Merge branch '5029-support-cluster-metrics-ce' into 'master'

Refactoring changes to support cluster metrics in EE

Closes #42820

See merge request gitlab-org/gitlab-ce!17336

11 files changed, 124 insertions, 110 deletions

diff --git a/app/assets/javascripts/monitoring/utils/multiple_time_series.js b/app/assets/javascripts/monitoring/utils/multiple_time_series.js
index 4ce3dad440c..b5b8e3c255d 100644
--- a/app/assets/javascripts/monitoring/utils/multiple_time_series.js
+++ b/app/assets/javascripts/monitoring/utils/multiple_time_series.js
@@ -76,7 +76,7 @@ function queryTimeSeries(query, graphWidth, graphHeight, graphHeightOffset, xDom
       metricTag = seriesCustomizationData.value || timeSeriesMetricLabel;
       [lineColor, areaColor] = pickColor(seriesCustomizationData.color);
     } else {
-      metricTag = timeSeriesMetricLabel || `series ${timeSeriesNumber + 1}`;
+      metricTag = timeSeriesMetricLabel || query.label || `series ${timeSeriesNumber + 1}`;
       [lineColor, areaColor] = pickColor();
     }
 
diff --git a/app/controllers/projects/deployments_controller.rb b/app/controllers/projects/deployments_controller.rb
index 1a418d0f15a..b68cdc39cb8 100644
--- a/app/controllers/projects/deployments_controller.rb
+++ b/app/controllers/projects/deployments_controller.rb
@@ -24,7 +24,7 @@ class Projects::DeploymentsController < Projects::ApplicationController
   end
 
   def additional_metrics
-    return render_404 unless deployment.has_additional_metrics?
+    return render_404 unless deployment.has_metrics?
 
     respond_to do |format|
       format.json do
diff --git a/app/controllers/projects/prometheus/metrics_controller.rb b/app/controllers/projects/prometheus/metrics_controller.rb
index b739d0f0f90..1dd886409a5 100644
--- a/app/controllers/projects/prometheus/metrics_controller.rb
+++ b/app/controllers/projects/prometheus/metrics_controller.rb
@@ -2,11 +2,12 @@ module Projects
   module Prometheus
     class MetricsController < Projects::ApplicationController
       before_action :authorize_admin_project!
+      before_action :require_prometheus_metrics!
 
       def active_common
         respond_to do |format|
           format.json do
-            matched_metrics = prometheus_service.matched_metrics || {}
+            matched_metrics = prometheus_adapter.query(:matched_metrics) || {}
 
             if matched_metrics.any?
               render json: matched_metrics
@@ -19,8 +20,12 @@ module Projects
 
       private
 
-      def prometheus_service
-        @prometheus_service ||= project.find_or_initialize_service('prometheus')
+      def prometheus_adapter
+        @prometheus_adapter ||= ::Prometheus::AdapterService.new(project).prometheus_adapter
+      end
+
+      def require_prometheus_metrics!
+        render_404 unless prometheus_adapter.can_query?
       end
     end
   end
diff --git a/app/models/clusters/applications/prometheus.rb b/app/models/clusters/applications/prometheus.rb
index 89ebd63e605..7b25d8c4089 100644
--- a/app/models/clusters/applications/prometheus.rb
+++ b/app/models/clusters/applications/prometheus.rb
@@ -1,6 +1,8 @@
 module Clusters
   module Applications
     class Prometheus < ActiveRecord::Base
+      include PrometheusAdapter
+
       VERSION = "2.0.0".freeze
 
       self.table_name = 'clusters_applications_prometheus'
@@ -39,7 +41,7 @@ module Clusters
         )
       end
 
-      def proxy_client
+      def prometheus_client
         return unless kube_client
 
         proxy_url = kube_client.proxy_url('service', service_name, service_port, Gitlab::Kubernetes::Helm::NAMESPACE)
diff --git a/app/models/clusters/cluster.rb b/app/models/clusters/cluster.rb
index 1c0046107d7..49eb069016a 100644
--- a/app/models/clusters/cluster.rb
+++ b/app/models/clusters/cluster.rb
@@ -51,9 +51,6 @@ module Clusters
     scope :enabled, -> { where(enabled: true) }
     scope :disabled, -> { where(enabled: false) }
 
