Spaces:
Runtime error
Runtime error
VenkateshRoshan
commited on
Commit
·
7115bd6
1
Parent(s):
ff4ac0d
health check added
Browse files
app.py
CHANGED
|
@@ -85,7 +85,6 @@ class CustomerSupportBot:
|
|
| 85 |
}
|
| 86 |
return usage
|
| 87 |
|
| 88 |
-
|
| 89 |
def create_chat_interface():
|
| 90 |
bot = CustomerSupportBot(model_path="/app/models")
|
| 91 |
|
|
@@ -164,6 +163,46 @@ def create_chat_interface():
|
|
| 164 |
# Add keyboard shortcut for submit
|
| 165 |
msg.change(lambda x: gr.update(interactive=bool(x.strip())), inputs=[msg], outputs=[submit])
|
| 166 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 167 |
return interface
|
| 168 |
|
| 169 |
if __name__ == "__main__":
|
|
@@ -173,4 +212,4 @@ if __name__ == "__main__":
|
|
| 173 |
server_name="0.0.0.0", # Makes the server accessible from other machines
|
| 174 |
server_port=7860, # Specify the port
|
| 175 |
debug=True
|
| 176 |
-
)
|
|
|
|
| 85 |
}
|
| 86 |
return usage
|
| 87 |
|
|
|
|
| 88 |
def create_chat_interface():
|
| 89 |
bot = CustomerSupportBot(model_path="/app/models")
|
| 90 |
|
|
|
|
| 163 |
# Add keyboard shortcut for submit
|
| 164 |
msg.change(lambda x: gr.update(interactive=bool(x.strip())), inputs=[msg], outputs=[submit])
|
| 165 |
|
| 166 |
+
# Add health check endpoint
|
| 167 |
+
@interface.route("/ping", methods=["GET"])
|
| 168 |
+
def ping():
|
| 169 |
+
try:
|
| 170 |
+
# Check if model and tokenizer are loaded
|
| 171 |
+
if not hasattr(bot, 'model') or not hasattr(bot, 'tokenizer'):
|
| 172 |
+
return {"status": "unhealthy", "reason": "Model or tokenizer not loaded"}, 503
|
| 173 |
+
|
| 174 |
+
# Check if CUDA is available and model is on the correct device
|
| 175 |
+
if torch.cuda.is_available():
|
| 176 |
+
if not bot.model.device.type == 'cuda':
|
| 177 |
+
return {"status": "unhealthy", "reason": "Model not on GPU"}, 503
|
| 178 |
+
|
| 179 |
+
# Check memory usage
|
| 180 |
+
usage = bot.monitor_resources()
|
| 181 |
+
if usage["RAM (GB)"] > 30: # Example threshold
|
| 182 |
+
return {"status": "unhealthy", "reason": "High memory usage"}, 503
|
| 183 |
+
|
| 184 |
+
# Try a quick model inference to ensure it's working
|
| 185 |
+
try:
|
| 186 |
+
test_response = bot.generate_response("Test message")
|
| 187 |
+
if test_response.startswith("An error occurred"):
|
| 188 |
+
return {"status": "unhealthy", "reason": "Model inference failed"}, 503
|
| 189 |
+
except Exception as e:
|
| 190 |
+
return {"status": "unhealthy", "reason": f"Model inference error: {str(e)}"}, 503
|
| 191 |
+
|
| 192 |
+
return {
|
| 193 |
+
"status": "healthy",
|
| 194 |
+
"model_loaded": True,
|
| 195 |
+
"device": bot.device,
|
| 196 |
+
"resources": usage
|
| 197 |
+
}
|
| 198 |
+
except Exception as e:
|
| 199 |
+
return {"status": "unhealthy", "reason": str(e)}, 503
|
| 200 |
+
|
| 201 |
+
# Add secondary health endpoint
|
| 202 |
+
@interface.route("/health", methods=["GET"])
|
| 203 |
+
def health():
|
| 204 |
+
return {"status": "healthy"}
|
| 205 |
+
|
| 206 |
return interface
|
| 207 |
|
| 208 |
if __name__ == "__main__":
|
|
|
|
| 212 |
server_name="0.0.0.0", # Makes the server accessible from other machines
|
| 213 |
server_port=7860, # Specify the port
|
| 214 |
debug=True
|
| 215 |
+
)
|