Merge branch 'port-prometheus-health-check-to-integrations' into 'master'

Port prometheus health check to cluster integration

See merge request gitlab-org/gitlab!72041

app/models/clusters/cluster.rb

@@ -150,9 +150,7 @@ module Clusters
     scope :for_project_namespace, -> (namespace_id) { joins(:projects).where(projects: { namespace_id: namespace_id }) }
     scope :with_name, -> (name) { where(name: name) }
 
-    # with_application_prometheus scope is deprecated, and scheduled for removal
-    # in %14.0. See https://gitlab.com/groups/gitlab-org/-/epics/4280
-    scope :with_application_prometheus, -> { includes(:application_prometheus).joins(:application_prometheus) }
+    scope :with_integration_prometheus, -> { includes(:integration_prometheus).joins(:integration_prometheus) }
     scope :with_project_http_integrations, -> (project_ids) do
       conditions = { projects: :alert_management_http_integrations }
       includes(conditions).joins(conditions).where(projects: { id: project_ids })

app/models/clusters/integrations/prometheus.rb

@@ -14,6 +14,13 @@ module Clusters
       validates :cluster, presence: true
       validates :enabled, inclusion: { in: [true, false] }
 
+      # Periodically checked and kept up to date for Monitor demo projects
+      enum health_status: {
+        unknown: 0,
+        healthy: 1,
+        unhealthy: 2
+      }
+
       attr_encrypted :alert_manager_token,
         mode: :per_attribute_iv,
         key: Settings.attr_encrypted_db_key_base_32,

app/services/clusters/applications/prometheus_health_check_service.rb → app/services/clusters/integrations/prometheus_health_check_service.rb

 # frozen_string_literal: true
 
 module Clusters
-  module Applications
+  module Integrations
     class PrometheusHealthCheckService
       include Gitlab::Utils::StrongMemoize
       include Gitlab::Routing
@@ -14,7 +14,7 @@ module Clusters
       def execute
         raise 'Invalid cluster type. Only project types are allowed.' unless @cluster.project_type?
 
-        return unless prometheus_application.installed?
+        return unless prometheus_integration.enabled
 
         project = @cluster.clusterable
 
@@ -28,32 +28,46 @@ module Clusters
 
         send_notification(project) if became_unhealthy?
 
-        prometheus_application.update_columns(healthy: currently_healthy?) if health_changed?
+        prometheus_integration.update_columns(health_status: current_health_status) if health_changed?
       end
 
       private
 
-      def prometheus_application
-        strong_memoize(:prometheus_application) do
-          @cluster.application_prometheus
+      def prometheus_integration
+        strong_memoize(:prometheus_integration) do
+          @cluster.integration_prometheus
+        end
+      end
+
+      def current_health_status
+        if currently_healthy?
+          :healthy
+        else
+          :unhealthy
         end
       end
 
       def currently_healthy?
         strong_memoize(:currently_healthy) do
-          prometheus_application.prometheus_client.healthy?
+          prometheus_integration.prometheus_client.healthy?
         end
       end
 
       def became_unhealthy?
         strong_memoize(:became_unhealthy) do
-          (was_healthy? || was_healthy?.nil?) && !currently_healthy?
+          (was_healthy? || was_unknown?) && !currently_healthy?
         end
       end
 
       def was_healthy?
         strong_memoize(:was_healthy) do
-          prometheus_application.healthy
+          prometheus_integration.healthy?
+        end
+      end
+
+      def was_unknown?
+        strong_memoize(:was_unknown) do
+          prometheus_integration.unknown?
         end
       end
 

app/workers/all_queues.yml

@@ -1069,8 +1069,8 @@
   :idempotent:
   :tags:
   - :needs_own_queue
-- :name: incident_management:clusters_applications_check_prometheus_health
-  :worker_name: Clusters::Applications::CheckPrometheusHealthWorker
+- :name: incident_management:clusters_integrations_check_prometheus_health
+  :worker_name: Clusters::Integrations::CheckPrometheusHealthWorker
   :feature_category: :incident_management
   :has_external_dependencies: true
   :urgency: :low

app/workers/clusters/applications/check_prometheus_health_worker.rb → app/workers/clusters/integrations/check_prometheus_health_worker.rb

