Adds project setup and initial Prometheus and OpenSearch tests

[markgoddard]

No description provided.

[Alex-Welsh]

I approve of the code, but I'm also wondering what your thoughts are with regards to versioning?

We could go the standard stackhpc/release branch route, which would make integration with release train easy. I think I'd also rather pin to branches in SKC than commits, to save the effort of changing it constantly.

[markgoddard]

I approve of the code, but I'm also wondering what your thoughts are with regards to versioning?

We could go the standard stackhpc/release branch route, which would make integration with release train easy. I think I'd also rather pin to branches in SKC than commits, to save the effort of changing it constantly.

It's a good question. This is really at quite an early stage, but I wanted to get something started that can evolve. I was thinking more along the lines of a single branch with tags, but perhaps we would end up needing to diverge as things change over time. I was trying to keep the tests fairly generic, and pass in various things through environment variables. Potentially we could pass in the release name and encode any logic in the test suite about which tests need to run. Would that be easier to maintain than multiple branches?

[Alex-Welsh]

I approve of the code, but I'm also wondering what your thoughts are with regards to versioning?
We could go the standard stackhpc/release branch route, which would make integration with release train easy. I think I'd also rather pin to branches in SKC than commits, to save the effort of changing it constantly.

It's a good question. This is really at quite an early stage, but I wanted to get something started that can evolve. I was thinking more along the lines of a single branch with tags, but perhaps we would end up needing to diverge as things change over time. I was trying to keep the tests fairly generic, and pass in various things through environment variables. Potentially we could pass in the release name and encode any logic in the test suite about which tests need to run. Would that be easier to maintain than multiple branches?

Keeping the logic inside the test suite sounds good. Saves a lot of complexity vs managing multiple branches. Easy to extend as well since I assume at some point we will have environment-specific options e.g. OVS/OVN, TLS, OS