Added metrics for number of kernels running labeled by type

jupyter_server/prometheus/metrics.py

@@ -19,3 +19,9 @@
     'terminal_currently_running_total',
     'counter for how many terminals are running',
 )
+
+KERNEL_CURRENTLY_RUNNING_TOTAL = Gauge(
+    'kernel_currently_running_total',
+    'counter for how many kernels are running labeled by type',
+    ['type']
+)

jupyter_server/services/kernels/kernelmanager.py

@@ -26,6 +26,8 @@
 from jupyter_server._tz import utcnow, isoformat
 from ipython_genutils.py3compat import getcwd
 
+from notebook.prometheus.metrics import KERNEL_CURRENTLY_RUNNING_TOTAL
+
 
 class MappingKernelManager(MultiKernelManager):
     """A KernelManager that handles
@@ -178,6 +180,13 @@ def start_kernel(self, kernel_id=None, path=None, **kwargs):
                 lambda : self._handle_kernel_died(kernel_id),
                 'dead',
             )
+
+            # Increase the metric of number of kernels running
+            # for the relevant kernel type by 1
+            KERNEL_CURRENTLY_RUNNING_TOTAL.labels(
+                type=self._kernels[kernel_id].kernel_name
+            ).inc()
+
         else:
             self._check_kernel_id(kernel_id)
             self.log.info("Using existing kernel: %s" % kernel_id)
@@ -288,6 +297,13 @@ def shutdown_kernel(self, kernel_id, now=False):
         self.stop_buffering(kernel_id)
         self._kernel_connections.pop(kernel_id, None)
         self.last_kernel_activity = utcnow()
+
+        # Decrease the metric of number of kernels
+        # running for the relevant kernel type by 1
+        KERNEL_CURRENTLY_RUNNING_TOTAL.labels(
+            type=self._kernels[kernel_id].kernel_name
+        ).dec()
+
         return super(MappingKernelManager, self).shutdown_kernel(kernel_id, now=now)
 
     def restart_kernel(self, kernel_id):

jupyter_server/terminal/api_handlers.py

@@ -12,6 +12,7 @@ def get(self):
         tm = self.terminal_manager
         terms = [{'name': name} for name in tm.terminals]
         self.finish(json.dumps(terms))
+
         # Update the metric below to the length of the list 'terms'
         TERMINAL_CURRENTLY_RUNNING_TOTAL.set(
             len(terms)
@@ -22,6 +23,7 @@ def post(self):
         """POST /terminals creates a new terminal and redirects to it"""
         name, _ = self.terminal_manager.new_named_terminal()
         self.finish(json.dumps({'name': name}))
+
         # Increase the metric by one because a new terminal was created
         TERMINAL_CURRENTLY_RUNNING_TOTAL.inc()
 
@@ -45,7 +47,10 @@ def delete(self, name):
             yield tm.terminate(name, force=True)
             self.set_status(204)
             self.finish()
-            # Decrease the metric below by one because a terminal has been shutdown
-            TERMINAL_CURRENTLY_RUNNING_TOTAL.dec(1)
+
+            # Decrease the metric below by one
+            # because a terminal has been shutdown
+            TERMINAL_CURRENTLY_RUNNING_TOTAL.dec()
+
         else:
             raise web.HTTPError(404, "Terminal not found: %r" % name)