[Frontend] correctly record prefill and decode time metrics (#10853)

Signed-off-by: Tomer Asida <tomera@ai21.com>
vllm-project · Dec 3, 2024 · 7c32b68 · 7c32b68
1 parent 7090c27
commit 7c32b68
Showing 1 changed file with 2 additions and 2 deletions.
diff --git a/vllm/engine/metrics.py b/vllm/engine/metrics.py
@@ -599,9 +599,9 @@ def _log_prometheus(self, stats: Stats) -> None:
                             stats.time_queue_requests)
         self._log_histogram(self.metrics.histogram_inference_time_request,
                             stats.time_inference_requests)
-        self._log_histogram(self.metrics.histogram_decode_time_request,
-                            stats.time_prefill_requests)
         self._log_histogram(self.metrics.histogram_prefill_time_request,
+                            stats.time_prefill_requests)
+        self._log_histogram(self.metrics.histogram_decode_time_request,
                             stats.time_decode_requests)
         self._log_histogram(self.metrics.histogram_time_in_queue_request,
                             stats.time_in_queue_requests)