Add jetstream_request_success_count metric (#124)

* first commit

* Count -> Counter

* typo

jetstream/core/metrics/prometheus.py

@@ -16,7 +16,7 @@
 
 import os
 import shortuuid
-from prometheus_client import Gauge
+from prometheus_client import Counter, Gauge
 
 
 class JetstreamMetricsCollector:
@@ -55,6 +55,11 @@ def __new__(cls):
       documentation="Total time taken to start the Jetstream server",
       labelnames=["id"],
   )
+  _request_success_count = Counter(
+      name="jetstream_request_success_count",
+      documentation="Number of requests successfully completed",
+      labelnames=["id"],
+  )
 
   def get_prefill_backlog_metric(self):
     return self._prefill_backlog.labels(id=self._id)
@@ -70,3 +75,6 @@ def get_slots_used_percentage_metric(self, idx: int):
 
   def get_server_startup_latency_metric(self):
     return self._server_startup_latency.labels(id=self._id)
+
+  def get_request_success_count_metric(self):
+    return self._request_success_count.labels(id=self._id)

jetstream/core/orchestrator.py

@@ -780,6 +780,8 @@ def _detokenize_thread(self, idx: int):
             # Return some output samples.
             request.enqueue_samples(results)
             if request.complete.all():
+              if self._metrics_collector:
+                self._metrics_collector.get_request_success_count_metric().inc()
               request.return_channel.close()
               # Place the slot back on the free queue.
               my_live_requests[slot] = None