Allow configuration of stop duration metric buckets

Issue #4833 proposes allowing configuration of buckets for server spawn
duration. It was resolved with PR #4967

This follows a similar pattern to support the same kind of configuration
for server stop duration

docs/source/reference/monitoring.md

@@ -33,14 +33,19 @@ export JUPYTERHUB_METRICS_PREFIX=jupyterhub_prod
 
 would result in the metric `jupyterhub_prod_active_users`, etc.
 
-## Customizing spawn bucket sizes
+## Customizing bucket sizes
 
-As of JupyterHub 5.3, override `JUPYTERHUB_SERVER_SPAWN_DURATION_SECONDS_BUCKETS` env variable in Hub's environment to allow custom bucket sizes. Otherwise default to, [0.5, 1, 2.5, 5, 10, 15, 30, 60, 120, 180, 300, 600, float("inf")]
+As of JupyterHub 5.3, the following environment variables in the Hub's environment can be overridden to support custom bucket sizes - below are the defaults:
+
+| Variable                                           | Default                                                            |
+| -------------------------------------------------- | ------------------------------------------------------------------ |
+| `JUPYTERHUB_SERVER_SPAWN_DURATION_SECONDS_BUCKETS` | `0.5,1,2.5,5,10,15,30,60,120,180,300,600,inf`                      |
+| `JUPYTERHUB_SERVER_STOP_DURATION_SECONDS_BUCKETS`  | `0.005,0.01,0.025,0.05,0.075,0.1,0.25,0.5,0.75,1,2.5,5,7.5,10,inf` |
 
 For example,
 
 ```bash
-export JUPYTERHUB_SERVER_SPAWN_DURATION_SECONDS_BUCKETS="1,2,4,6,12,30,60,120"
+export JUPYTERHUB_SERVER_SPAWN_DURATION_SECONDS_BUCKETS="1,2,4,6,12,30,60,120,inf"
 ```
 
 ## Configuring metrics

jupyterhub/metrics.py

@@ -37,12 +37,16 @@ from . import orm
 from .utils import utcnow
 
 metrics_prefix = os.getenv('JUPYTERHUB_METRICS_PREFIX', 'jupyterhub')
-_env_buckets = os.environ.get(
+
+_env_spawn_duration_buckets = os.environ.get(
     'JUPYTERHUB_SERVER_SPAWN_DURATION_SECONDS_BUCKETS', ""
 ).strip()
+_env_stop_duration_buckets = os.environ.get(
+    "JUPYTERHUB_SERVER_STOP_DURATION_SECONDS_BUCKETS", ""
+).strip()
 
-if _env_buckets:
-    spawn_duration_buckets = [float(_s) for _s in _env_buckets.split(",")]
+if _env_spawn_duration_buckets:
+    spawn_duration_buckets = [float(_s) for _s in _env_spawn_duration_buckets.split(",")]
 else:
     spawn_duration_buckets = [
         0.5,
@@ -60,6 +64,29 @@ else:
         float("inf"),
     ]
 
+if _env_stop_duration_buckets:
+    stop_duration_buckets = [float(_s) for _s in _env_stop_duration_buckets.split(",")]
+else:
+    # We default to the same buckets as upstream Prometheus (as it was before) so we don't
+    # break anything that was consuming this metric before bucket configuration was possible
+    stop_duration_buckets = [
+        0.005,
+        0.01,
+        0.025,
+        0.05,
+        0.075,
+        0.1,
+        0.25,
+        0.5,
+        0.75,
+        1,
+        2.5,
+        5,
+        7.5,
+        10,
+        float("inf"),
+    ]
+
 REQUEST_DURATION_SECONDS = Histogram(
     'request_duration_seconds',
     'Request duration for all HTTP requests',
@@ -197,6 +224,7 @@ SERVER_STOP_DURATION_SECONDS = Histogram(
     'server_stop_seconds',
     'Time taken for server stopping operation',
     ['status'],
+    buckets=stop_duration_buckets,
     namespace=metrics_prefix,
 )
 
@@ -341,7 +369,7 @@ class PeriodicMetricsCollector(LoggingConfigurable):
         config=True,
         help="""
         Enable event_loop_interval_seconds metric.
-        
+
         Measures event-loop responsiveness.
         """,
     )
@@ -350,7 +378,7 @@ class PeriodicMetricsCollector(LoggingConfigurable):
         config=True,
         help="""
         Interval (in seconds) on which to measure the event loop interval.
-        
+
         This is the _sensitivity_ of the `event_loop_interval` metric.
         Setting it too low (e.g. below 20ms) can end up slowing down the whole event loop
         by measuring too often,