-    scope :for_environment, -> (env) { where(environment_scope: ['*', '', env.slug]) }
-    scope :for_all_environments, -> { where(environment_scope: ['*', '']) }
-
     def status_name
       if provider
         provider.status_name
diff --git a/app/models/concerns/prometheus_adapter.rb b/app/models/concerns/prometheus_adapter.rb
new file mode 100644
index 00000000000..18cbbd871a1
--- /dev/null
+++ b/app/models/concerns/prometheus_adapter.rb
@@ -0,0 +1,48 @@
+module PrometheusAdapter
+  extend ActiveSupport::Concern
+
+  included do
+    include ReactiveCaching
+
+    self.reactive_cache_key = ->(adapter) { [adapter.class.model_name.singular, adapter.id] }
+    self.reactive_cache_lease_timeout = 30.seconds
+    self.reactive_cache_refresh_interval = 30.seconds
+    self.reactive_cache_lifetime = 1.minute
+
+    def prometheus_client
+      raise NotImplementedError
+    end
+
+    def prometheus_client_wrapper
+      Gitlab::PrometheusClient.new(prometheus_client)
+    end
+
+    def can_query?
+      prometheus_client.present?
+    end
+
+    def query(query_name, *args)
+      return unless can_query?
+
+      query_class = Gitlab::Prometheus::Queries.const_get("#{query_name.to_s.classify}Query")
+
+      args.map!(&:id)
+
+      with_reactive_cache(query_class.name, *args, &query_class.method(:transform_reactive_result))
+    end
+
+    # Cache metrics for specific environment
+    def calculate_reactive_cache(query_class_name, *args)
+      return unless prometheus_client
+
+      data = Kernel.const_get(query_class_name).new(prometheus_client_wrapper).query(*args)
+      {
+        success: true,
+        data: data,
+        last_update: Time.now.utc
+      }
+    rescue Gitlab::PrometheusClient::Error => err
+      { success: false, result: err.message }
+    end
+  end
+end
diff --git a/app/models/deployment.rb b/app/models/deployment.rb
index b6cf168d60e..66e61c06765 100644
--- a/app/models/deployment.rb
+++ b/app/models/deployment.rb
@@ -98,28 +98,29 @@ class Deployment < ActiveRecord::Base
   end
 
   def has_metrics?
-    project.monitoring_service.present?
+    prometheus_adapter&.can_query?
   end
 
   def metrics
     return {} unless has_metrics?
 
-    project.monitoring_service.deployment_metrics(self)
-  end
-
-  def has_additional_metrics?
-    project.prometheus_service.present?
+    metrics = prometheus_adapter.query(:deployment, self)
+    metrics&.merge(deployment_time: created_at.to_i) || {}
   end
 
   def additional_metrics
-    return {} unless project.prometheus_service.present?
+    return {} unless has_metrics?
 
-    metrics = project.prometheus_service.additional_deployment_metrics(self)
+    metrics = prometheus_adapter.query(:additional_metrics_deployment, self)
     metrics&.merge(deployment_time: created_at.to_i) || {}
   end
 
   private
 
+  def prometheus_adapter
+    environment.prometheus_adapter
+  end
+
   def ref_path
     File.join(environment.ref_path, 'deployments', iid.to_s)
   end
diff --git a/app/models/environment.rb b/app/models/environment.rb
index f78c21aebe5..24d4f1d8761 100644
--- a/app/models/environment.rb
+++ b/app/models/environment.rb
@@ -146,21 +146,19 @@ class Environment < ActiveRecord::Base
   end
 
   def has_metrics?
-    project.monitoring_service.present? && available? && last_deployment.present?
+    prometheus_adapter&.can_query? && available? && last_deployment.present?
   end
 
   def metrics
-    project.monitoring_service.environment_metrics(self) if has_metrics?
+    prometheus_adapter.query(:environment, self) if has_metrics?
   end
 
-  def has_additional_metrics?
-    project.prometheus_service.present? && available? && last_deployment.present?
+  def additional_metrics
+    prometheus_adapter.query(:additional_metrics_environment, self) if has_metrics?
   end
 
-  def additional_metrics
-    if has_additional_metrics?
-      project.prometheus_service.additional_environment_metrics(self)
-    end
+  def prometheus_adapter
+    @prometheus_adapter ||= Prometheus::AdapterService.new(project, deployment_platform).prometheus_adapter
   end
 
   def slug
@@ -226,6 +224,10 @@ class Environment < ActiveRecord::Base
     self.environment_type || self.name
   end
 
+  def deployment_platform
+    project.deployment_platform
+  end
+
   private
 
   # Slugifying a name may remove the uniqueness guarantee afforded by it being
diff --git a/app/models/project_services/monitoring_service.rb b/app/models/project_services/monitoring_service.rb
index ee9cd78327a..9af68b4e821 100644
--- a/app/models/project_services/monitoring_service.rb
+++ b/app/models/project_services/monitoring_service.rb
@@ -9,11 +9,11 @@ class MonitoringService < Service
     %w()
   end
 
-  def environment_metrics(environment)
+  def can_query?
     raise NotImplementedError
   end
 
-  def deployment_metrics(deployment)
+  def query(_, *_)
     raise NotImplementedError
   end
 end
diff --git a/app/models/project_services/prometheus_service.rb b/app/models/project_services/prometheus_service.rb
index 58731451429..dcaeb65dc32 100644
--- a/app/models/project_services/prometheus_service.rb
+++ b/app/models/project_services/prometheus_service.rb
@@ -1,9 +1,5 @@
 class PrometheusService < MonitoringService
-  include ReactiveService
-
-  self.reactive_cache_lease_timeout = 30.seconds
-  self.reactive_cache_refresh_interval = 30.seconds
-  self.reactive_cache_lifetime = 1.minute
+  include PrometheusAdapter
 
   #  Access to prometheus is directly through the API
   prop_accessor :api_url
@@ -13,7 +9,7 @@ class PrometheusService < MonitoringService
     validates :api_url, url: true
   end
 
-  before_save :synchronize_service_state!
+  before_save :synchronize_service_state
 
   after_save :clear_reactive_cache!
 
