VNFRQTS CICD Automation

Abstract

There is a problem that we are experiencing when editing requirements. Because the requirements are generating a JSON file to maintain history of the requirements between releases, when users updated different parts of the documents, we are still getting merge conflicts even though they aren't editing the same documentation.

Purpose of this proposal is to create some additional automation to the document generation and to decouple the requirements to remove the possibility of merge conflicts because of requirements being edited. As well as to automate the generation of some of the tables we are manually creating today.

Rationale

Decoupling dependencies are important to optimize workflows plus gives more flexibility for future changes.

Other Options Considered

Other options discussed were to get rid of the JSON, but this JSON is used today to generate reports between the VNFRQTS and VVP projects, specifically to map validation scripts to requirements. These reports would not be able to be generated without the JSON file.

Proposal

Proposal is to upload the JSON artifact into the ONAP Nexus as a RAW artifact in the RAW repository. To make sure this artifact will be up to date, I also propose to create Jenkins Jobs to do daily builds of the VNFRQTS projects, which builds the JSON artifact and upload this artifact to Nexus. This workflow would also be updated for the VVP project to pull down this proposed JSON artifact, use it to generate reports (CSV files) and also upload those CSV files to Nexus as RAW artifacts to be re-ingested by the VNFRQTS project. By re-ingested, this refers to tables in the VNFRQTS project that are generated manually and uploaded to the gerrit repository today, but this workflow would automate this effort as well by having the RST in the VNFRQTS project point to the Nexus artifact.

Below is a diagram explaining this new workflow utilizing the CICD tools.