[AGENT-6113] Increase timeout and limit payload size of nginx (#4)

* [AGENT-6113] Increase timeout and limit payload size of nginx

* update test

environments/java_codegen_monitoring/VERSION

@@ -1 +1 @@
-1.0.1
+1.0.2

environments/java_codegen_monitoring/nginx.conf

@@ -6,12 +6,17 @@ events {
 }
 
 http {
+    log_format  main  '$remote_addr - $remote_user [$time_local] [$request] status: $status request_time: $request_time  '
+        'bytes: $body_bytes_sent upstream: [$upstream_addr]';
+    access_log /dev/stdout main;
+    error_log /dev/stdout;
+
     server {
         listen 9001;
         large_client_header_buffers 4 32k;
         sendfile            on;
         keepalive_timeout   65;
-        client_max_body_size 20M;
+        client_max_body_size 1M;
 
 
         # Remove v1 prefix from URL
@@ -21,6 +26,9 @@ http {
         # Redirect to port 8080 where drum is running
         location / {
             proxy_pass http://localhost:8080/;
+            proxy_send_timeout 120s;
+            proxy_read_timeout 120s;
+            proxy_connect_timeout 120s;
         }
     }
 }

tests/integration/test_make_predictions.py

@@ -8,6 +8,7 @@
 import contextlib
 import logging
 import time
+from io import BytesIO
 
 import pandas as pd
 import pytest
@@ -19,6 +20,7 @@
 SAP_MLOPS_INTEGRATION_IMAGE = (
     "ghcr.io/datarobot-oss/mlops-sap-monitoring-scoring-code:latest"
 )
+MAX_ROWS_PER_REQUEST = 1000
 
 
 def _wait_for_container_healthy(container, ping_url, timeout=100, interval=2):
@@ -96,14 +98,47 @@ def test_run_server(
 
     base_url = "http://localhost:9001"
     with run_sap_integration_image(env_vars, f"{base_url}/v1/ping/"):
-        # Make predictions
         dataset = model_packages[model_package_id].get("model_dataset")
+        data_chunks = pd.read_csv(dataset, chunksize=MAX_ROWS_PER_REQUEST)
+
+        for chunk in data_chunks:
+            # Convert the chunk to a CSV byte stream
+            csv_chunk = BytesIO()
+            chunk.to_csv(csv_chunk, index=False)
+            csv_chunk.seek(0)
+
+            # Make predictions
+            response = requests.post(
+                f"{base_url}/v1/predict/",
+                data=csv_chunk,
+            )
+            assert response.status_code == 200
+
+            # assert number of predictions
+            predictions = response.json().get("predictions", [])
+            assert len(predictions) == len(chunk)
+
+
+def test_run_server_large_request(
+    mock_dr_app,
+    mock_dr_app_port,
+    run_sap_integration_image,
+    model_packages,
+):
+    env_vars = {
+        "MLOPS_MODEL_PACKAGE_ID": "10k_diabetes_package",
+        "TARGET_TYPE": "binary",
+        "DATAROBOT_ENDPOINT": f"http://host.docker.internal:{mock_dr_app_port}",
+        "DATAROBOT_API_TOKEN": "secret_token",
+        "POSITIVE_CLASS_LABEL": "True",
+        "NEGATIVE_CLASS_LABEL": "False",
+    }
+
+    base_url = "http://localhost:9001"
+    with run_sap_integration_image(env_vars, f"{base_url}/v1/ping/"):
+        # Make large number of predictions in a single request
+        dataset = model_packages["10k_diabetes_package"].get("model_dataset")
         response = requests.post(
             f"{base_url}/v1/predict/", data=dataset.open(mode="rb")
         )
-        assert response.status_code == 200
-
-        # assert number of predictions
-        predictions = response.json().get("predictions", [])
-        dataset = pd.read_csv(dataset)
-        assert len(predictions) == len(dataset)
+        assert response.status_code == 413