add llama log

lib/backend.py

@@ -431,4 +431,4 @@ class Backend:
             if os.path.isfile(self.model_log_file) is True:
                 return await tail_log()
             else:
-                await asyncio.sleep(1)
+                await sleep(1)

start_server.sh

@@ -124,5 +124,9 @@ cd "$SERVER_DIR"
 
 echo "launching PyWorker server"
 
+# if instance is rebooted, we want to clear out the log file so pyworker doesn't read lines
+# from the run prior to reboot. past logs are saved in $MODEL_LOG.old for debugging only
+[ -e "$MODEL_LOG" ] && cat "$MODEL_LOG" >> "$MODEL_LOG.old" && : > "$MODEL_LOG"
+
 (python3 -m "workers.$BACKEND.server" |& tee -a "$PYWORKER_LOG") &
 echo "launching PyWorker server done"

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 = [