update service

llm_server/pre_fork.py

@@ -1,3 +1,4 @@
+import time
 import sys
 
 from llm_server.custom_redis import redis
@@ -5,7 +6,7 @@ from llm_server.logging import create_logger
 
 
 def server_startup(s):
-    if not redis.get('daemon_started', dtype=bool):
     logger = create_logger('gunicorn')
-        logger.fatal('Could not find the key daemon_started in Redis. Did you forget to start the daemon process?')
-        sys.exit(1)
+    while not redis.get('daemon_started', dtype=bool):
+        logger.warning('Could not find the key daemon_started in Redis. Did you forget to start the daemon process?')
+        time.sleep(10)

other/local-llm-daemon.service

@@ -0,0 +1,15 @@
+[Unit]
+Description=Local LLM Proxy Daemon
+Wants=basic.target
+After=basic.target network.target
+
+[Service]
+User=server
+Group=server
+ExecStart=/srv/server/local-llm-server/venv/bin/python /srv/server/local-llm-server/daemon.py
+Restart=always
+RestartSec=2
+SyslogIdentifier=local-llm-daemon
+
+[Install]
+WantedBy=multi-user.target

other/local-llm-server.service

@@ -1,7 +1,8 @@
 [Unit]
-Description=Local LLM Proxy
+Description=Local LLM Proxy Server
 Wants=basic.target
-After=basic.target network.target
+After=basic.target network.target local-llm-daemon.service
+Requires=local-llm-daemon.service
 
 [Service]
 User=server
@@ -9,9 +10,10 @@ Group=server
 WorkingDirectory=/srv/server/local-llm-server
 # Need a lot of workers since we have long-running requests
 # Takes about 3.5G memory
-ExecStart=/srv/server/local-llm-server/venv/bin/gunicorn --workers 20 --bind 0.0.0.0:5000 server:app --timeout 60 --worker-class gevent
+ExecStart=/srv/server/local-llm-server/venv/bin/gunicorn --workers 20 --bind 0.0.0.0:5000 server:app --timeout 60 --worker-class gevent -c /srv/server/local-llm-server/other/server.conf
 Restart=always
 RestartSec=2
+SyslogIdentifier=local-llm-server
 
 [Install]
 WantedBy=multi-user.target

other/server.conf

@@ -0,0 +1,2 @@
+accesslog="-"
+errorlog="-"