Add busy_threads metric to Puma integration (DataDog#2697)

* Add `busy_threads` metric to Puma integration

This new stat was [added about 6 months
ago](puma/puma#3517) to track "the overall
current state of work to be done and the capacity to do it". It seems
like it could be useful to track, potentially for autoscaling or other
applications.

* Update puma/metadata.csv

Co-authored-by: May Lee <[email protected]>

* Update puma/metadata.csv

Co-authored-by: Steven Yuen <[email protected]>

---------

Co-authored-by: May Lee <[email protected]>
Co-authored-by: Steven Yuen <[email protected]>

Author: 3 people

puma/datadog_checks/puma/puma.py

@@ -7,6 +7,7 @@
 METRICS = [
     ('backlog', 'backlog', AgentCheck.gauge),
     ('booted_workers', 'booted_workers', AgentCheck.gauge),
+    ('busy_threads', 'busy_threads', AgentCheck.gauge),
     ('max_threads', 'max_threads', AgentCheck.gauge),
     ('pool_capacity', 'pool_capacity', AgentCheck.gauge),
     ('requests_count', 'requests_count', AgentCheck.gauge),
@@ -32,6 +33,7 @@ def _extract_metrics(self, response):
         metrics = {
             'backlog': 0,
             'booted_workers': 0,
+            'busy_threads': 0,
             'max_threads': 0,
             'pool_capacity': 0,
             'requests_count': 0,
@@ -46,13 +48,15 @@ def _extract_metrics(self, response):
             for worker in response['worker_status']:
                 last_status = worker.get('last_status')
                 metrics['backlog'] += int(last_status.get('backlog', 0))
+                metrics['busy_threads'] += int(last_status.get('busy_threads', 0))
                 metrics['max_threads'] += int(last_status.get('max_threads', 0))
                 metrics['pool_capacity'] += int(last_status.get('pool_capacity', 0))
                 metrics['requests_count'] += int(last_status.get('requests_count', 0))
                 metrics['running'] += int(last_status.get('running', 0))
 
         else:  # Puma is not clustered
             metrics['backlog'] = int(response.get('backlog', 0))
+            metrics['busy_threads'] = int(response.get('busy_threads', 0))
             metrics['max_threads'] = int(response.get('max_threads', 0))
             metrics['pool_capacity'] = int(response.get('pool_capacity', 0))
             metrics['requests_count'] = int(response.get('requests_count', 0))

puma/metadata.csv

@@ -1,6 +1,7 @@
 metric_name,metric_type,interval,unit_name,per_unit_name,description,orientation,integration,short_name,curated_metric
 puma.backlog,gauge,,unit,,Pending request backlog,0,puma,request backlog,
 puma.booted_workers,gauge,,unit,,Number of booted puma workers,0,puma,booted workers,
+puma.busy_threads,gauge,,unit,,"Number of threads that are running, including requests waiting on a thread, but excluding requests waiting for work.",0,puma,booted workers,
 puma.max_threads,gauge,,unit,,Maximum threads,0,puma,max threads,
 puma.pool_capacity,gauge,,unit,,Pool capacity,0,puma,pool capacity,
 puma.requests_count,gauge,,unit,,Request count,0,puma,request count,

puma/tests/test_puma.py

@@ -29,6 +29,7 @@ def test_metrics_for_puma_in_cluster_mode(aggregator, instance):
                 {
                     'last_status': {
                         'backlog': 0,
+                        'busy_threads': 6,
                         'max_threads': 20,
                         'pool_capacity': 15,
                         'running': 20,
@@ -38,6 +39,7 @@ def test_metrics_for_puma_in_cluster_mode(aggregator, instance):
                 {
                     'last_status': {
                         'backlog': 1,
+                        'busy_threads': 4,
                         'max_threads': 20,
                         'pool_capacity': 15,
                         'running': 20,
@@ -58,6 +60,7 @@ def test_metrics_for_puma_in_cluster_mode(aggregator, instance):
     aggregator.assert_metric('puma.workers', value=2.0, tags=[])
     aggregator.assert_metric('puma.backlog', value=1.0, tags=[])
     aggregator.assert_metric('puma.booted_workers', value=2.0, tags=[])
+    aggregator.assert_metric('puma.busy_threads', value=10.0, tags=[])
     aggregator.assert_metric('puma.pool_capacity', value=30.0, tags=[])
     aggregator.assert_metric('puma.running', value=40.0, tags=[])
     aggregator.assert_metric('puma.requests_count', value=270.0, tags=[])
@@ -71,7 +74,9 @@ def test_metrics_for_puma_in_single_mode(aggregator, instance):
         'content-type': 'application/json',
     }
 
-    content = json.dumps({"backlog": 1, "running": 2, "pool_capacity": 15, "max_threads": 20, "requests_count": 120})
+    content = json.dumps(
+        {"backlog": 1, "busy_threads": 6, "running": 2, "pool_capacity": 15, "max_threads": 20, "requests_count": 120}
+    )
 
     response = Mock(headers=headers, content=content)
 
@@ -81,6 +86,7 @@ def test_metrics_for_puma_in_single_mode(aggregator, instance):
 
     aggregator.assert_metric('puma.max_threads', value=20.0, tags=[])
     aggregator.assert_metric('puma.backlog', value=1.0, tags=[])
+    aggregator.assert_metric('puma.busy_threads', value=6.0, tags=[])
     aggregator.assert_metric('puma.pool_capacity', value=15.0, tags=[])
     aggregator.assert_metric('puma.running', value=2.0, tags=[])
     aggregator.assert_metric('puma.requests_count', value=120.0, tags=[])
@@ -107,6 +113,7 @@ def test_check(aggregator, instance, test_server):
     aggregator.assert_metric('puma.workers', value=2.0, tags=[])
     aggregator.assert_metric('puma.backlog', value=0.0, tags=[])
     aggregator.assert_metric('puma.booted_workers', value=2.0, tags=[])
+    aggregator.assert_metric('puma.busy_threads', value=0.0, tags=[])
     aggregator.assert_metric('puma.pool_capacity', value=10.0, tags=[])
     aggregator.assert_metric('puma.running', value=10.0, tags=[])
     aggregator.assert_metric('puma.requests_count', value=1.0, tags=[])