Make load_sample_data more resilient (#2479)

Co-authored-by: Zack Krida <[email protected]>

ingestion_server/justfile

@@ -62,12 +62,17 @@ wait-up: up
 
 # Make a cURL POST request to the service with the given data
 _curl-post data host="localhost:50281":
-    curl \
+    STATUS_CODE=$(curl \
       -X POST \
       -H 'Content-Type: application/json' \
       -d '{{ data }}' \
-      -w "\n" \
-      'http://{{ host }}/task'
+      -o /dev/stderr \
+      -w "%{http_code}" \
+      'http://{{ host }}/task'); \
+    if [ $STATUS_CODE -lt 200 ] || [ $STATUS_CODE -ge 300 ]; then \
+      echo "Status: $STATUS_CODE"; \
+      exit 1; \
+    fi
 
 # Load sample data into temp table in API and new index in Elasticsearch
 ingest-upstream model="image" suffix="init":

load_sample_data.sh

@@ -12,16 +12,28 @@ DB_SERVICE_NAME="${DB_SERVICE_NAME:-db}"
 # Load sample data
 function load_sample_data {
   docker-compose exec -T "$UPSTREAM_DB_SERVICE_NAME" bash -c "psql <<-EOF
-    DELETE FROM $1;
+		DELETE FROM $1;
 		\copy $1 \
 			from './sample_data/sample_$1.csv' \
 			with (FORMAT csv, HEADER true);
 		REFRESH MATERIALIZED VIEW $1_view;
 		EOF"
 }
 
+function verify_loaded_data {
+  COUNT=$(docker-compose exec -T "$UPSTREAM_DB_SERVICE_NAME" bash -c "psql -AXqt <<-EOF
+		SELECT COUNT(*) FROM $1;
+		EOF")
+  if [ "$COUNT" -ne 5000 ]; then
+    echo "Error: table $1 count differs from expected."
+    exit 1
+  fi
+}
+
 load_sample_data "image"
+verify_loaded_data "image"
 load_sample_data "audio"
+verify_loaded_data "audio"
 
 #######
 # API #