Fix azure ci being flakey for restarting nginx

Instead of restarting a service in the container, put the proper
configuration in place when generating the image.

Author: mdellweg

.ci/ansible/Containerfile.j2

@@ -5,9 +5,9 @@ FROM {{ ci_base | default(pulp_default_container) }}
 ADD ./{{ item.name }} ./{{ item.name }}
 {% endfor %}
 
-# Install python packages
-# S3 botocore needs to be patched to handle responses from minio during 0-byte uploads
-# Hacking botocore (https://github.com/boto/botocore/pull/1990)
+{% for item in extra_files | default([]) %}
+ADD ./{{ item.origin }} {{ item.destination }}
+{% endfor %}
 
 # This MUST be the ONLY call to pip install in inside the container.
 RUN pip3 install --upgrade pip setuptools wheel && \

.github/workflows/scripts/install.sh

@@ -62,6 +62,9 @@ VARSYAML
 fi
 
 cat >> vars/main.yaml << VARSYAML
+extra_files:
+  - origin: "pulpcore/pulpcore/tests/functional/assets/api_root_rewrite.conf"
+    destination: "/etc/nginx/pulp/api_root_rewrite.conf"
 services:
   - name: pulp
     image: "pulp:ci_build"

.github/workflows/scripts/pre_before_script.sh

@@ -21,6 +21,9 @@ deploy_to_pypi: true
 disabled_redis_runners:
 - s3
 docker_fixtures: true
+extra_files:
+  - origin: pulpcore/pulpcore/tests/functional/assets/api_root_rewrite.conf
+    destination: /etc/nginx/pulp/api_root_rewrite.conf
 flake8: true
 flake8_ignore:
 - ./pulpcore/app/protobuf/*