cache the home page in the background

llm_server/config/config.py

@@ -34,7 +34,8 @@ config_default_vars = {
     'openai_moderation_enabled': True,
     'netdata_root': None,
     'show_backends': True,
-    'cluster_workers': 30
+    'cluster_workers': 30,
+    'background_homepage_cacher': True
 }
 config_required_vars = ['cluster', 'mode', 'llm_middleware_name']
 

llm_server/config/load.py

@@ -53,6 +53,7 @@ def load_config(config_path):
     opts.openai_moderation_enabled = config['openai_moderation_enabled']
     opts.show_backends = config['show_backends']
     opts.cluster_workers = config['cluster_workers']
+    opts.background_homepage_cacher = config['background_homepage_cacher']
 
     if opts.openai_expose_our_model and not opts.openai_api_key:
         print('If you set openai_epose_our_model to false, you must set your OpenAI key in openai_api_key.')

llm_server/opts.py

@@ -38,3 +38,4 @@ openai_moderation_enabled = True
 cluster = {}
 show_backends = True
 cluster_workers = 30
+background_homepage_cacher = True

llm_server/routes/v1/__init__.py

@@ -14,5 +14,5 @@ def fallback(path):
     response_msg = format_sillytavern_err(error_msg, error_type='API')
     return jsonify({
         'results': [{'text': response_msg}],
-        'result': f'Wrong API path, visit {base_client_api} for more info'
+        'result': f'Wrong API path, visit {base_client_api} for more info.'
     }), 200  # return 200 so we don't trigger an error message in the client's ST

llm_server/workers/mainer.py

@@ -1,11 +1,13 @@
 import time
 
+import requests
+
 from llm_server import opts
-from llm_server.cluster.backend import get_a_cluster_backend, get_backends
+from llm_server.cluster.backend import get_backends
 from llm_server.cluster.cluster_config import cluster_config
 from llm_server.custom_redis import redis
 from llm_server.database.database import weighted_average_column_for_model
-from llm_server.llm.info import get_info, get_running_model
+from llm_server.llm.info import get_info
 
 
 def main_background_thread():
@@ -26,6 +28,14 @@ def main_background_thread():
                 cluster_config.set_backend_value(backend_url, 'average_output_tokens', average_output_tokens)
             if average_generation_elapsed_sec and average_output_tokens:
                 cluster_config.set_backend_value(backend_url, 'estimated_avg_tps', estimated_avg_tps)
+
+        if opts.background_homepage_cacher:
+            try:
+                base_client_api = redis.get('base_client_api', dtype=str)
+                r = requests.get('https://' + base_client_api, timeout=5)
+            except Exception as e:
+                print(f'Failed fetch the homepage - {e.__class__.__name__}: {e}')
+
         time.sleep(30)
 
 

templates/home.html

@@ -111,7 +111,7 @@
         <h3>Instructions</h3>
         <div id="instructions">
             <ol>
-                <li>In Settings > Power User Options, enable <kbd>Relaxed API URLS</kbd></li>
+                <li>In Settings > Power User Options, enable <kbd>Relaxed API URLS</kbd>.</li>
                 <li>Set your API type to <kbd>{{ mode_name }}</kbd></li>
                 <li>Enter <kbd>{{ client_api }}</kbd> in the <kbd>{{ api_input_textbox }}</kbd> textbox.</li>
                 {% if enable_streaming %}
@@ -167,17 +167,10 @@
     </div>
     <br>
     {% endfor %}
-
-    <!--
-    <div class="info-box">
-        <pre><code class="language-json" style="background-color: white">{{ stats_json|safe }}</code></pre>
-    </div>
-    -->
 </div>
 <div class="footer">
     <a href="https://git.evulid.cc/cyberes/local-llm-server" target="_blank">git.evulid.cc/cyberes/local-llm-server</a>
 </div>
-<script>hljs.highlightAll();</script>
 </body>
 
 </html>