 # frozen_string_literal: true
 
 module Clusters
-  module Applications
+  module Integrations
     class CheckPrometheusHealthWorker
       include ApplicationWorker
 
@@ -22,11 +22,11 @@ module Clusters
       def perform
         demo_project_ids = Gitlab::Monitor::DemoProjects.primary_keys
 
-        clusters = Clusters::Cluster.with_application_prometheus
+        clusters = Clusters::Cluster.with_integration_prometheus
           .with_project_http_integrations(demo_project_ids)
 
         # Move to a seperate worker with scoped context if expanded to do work on customer projects
-        clusters.each { |cluster| Clusters::Applications::PrometheusHealthCheckService.new(cluster).execute }
+        clusters.each { |cluster| Clusters::Integrations::PrometheusHealthCheckService.new(cluster).execute }
       end
     end
   end

db/migrate/20211008193137_add_health_status_column_on_clusters_integration_prometheus.rb

+# frozen_string_literal: true
+
+class AddHealthStatusColumnOnClustersIntegrationPrometheus < Gitlab::Database::Migration[1.0]
+  def change
+    # For now, health checks will only run on monitor demo projects
+    add_column :clusters_integration_prometheus, :health_status, :smallint, limit: 2, default: 0, null: false
+  end
+end

db/schema_migrations/20211008193137

+97efc3bb2039b66dac98135d93baefc780a62571bd80aa39d7458f37ce92905b
\ No newline at end of file

db/structure.sql

@@ -12619,7 +12619,8 @@ CREATE TABLE clusters_integration_prometheus (
     cluster_id bigint NOT NULL,
     enabled boolean DEFAULT false NOT NULL,
     encrypted_alert_manager_token text,
-    encrypted_alert_manager_token_iv text
+    encrypted_alert_manager_token_iv text,
+    health_status smallint DEFAULT 0 NOT NULL
 );
 
 CREATE TABLE clusters_kubernetes_namespaces (

spec/models/clusters/cluster_spec.rb

@@ -178,13 +178,13 @@ RSpec.describe Clusters::Cluster, :use_clean_rails_memory_store_caching do
     end
   end
 
-  describe '.with_application_prometheus' do
-    subject { described_class.with_application_prometheus }
+  describe '.with_integration_prometheus' do
+    subject { described_class.with_integration_prometheus }
 
     let!(:cluster) { create(:cluster) }
 
     context 'cluster has prometheus application' do
-      let!(:application) { create(:clusters_applications_prometheus, :installed, cluster: cluster) }
+      let!(:application) { create(:clusters_integrations_prometheus, cluster: cluster) }
 
       it { is_expected.to include(cluster) }
     end

spec/services/clusters/applications/prometheus_health_check_service_spec.rb → spec/services/clusters/integrations/prometheus_health_check_service_spec.rb

@@ -2,7 +2,7 @@
 
 require 'spec_helper'
 
-RSpec.describe Clusters::Applications::PrometheusHealthCheckService, '#execute' do
+RSpec.describe Clusters::Integrations::PrometheusHealthCheckService, '#execute' do
   let(:service) { described_class.new(cluster) }
 
   subject { service.execute }
@@ -26,10 +26,10 @@ RSpec.describe Clusters::Applications::PrometheusHealthCheckService, '#execute'
   end
 
   RSpec.shared_examples 'correct health stored' do
-    it 'stores the correct health of prometheus app' do
+    it 'stores the correct health of prometheus' do
       subject
 
-      expect(prometheus.healthy).to eq(client_healthy)
+      expect(prometheus.healthy?).to eq(client_healthy)
     end
   end
 
@@ -43,19 +43,19 @@ RSpec.describe Clusters::Applications::PrometheusHealthCheckService, '#execute'
     let_it_be(:project) { create(:project) }
     let_it_be(:integration) { create(:alert_management_http_integration, project: project) }
 
-    let(:applications_prometheus_healthy) { true }
-    let(:prometheus) { create(:clusters_applications_prometheus, status: prometheus_status_value, healthy: applications_prometheus_healthy) }
-    let(:cluster) { create(:cluster, :project, application_prometheus: prometheus, projects: [project]) }
+    let(:previous_health_status) { :healthy }
+    let(:prometheus) { create(:clusters_integrations_prometheus, enabled: prometheus_enabled, health_status: previous_health_status) }
+    let(:cluster) { create(:cluster, :project, integration_prometheus: prometheus, projects: [project]) }
 
