add llama log

lib/backend.py

@@ -30,7 +30,7 @@ from lib.data_types import (
     BenchmarkResult
 )
 
-VERSION = "0.2.1"
+VERSION = "0.2.0"
 
 MSG_HISTORY_LEN = 100
 log = logging.getLogger(__file__)

lib/metrics.py

@@ -146,7 +146,6 @@ class Metrics:
     def _set_mtoken(self, mtoken: str) -> None:
         self.mtoken = mtoken
 
-
     #######################################Private#######################################
 
     async def __send_delete_requests_and_reset(self):
@@ -217,7 +216,7 @@ class Metrics:
                 id=self.id,
                 mtoken=self.mtoken,
                 version=self.version,
-                loadtime=(loadtime_snapshot or 0.0),
+                loadtime=(loadtime_snapshot or 0.0), 
                 new_load=self.model_metrics.workload_processing,
                 cur_load=self.model_metrics.cur_load,
                 rej_load=self.model_metrics.workload_rejected,
@@ -281,6 +280,7 @@ class Metrics:
 
         if sent:
             # clear the one-shot loadtime only if we actually sent *this* value
+            self.system_metrics.reset(expected=loadtime_snapshot)
             self.update_pending = False
             self.model_metrics.reset()
             self.last_metric_update = time.time()

workers/openai/server.py

@@ -11,6 +11,7 @@ MODEL_SERVER_START_LOG_MSG = [
     "llama runner started",  # Ollama
     '"message":"Connected","target":"text_generation_router"',  # TGI
     '"message":"Connected","target":"text_generation_router::server"',  # TGI
+    "main: model loaded" # llama.cpp
 ]
 
 MODEL_SERVER_ERROR_LOG_MSGS = [