if built_at.blank?
+
+    "#{built_at} #{git_sha}"
+  end
 end

data/lib/app/jobs/cron/server.rb

@@ -16,8 +16,6 @@ module Cron
     #
     # Fields
     #
-    field :desired_server_count, type: Integer, default: 0
-    field :current_server_count, type: Integer, default: 0
     field :state, type: String, default: STATE_SECONDARY
     #
     # Validations
@@ -52,7 +50,7 @@ module Cron
 
     def run_cron_jobs
       run_jobs
-      check_auto_scale
+      AutoScaler.check
     end
 
     # @abstract Run all cron tab and emitter jobs
@@ -85,15 +83,6 @@ module Cron
       Cron::Server.where(state: STATE_PRIMARY).first
     end
 
-    #
-    # Warm up a server on the next evaluation
-    #
-    def self.warm_up_server
-      return unless auto_scaling_configured?
-
-      primary_server.auto_scale([primary_server.desired_server_count + 1, 10].min)
-    end
-
     #
     # Become primary, making others secondary
     #
@@ -148,124 +137,6 @@ module Cron
       !alive?
     end
 
-    #
-    # Auto scale environment
-    #
-    def check_auto_scale
-      return unless auto_scaling_configured?
-
-      if delayed_jobs_count.zero?
-        handle_zero_job_count
-      else
-        handle_auto_scale_jobs
-      end
-    end
-
-    def sys_config
-      @sys_config ||= SystemConfiguration.configuration
-    end
-
-    #
-    # Test if autoscaling is configured, return false if there is an error
-    #
-    def auto_scaling_configured?
-      @auto_scaling_configured ||= sys_config.aws_auto_scaling_configured?
-    rescue StandardError
-      false
-    end
-
-    #
-    # Test if autoscaling is configured, return false if there is an error
-    #
-    def self.auto_scaling_configured?
-      SystemConfiguration.aws_auto_scaling_configured?
-    rescue StandardError
-      false
-    end
-
-    # @abstract Returns the AWS ECS service
-    # @return [Aws::ECS::Types::Service]
-    def ecs_service
-      filter = { cluster: sys_config.aws_ecs_cluster_worker_name, services: [sys_config.aws_ecs_service_worker_name] }
-      @ecs_service ||= ecs_client.describe_services(filter).services.first
-    end
-
-    # @abstract Returns a count of the Delayed Jobs in queue that have not failed
-    # @return [Integer]
-    def delayed_jobs_count
-      @delayed_jobs_count ||= Delayed::Backend::Mongoid::Job.where(failed_at: nil).read(mode: :primary).count
-    end
-
-    # @abstract Returns the current value of 'desired capacity' for the ECS Service
-    # @return [Integer]
-    def current_desired_capacity
-      current = ecs_service.desired_count
-      set current_server_count: current
-      current
-    rescue StandardError
-      puts 'error on desirect capacy'
-      0
-    end
-
-    #
-    # Calls the 'auto_scale' method with a 'desired_count' of 0 unless the capacity is already at 0
-    #
-    def handle_zero_job_count
-      return if current_desired_capacity.zero?
-
-      auto_scale
-    end
-
-    #
-    # Calls the 'auto_scale' method with a variable 'desired_count' based on how many jobs are running
-    # We don't need any more workers if the job count is less than 1,000
-    #
-    def handle_auto_scale_jobs
-      return if delayed_jobs_count < sys_config.aws_min_worker_job_count
-
-      case delayed_jobs_count
-      when 1..200
-        auto_scale(1)
-      when 201..500
-        auto_scale(2)
-      when 501..800
-        auto_scale(3)
-      when 801..1_200
-        auto_scale(4)
-      when 1_201..2_000
-        auto_scale(4)
-      when 2_001..3_000
-        auto_scale(5)
-      when 3_001..4_000
-        auto_scale(5)
-      when 4_001..6_000
-        auto_scale(6)
-      when 6_001..8_000
-        auto_scale(7)
-      when 8_001..10_000
-        auto_scale(8)
-      when 10_001..13_000
-        auto_scale(9)
-      else
-        auto_scale(10)
-      end
-    end
-
-    #
-    # Sets the desired and minimum number of EC2 instances to run
-    #
-    def auto_scale(desired_count = 0)
-      set desired_server_count: desired_count
-      # Make sure we don't remove any workers with assigned jobs by accident
-      return if desired_count.positive? && desired_count <= current_desired_capacity
-
-      ecs_client.update_service(cluster: sys_config.aws_ecs_cluster_worker_name,
-                                service: sys_config.aws_ecs_service_worker_name,
-                                desired_count: desired_count)
-    rescue StandardError => error
-      App47Logger.log_error "Unable to set ECS service auto scaler to #{desired_count}", error
-    end
-
     #
     # Look to make sure there is only one primary
     #
@@ -276,28 +147,8 @@ module Cron
       errors.add(:state, 'there can only be one primary') unless primary.blank? || primary.eql?(self)
     end
 
