(fix) don't block proxy startup if license check fails & using prometheus (#6839)

* fix - don't block proxy startup if not a premium user

* test_litellm_proxy_server_config_with_prometheus

* add test for proxy startup

* fix remove unused test

* fix startup test

* add comment on bad-license

.circleci/config.yml

@@ -1039,6 +1039,48 @@ jobs:
             ls
             python -m pytest -vv tests/otel_tests -x --junitxml=test-results/junit.xml --durations=5
           no_output_timeout: 120m
+           # Clean up first container
+      - run:
+          name: Stop and remove first container
+          command: |
+            docker stop my-app
+            docker rm my-app
+
+      # Second Docker Container Run with Different Config
+      # NOTE: We intentionally pass a "bad" license here. We need to ensure proxy starts and serves request even with bad license
+      - run:
+          name: Run Second Docker container
+          command: |
+            docker run -d \
+              -p 4000:4000 \
+              -e DATABASE_URL=$PROXY_DATABASE_URL \
+              -e REDIS_HOST=$REDIS_HOST \
+              -e REDIS_PASSWORD=$REDIS_PASSWORD \
+              -e REDIS_PORT=$REDIS_PORT \
+              -e LITELLM_MASTER_KEY="sk-1234" \
+              -e OPENAI_API_KEY=$OPENAI_API_KEY \
+              -e LITELLM_LICENSE="bad-license" \
+              --name my-app-3 \
+              -v $(pwd)/litellm/proxy/example_config_yaml/enterprise_config.yaml:/app/config.yaml \
+              my-app:latest \
+              --config /app/config.yaml \
+              --port 4000 \
+              --detailed_debug
+
+      - run:
+          name: Start outputting logs for second container
+          command: docker logs -f my-app-2
+          background: true
+
+      - run:
+          name: Wait for second app to be ready
+          command: dockerize -wait http://localhost:4000 -timeout 5m
+
+      - run:
+          name: Run second round of tests
+          command: |
+            python -m pytest -vv tests/basic_proxy_startup_tests -x --junitxml=test-results/junit-2.xml --durations=5
+          no_output_timeout: 120m
 
       # Store test results
       - store_test_results:

litellm/proxy/common_utils/callback_utils.py

@@ -242,7 +242,9 @@ def initialize_callbacks_on_proxy(  # noqa: PLR0915
 
         if "prometheus" in value:
             if premium_user is not True:
-                raise Exception(CommonProxyErrors.not_premium_user.value)
+                verbose_proxy_logger.warning(
+                    f"Prometheus metrics are only available for premium users. {CommonProxyErrors.not_premium_user.value}"
+                )
             from litellm.proxy.proxy_server import app
 
             verbose_proxy_logger.debug("Starting Prometheus Metrics on /metrics")

litellm/proxy/example_config_yaml/enterprise_config.yaml

@@ -0,0 +1,17 @@
+model_list:
+ - model_name: gpt-4
+   litellm_params:
+     model: openai/fake
+     api_key: fake-key
+     api_base: https://exampleopenaiendpoint-production.up.railway.app/
+     tags: ["teamA"]
+   model_info:
+     id: "team-a-model"
+
+litellm_settings:
+  cache: true
+  callbacks: ["prometheus"]
+
+router_settings:
+  enable_tag_filtering: True # 👈 Key Change
+

tests/basic_proxy_startup_tests/test_basic_proxy_startup.py

@@ -0,0 +1,51 @@
+"""
+This test ensures that the proxy starts and serves requests even with a bad license.
+
+
+in ci/cd config.yml, we set the license to "bad-license"
+"""
+
+import pytest
+import aiohttp
+from typing import Optional
+
+
+@pytest.mark.asyncio
+async def test_health_and_chat_completion():
+    """
+    Test health endpoints and chat completion:
+    1. Check /health/readiness
+    2. Check /health/liveness
+    3. Make a chat completion call
+    """
+    async with aiohttp.ClientSession() as session:
+        # Test readiness endpoint
+        async with session.get("http://0.0.0.0:4000/health/readiness") as response:
+            assert response.status == 200
+            readiness_response = await response.json()
+            assert readiness_response["status"] == "connected"
+
+        # Test liveness endpoint
+        async with session.get("http://0.0.0.0:4000/health/liveness") as response:
+            assert response.status == 200
+            liveness_response = await response.json()
+            print("liveness_response", liveness_response)
+
+        # Make a chat completion call
+        url = "http://0.0.0.0:4000/chat/completions"
+        headers = {
+            "Authorization": "Bearer sk-1234",
+            "Content-Type": "application/json",
+        }
+        data = {
+            "model": "gpt-4",
+            "messages": [
+                {"role": "system", "content": "You are a helpful assistant."},
+                {"role": "user", "content": "Hello!"},
+            ],
+        }
+
+        async with session.post(url, headers=headers, json=data) as response:
+            assert response.status == 200
+            completion_response = await response.json()
+            assert "choices" in completion_response