Add supervisor specific metrics (#54)

README.md

@@ -21,6 +21,7 @@ Supervisor is a Docker GitOps service that allows you to manage `docker-compose`
       * [Delete Stack](#delete-stack)
       * [Get Stack Log](#get-stack-log)
       * [Control Stack](#control-stack)
+   * [Metrics](#metrics)
    * [Dashboard](#dashboard)
    * [License](#license)
    * [Is it any good?](#is-it-any-good)
@@ -237,6 +238,19 @@ curl --request POST \
   https://supervisor.example.com/stacks/<stack_uuid>/control
 ```
 
+## Metrics
+
+To retrieve Prometheus metrics, you can access the
+`http://supervisor.example.com:9394/metrics` endpoint.
+
+  * `supervisor_total_stacks`: The total number of stacks. (gauge)
+  * `supervisor_total_healthy_stacks`: The total number of healthy stacks. (gauge)
+  * `supervisor_total_unhealthy_stacks`: The total number of unhealthy stacks. (gauge)
+  * `supervisor_jobs_execution_time`: The time taken to execute stack jobs, measured in seconds. (histogram)
+  * `supervisor_jobs_executed_total`: The total number of stack jobs executed. (counter)
+  * `supervisor_jobs_succeeded_total`: The total number of stack jobs that succeeded. (counter)
+  * `supervisor_jobs_failed_total`: The total number of stack jobs that failed. (counter)
+
 ## Dashboard
 
 Supervisor provides a simple dashboard to view and monitor stacks. The

app/jobs/stack_job.rb

@@ -21,7 +21,9 @@ def perform(stack)
 
   def execute
     script = render_script(@stack, @assets)
-    run_script(script)
+    Yabeda.supervisor.stack_jobs_execution_time.measure do
+      run_script(script)
+    end
     return if instance_of?(StackDestroyJob)
 
     stack_log

app/jobs/stack_job/handles_execute_result.rb

@@ -26,11 +26,11 @@ def stack_log
       end
 
       def stack_stats
-        @stack.update_stats(failed: error?, action: __action.split.second)
+        @stack.update_stats(succeeded: success?, action: __action.split.second)
       end
 
       def stack_health
-        @stack.update(healthy: !error?)
+        @stack.update(healthy: success?)
       end
 
       def __action

app/models/stack/has_stats.rb

@@ -12,17 +12,31 @@ def stats
         )
       end
 
-      def update_stats(failed: false, action: nil)
-        processed = self.processed + 1
-        failed = failed ? self.failed + 1 : self.failed
-        last_run = processed ? Time.current : self.last_run
-        last_action = action || 'unknown'
-
+      def update_stats(succeeded: false, action: nil)
         update(
-          processed: processed,
-          failed: failed,
-          last_run: last_run,
-          last_action: last_action
+          processed: processed + 1,
+          failed: succeeded ? failed : failed + 1,
+          last_run: Time.current,
+          last_action: action || 'unknown'
+        )
+
+        status = succeeded ? 'succeeded' : 'failed'
+
+        Yabeda.supervisor.stack_jobs_executed_total.increment(
+          {
+            name: name,
+            action: last_action,
+            status: succeeded ? 'succeeded' : 'failed'
+          },
+          by: 1
+        )
+
+        Yabeda.supervisor.send(:"stack_jobs_#{status}_total").increment(
+          {
+            name: name,
+            action: last_action
+          },
+          by: 1
         )
       end
     end

config/initializers/metrics.rb

@@ -0,0 +1,43 @@
+require 'prometheus/client/support/puma'
+
+Prometheus::Client.configuration.pid_provider = Prometheus::Client::Support::Puma.method(:worker_pid_provider)
+
+Yabeda.configure do
+  group :supervisor do
+    counter :stack_jobs_executed_total do
+      comment 'The total number of stack jobs executed.'
+      tags %i[name action status]
+    end
+    counter :stack_jobs_failed_total do
+      comment 'The total number of stack jobs that failed.'
+      tags %i[name action]
+    end
+    counter :stack_jobs_succeeded_total do
+      comment 'The total number of stack jobs that succeeded.'
+      tags %i[name action]
+    end
+    histogram :stack_jobs_execution_time do
+      comment 'The time taken to execute stack jobs, measured in seconds.'
+      unit :seconds
+      buckets [
+        0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10,
+        30, 60, 120, 300, 1800, 3600, 86_400
+      ].freeze
+    end
+    gauge :total_stacks do
+      comment 'The total number of stacks.'
+    end
+    gauge :total_healthy_stacks do
+      comment 'The total number of healthy stacks.'
+    end
+    gauge :total_unhealthy_stacks do
+      comment 'The total number of unhealthy stacks.'
+    end
+
+    collect do
+      supervisor.total_stacks.set({}, Stack.count)
+      supervisor.total_healthy_stacks.set({}, Stack.where(healthy: true).count)
+      supervisor.total_unhealthy_stacks.set({}, Stack.where(healthy: false).count)
+    end
+  end
+end