@@ -66,63 +62,15 @@ class PrometheusService < MonitoringService
 
   # Check we can connect to the Prometheus API
   def test(*args)
-    client.ping
+    Gitlab::PrometheusClient.new(prometheus_client).ping
 
     { success: true, result: 'Checked API endpoint' }
   rescue Gitlab::PrometheusClient::Error => err
     { success: false, result: err }
   end
 
-  def environment_metrics(environment)
-    with_reactive_cache(Gitlab::Prometheus::Queries::EnvironmentQuery.name, environment.id, &rename_field(:data, :metrics))
-  end
-
-  def deployment_metrics(deployment)
-    metrics = with_reactive_cache(Gitlab::Prometheus::Queries::DeploymentQuery.name, deployment.environment.id, deployment.id, &rename_field(:data, :metrics))
-    metrics&.merge(deployment_time: deployment.created_at.to_i) || {}
-  end
-
-  def additional_environment_metrics(environment)
-    with_reactive_cache(Gitlab::Prometheus::Queries::AdditionalMetricsEnvironmentQuery.name, environment.id, &:itself)
-  end
-
-  def additional_deployment_metrics(deployment)
-    with_reactive_cache(Gitlab::Prometheus::Queries::AdditionalMetricsDeploymentQuery.name, deployment.environment.id, deployment.id, &:itself)
-  end
-
-  def matched_metrics
-    with_reactive_cache(Gitlab::Prometheus::Queries::MatchedMetricsQuery.name, &:itself)
-  end
-
-  # Cache metrics for specific environment
-  def calculate_reactive_cache(query_class_name, *args)
-    return unless active? && project && !project.pending_delete?
-
-    environment_id = args.first
-    client = client(environment_id)
-
-    data = Kernel.const_get(query_class_name).new(client).query(*args)
-    {
-      success: true,
-      data: data,
-      last_update: Time.now.utc
-    }
-  rescue Gitlab::PrometheusClient::Error => err
-    { success: false, result: err.message }
-  end
-
-  def client(environment_id = nil)
-    if manual_configuration?
-      Gitlab::PrometheusClient.new(RestClient::Resource.new(api_url))
-    else
-      cluster = cluster_with_prometheus(environment_id)
-      raise Gitlab::PrometheusClient::Error, "couldn't find cluster with Prometheus installed" unless cluster
-
-      rest_client = client_from_cluster(cluster)
-      raise Gitlab::PrometheusClient::Error, "couldn't create proxy Prometheus client" unless rest_client
-
-      Gitlab::PrometheusClient.new(rest_client)
-    end
+  def prometheus_client
+    RestClient::Resource.new(api_url) if api_url && manual_configuration? && active?
   end
 
   def prometheus_installed?
@@ -134,32 +82,7 @@ class PrometheusService < MonitoringService
 
   private
 
-  def cluster_with_prometheus(environment_id = nil)
-    clusters = if environment_id
-                 ::Environment.find_by(id: environment_id).try do |env|
-                   # sort results by descending order based on environment_scope being longer
-                   # thus more closely matching environment slug
-                   project.clusters.enabled.for_environment(env).sort_by { |c| c.environment_scope&.length }.reverse!
-                 end
-               else
-                 project.clusters.enabled.for_all_environments
-               end
-
-    clusters&.detect { |cluster| cluster.application_prometheus&.installed? }
-  end
-
-  def client_from_cluster(cluster)
-    cluster.application_prometheus.proxy_client
-  end
-
-  def rename_field(old_field, new_field)
-    -> (metrics) do
-      metrics[new_field] = metrics.delete(old_field)
-      metrics
-    end
-  end
-
-  def synchronize_service_state!
+  def synchronize_service_state
     self.active = prometheus_installed? || manual_configuration?
 
     true
diff --git a/app/services/prometheus/adapter_service.rb b/app/services/prometheus/adapter_service.rb
new file mode 100644
index 00000000000..4504d2ccfe6
--- /dev/null
+++ b/app/services/prometheus/adapter_service.rb
@@ -0,0 +1,36 @@
+module Prometheus
+  class AdapterService
+    def initialize(project, deployment_platform = nil)
+      @project = project
+
+      @deployment_platform = if deployment_platform
+                               deployment_platform
+                             else
+                               project.deployment_platform
+                             end
+    end
+
+    attr_reader :deployment_platform, :project
+
+    def prometheus_adapter
+      @prometheus_adapter ||= if service_prometheus_adapter.can_query?
+                                service_prometheus_adapter
+                              else
+                                cluster_prometheus_adapter
+                              end
+    end
+
+    def service_prometheus_adapter
+      project.find_or_initialize_service('prometheus')
+    end
+
+    def cluster_prometheus_adapter
+      return unless deployment_platform.respond_to?(:cluster)
+
+      cluster = deployment_platform.cluster
+      return unless cluster.application_prometheus&.installed?
+
+      cluster.application_prometheus
+    end
+  end
+end