[Metric] custom metrics refinement (#10861)

* In progress

* In Progress.

* Addressed code review.

* Add unit tests.

* Add a simple doc.

* Fixed test failure.

* Fix all test failures from serve.

* Addressed code review.

doc/source/package-ref.rst

@@ -167,6 +167,35 @@ get_current_placement_group
 
 .. autofunction:: ray.util.placement_group.get_current_placement_group
 
+.. _custom-metric-api-ref:
+
+Custom Metrics APIs
+-------------------
+
+Metric
+~~~~~~
+
+.. autoclass:: ray.util.metrics.Metric
+   :members:
+
+Count
+~~~~~
+
+.. autoclass:: ray.util.metrics.Count
+   :members:
+
+Gauge
+~~~~~
+
+.. autoclass:: ray.util.metrics.Gauge
+   :members:
+
+Histogram
+~~~~~~~~~
+
+.. autoclass:: ray.util.metrics.Histogram
+   :members:
+
 Experimental APIs
 -----------------
 

doc/source/ray-metrics.rst

@@ -1,9 +1,10 @@
-Ray Monitoring with Prometheus
-==============================
+Ray Monitoring
+==============
 To help monitoring Ray applications, Ray
 
 - Collects Ray's pre-selected system level metrics.
 - Exposes metrics in a Prometheus format. We'll call the endpoint to access these metrics a Prometheus endpoint.
+- Support custom metrics APIs that resemble Prometheus `metric types <https://prometheus.io/docs/concepts/metric_types/>`_.
 
 This page describes how to acces these metrics using Prometheus.
 
@@ -151,3 +152,11 @@ Now, modify a Prometheus config to scrape the file for service discovery.
         - '/tmp/ray/prom_metrics_service_discovery.json'
 
 Prometheus will automatically detect that the file contents are changing and update addresses it scrapes to based on the service discovery file generated by Ray.
+
+Custom Metrics
+--------------
+Ray supports custom metrics APIs to enable developers to have visibility to their applications.
+
+It current supports 3 metric types. All metric types have the same definition as `Prometheus metric types <https://prometheus.io/docs/concepts/metric_types/>`_.
+
+:ref:`Custom Metrics APIs Package Reference <custom-metric-api-ref>`