-    #
-    # Returns the count of active servers
-    #
-    def active_count
-      current_server_count
-    end
-
-    #
-    # Returns the count of inactive servers
-    #
-    def inactive_count
-      desired_server_count
-    end
-
     def ensure_last_check_in
       self.last_check_in_at ||= Time.now.utc
     end
-
-    # @abstract Returns the AWS ECS client
-    # @return [Aws::ECS::Client]
-    def ecs_client
-      @ecs_client ||= sys_config.aws_ecs_client
-    end
   end
 end

data/lib/app/models/concerns/aws_configuration.rb

@@ -13,9 +13,13 @@ module AwsConfiguration
       #
       field :aws_region, type: String
       field :aws_bucket_name, type: String
-      field :aws_min_worker_job_count, type: Integer, default: 1
+      # Worker fleet size asked for by AutoScaler. Not edited from the admin form.
+      field :aws_desired_worker_count, type: Integer, default: 0
+      # ECS service desired count from the last DescribeServices. Not the running task count.
+      field :aws_current_worker_count, type: Integer, default: 0
 
-      validates :aws_min_worker_job_count, numericality: { only_integer: true, greater_than: 0 }
+      validates :aws_desired_worker_count, numericality: { only_integer: true, greater_than_or_equal_to: 0 }
+      validates :aws_current_worker_count, numericality: { only_integer: true, greater_than_or_equal_to: 0 }
     end
   end
 