-    context 'when prometheus not installed' do
-      let(:prometheus_status_value) { Clusters::Applications::Prometheus.state_machine.states[:installing].value }
+    context 'when prometheus not enabled' do
+      let(:prometheus_enabled) { false }
 
       it { expect(subject).to eq(nil) }
       include_examples 'no alert'
     end
 
-    context 'when prometheus installed' do
-      let(:prometheus_status_value) { Clusters::Applications::Prometheus.state_machine.states[:installed].value }
+    context 'when prometheus enabled' do
+      let(:prometheus_enabled) { true }
 
       before do
         client = instance_double('PrometheusClient', healthy?: client_healthy)
@@ -63,7 +63,7 @@ RSpec.describe Clusters::Applications::PrometheusHealthCheckService, '#execute'
       end
 
       context 'when newly unhealthy' do
-        let(:applications_prometheus_healthy) { true }
+        let(:previous_health_status) { :healthy }
         let(:client_healthy) { false }
 
         include_examples 'sends alert'
@@ -71,7 +71,7 @@ RSpec.describe Clusters::Applications::PrometheusHealthCheckService, '#execute'
       end
 
       context 'when newly healthy' do
-        let(:applications_prometheus_healthy) { false }
+        let(:previous_health_status) { :unhealthy }
         let(:client_healthy) { true }
 
         include_examples 'no alert'
@@ -79,7 +79,7 @@ RSpec.describe Clusters::Applications::PrometheusHealthCheckService, '#execute'
       end
 
       context 'when continuously unhealthy' do
-        let(:applications_prometheus_healthy) { false }
+        let(:previous_health_status) { :unhealthy }
         let(:client_healthy) { false }
 
         include_examples 'no alert'
@@ -87,7 +87,7 @@ RSpec.describe Clusters::Applications::PrometheusHealthCheckService, '#execute'
       end
 
       context 'when continuously healthy' do
-        let(:applications_prometheus_healthy) { true }
+        let(:previous_health_status) { :healthy }
         let(:client_healthy) { true }
 
         include_examples 'no alert'
@@ -95,7 +95,7 @@ RSpec.describe Clusters::Applications::PrometheusHealthCheckService, '#execute'
       end
 
       context 'when first health check and healthy' do
-        let(:applications_prometheus_healthy) { nil }
+        let(:previous_health_status) { :unknown }
         let(:client_healthy) { true }
 
         include_examples 'no alert'
@@ -103,7 +103,7 @@ RSpec.describe Clusters::Applications::PrometheusHealthCheckService, '#execute'
       end
 
       context 'when first health check and not healthy' do
-        let(:applications_prometheus_healthy) { nil }
+        let(:previous_health_status) { :unknown }
         let(:client_healthy) { false }
 
         include_examples 'sends alert'

spec/workers/clusters/applications/check_prometheus_health_worker_spec.rb → spec/workers/clusters/integrations/check_prometheus_health_worker_spec.rb

@@ -2,16 +2,16 @@
 
 require 'spec_helper'
 
-RSpec.describe Clusters::Applications::CheckPrometheusHealthWorker, '#perform' do
+RSpec.describe Clusters::Integrations::CheckPrometheusHealthWorker, '#perform' do
   subject { described_class.new.perform }
 
   it 'triggers health service' do
     cluster = create(:cluster)
     allow(Gitlab::Monitor::DemoProjects).to receive(:primary_keys)
-    allow(Clusters::Cluster).to receive_message_chain(:with_application_prometheus, :with_project_http_integrations).and_return([cluster])
+    allow(Clusters::Cluster).to receive_message_chain(:with_integration_prometheus, :with_project_http_integrations).and_return([cluster])
 
-    service_instance = instance_double(Clusters::Applications::PrometheusHealthCheckService)
-    expect(Clusters::Applications::PrometheusHealthCheckService).to receive(:new).with(cluster).and_return(service_instance)
+    service_instance = instance_double(Clusters::Integrations::PrometheusHealthCheckService)
+    expect(Clusters::Integrations::PrometheusHealthCheckService).to receive(:new).with(cluster).and_return(service_instance)
     expect(service_instance).to receive(:execute)
 
     subject