Vm reinstate erd

[vrajmohan]

No description provided.

[mitchellhenke]

Suggested change

	stage: .post
	stage: .post
	needs:
	- job: install
	cache:
	- <<: *ruby_cache

[vrajmohan]

I was assuming that the artifacts from the publish_er_diagram job for the last successful main pipeline would be available at a well-known URL, something like https://docs.gitlab.com/ee/ci/jobs/job_artifacts.html#from-a-url. I now realize that this may not work and there is no prior art in our environment for publishing build artifacts to a standard location like an S3 bucket.

Would it be better for me to copy the way we generate the events for the analytics handbook - create a make task in the internal-handbook repo that would clone the identity-idp repo and run bundle exec rails erd?

[aduth]

I was assuming that the artifacts from the publish_er_diagram job for the last successful main pipeline would be available at a well-known URL, something like https://docs.gitlab.com/ee/ci/jobs/job_artifacts.html#from-a-url. I now realize that this may not work and there is no prior art in our environment for publishing build artifacts to a standard location like an S3 bucket.

What issues are you expecting won't allow it to work? I'd have imagined with a Pages configuration we could have a URL like https://lg.pages.production.gitlab.login.gov/identity-idp/idp_erd.svg to serve the latest artifact from main, although it would be access-restricted. Are you finding that that won't work?

I think doing it in a similar way to how analytics events could work too, if that ends up being a simpler path.

[vrajmohan]

I am trying to leverage the build artifacts and avoid issuing commits for redundant information.