data/lib/app/services/auto_scaler.rb

@@ -0,0 +1,185 @@
+# frozen_string_literal: true
+
+require 'singleton'
+
+#
+# Resize the ECS worker service from the current thread.
+#
+# check runs after the primary cron server has launched due jobs.
+# warm_up raises the fleet to a floor. bump_worker adds one worker.
+# Neither call lowers a fleet that is already larger, except check when the
+# ladder maps the queue to zero workers.
+#
+# A host project replaces the ladder at boot:
+#
+#   Web47core::Config.worker_ladder = [
+#     [0..49, 0],
+#     [50..500, 2],
+#     [501.., 6]
+#   ]
+#
+# Queue depth and the live ECS desired count are read on every call.
+# The singleton only keeps the ECS client.
+#
+class AutoScaler
+  include Singleton
+
+  unless defined? DEFAULT_WORKER_LADDER
+    DEFAULT_WORKER_LADDER = [
+      [0..0, 0],
+      [1..200, 1],
+      [201..500, 2],
+      [501..800, 3],
+      [801..2_000, 4],
+      [2_001..4_000, 5],
+      [4_001..6_000, 6],
+      [6_001..8_000, 7],
+      [8_001..10_000, 8],
+      [10_001..13_000, 9],
+      [13_001.., 10]
+    ].freeze
+  end
+
+  def self.check
+    instance.check
+  end
+
+  def self.warm_up(min: 1)
+    instance.warm_up(min: min)
+  end
+
+  def self.bump_worker
+    instance.bump_worker
+  end
+
+  def self.reset
+    instance.reset
+  end
+
+  def check
+    return unless configured?
+
+    apply_target(desired_for(delayed_jobs_count))
+  end
+
+  #
+  # Raise the ECS desired count to at least min. Never lowers the fleet.
+  #
+  def warm_up(min: 1)
+    return unless configured?
+
+    target = [[min.to_i, 1].max, max_worker_count].min
+    scale_to(target)
+  end
+
+  #
+  # Add one worker to the live ECS desired count, capped at the ladder maximum.
+  #
+  def bump_worker
+    return unless configured?
+
+    current = live_desired_count
+    return if current.nil?
+
+    scale_to([current + 1, max_worker_count].min, current: current)
+  end
+
+  def reset
+    @ecs_client = nil
+  end
+
+  private
+
+  def apply_target(target)
+    record_desired(target)
+    current = live_desired_count
+    return if current.nil?
+    return if target.positive? && target <= current
+    return if target.zero? && current.zero?
+
+    update_service(target)
+  end
+
+  def scale_to(target, current: nil)
+    current = live_desired_count if current.nil?
+    return if current.nil? || target <= current
+
+    record_desired(target)
+    update_service(target)
+  end
+
+  def desired_for(job_count)
+    worker_ladder.each do |range, workers|
+      return workers if range.cover?(job_count)
+    end
+    worker_ladder.last.last
+  end
+
+  def max_worker_count
+    worker_ladder.map(&:last).max
+  end
+
+  def worker_ladder
+    Web47core::Config.worker_ladder.presence || DEFAULT_WORKER_LADDER
+  end
+
+  #
+  # ECS service desired count. nil when the service cannot be read, so a
+  # failed describe does not look like a fleet of zero.
+  #
+  def live_desired_count
+    service = describe_service
+    return nil unless service
+
+    count = service.desired_count
+    configuration.set(aws_current_worker_count: count)
+    count
+  end
+
+  def describe_service
+    client = ecs_client
+    return nil unless client
+
+    filter = {
+      cluster: configuration.aws_ecs_cluster_worker_name,
+      services: [configuration.aws_ecs_service_worker_name]
+    }
+    client.describe_services(filter).services.first
+  rescue StandardError => error
+    App47Logger.log_error 'Unable to read ECS worker service', error
+    nil
+  end
+
+  def record_desired(count)
+    configuration.set(aws_desired_worker_count: count)
+  end
+
+  def update_service(desired_count)
+    ecs_client.update_service(cluster: configuration.aws_ecs_cluster_worker_name,
+                              service: configuration.aws_ecs_service_worker_name,
+                              desired_count: desired_count)
+  rescue StandardError => error
+    App47Logger.log_error "Unable to set ECS service auto scaler to #{desired_count}", error
+  end
+
+  def delayed_jobs_count
+    Delayed::Backend::Mongoid::Job.where(failed_at: nil).read(mode: :primary).count
+  end
+
+  def configured?
+    configuration.aws_auto_scaling_configured?
+  rescue StandardError
+    false
+  end
+
+  def configuration
+    SystemConfiguration.configuration
+  end
+
+  def ecs_client
+    @ecs_client ||= configuration.aws_ecs_client
+  rescue StandardError => error
+    App47Logger.log_error 'Unable to build ECS client', error
+    nil
+  end
+end

data/lib/web47core/config.rb

@@ -6,18 +6,20 @@ module Web47core
   #
   class Config
     include Singleton
-    attr_accessor :email_able_models, :switchboard_able_models, :audit_model
+    attr_accessor :email_able_models, :switchboard_able_models, :audit_model, :worker_ladder
 
     def initialize
       @email_able_models = []
       @switchboard_able_models = []
       @audit_model = :user
+      @worker_ladder = nil
     end
 
     def self.reset
       instance.email_able_models = []
       instance.switchboard_able_models = []
       instance.audit_model = :user
+      instance.worker_ladder = nil
     end
 
     def self.audit_model_log_class

data/lib/web47core/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Web47core
-  VERSION = '3.2.3.40'
+  VERSION = '3.2.3.42'
 end

data/lib/web47core.rb

@@ -61,6 +61,7 @@ require 'app/jobs/cron/record_delayed_job_metrics'
 require 'app/jobs/cron/restart_orphaned_delayed_jobs'
 require 'app/jobs/cron/trim_delayed_job_metrics'
 require 'app/jobs/cron/trim_delayed_job_workers'
+require 'app/services/auto_scaler'
 require 'app/jobs/cron/server'
 require 'app/jobs/cron/switchboard_sync_configuration'
 require 'app/jobs/cron/switchboard_sync_models'

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: web47core
 version: !ruby/object:Gem::Version
-  version: 3.2.3.40
+  version: 3.2.3.42
 platform: ruby
 authors:
 - Chris Schroeder
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2026-09-27 00:00:00.000000000 Z
+date: 2026-10-01 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activemodel
@@ -6606,6 +6606,7 @@ files:
 - lib/app/models/user_action_audit_log.rb
 - lib/app/models/user_audit_log.rb
 - lib/app/models/user_model_audit_log.rb
+- lib/app/services/auto_scaler.rb
 - lib/app/validators/email_format_validator.rb
 - lib/app/validators/http_url_validator.rb
 - lib/merge_coverage.rb