ci: migrate python image to MCR
Type of change
-
Bug Fix -
CI -
Feature
Does this introduce a change in the core logic?
- [No]
Does this introduce a change in the cloud provider implementation, if so which cloud?
-
AWS -
Azure -
GCP -
IBM
Updates description?
- Updated Azure Dockerfile python image to
python:3.8-slim
from Microsoft Container Registry - Copied DAG.Dockerfile to deployments/scripts/azure location and updated
FROM
in new Dockerfile to usepython:3.8-slim
image from Microsoft Container Registry (this is to avoid other cloud providers using MCR) - Replaced
apk
package manager in both Dockerfiles toapt-get
package manager which is available in slim tagged images - Updated condition in
override-stages.yml
to include Azure related jobs back in the pipeline following the changes made in CI-CD Pipelines repo for Azure deployments in general with this MR - Updated
azure_register_dag
job in pipeline to send valid workflow name to workflow API and not hardly depend on full branch name (this is due to workflow API having validation for the lenght in workflow name to be max 64 characters which can easily be exceeded by the pattern used in VDS Converter pipeline to construct workflow name) - Added more logs when registering dag to debug errors easier in the future
- ADO work item: Product Backlog Item 9649: Container Security for VDS Converter images
Testing
-
azure_build_dag job is pulling python:3.8-slim image from MCR for DAG task image
-
azure_build_dag job is pulling python:3.8-slim image from MCR for DAG load image
-
azure_register_dag job has run successfully in the pipeline
-
azure_copy_dag job has run successfully in the pipeline
-
Additional testing will be done on Azure side with build pipeline against m12-master branch
Edited by Marija Dukic