Internal IP address of host vs service name for local federation

[rmanaem]

In the local_federation setup, we use localhost for the API_QUERY_URL to refer to the federation API running on the host machine, This works fine for the requests made by clicking the submit query on the query tool interface from the browser (i.e., client side) but this is an issue for requests made in the fetch hook cycle that are made from the query tool container inside the docker network for which localhost refers to the query tool container. Not the host running the docker compose stack, therefore, requests to retrieve diagnosis, assessments, and nodes fail.

To resolve this

• [FIX] Implemented "no pipelines" button query-tool#336

See also #5

[surchs]

Hey @rmanaem, agree with the problem, but I think the solution here should be to ensure we're not fetching the information during the build step but in one of the post-nuxt lifecycle hooks / on the client-side. Maybe we can have a brief chat today?

edit: the main reason for not fetching during build is that even if we could successfully resolve the locally hosted node-API (e.g. by doing what you are proposing in #336) we would still freeze the node and instance list to what it was at build time. So if after the build step we add a new instance or a new node (or remove one), this would not currently be reflected (at least that's how I understand it). Instead we should fetch this information from the API when a user navigates to the query tool - and at that point the localhost should also resolve correctly.

[rmanaem]

Let's have a chat.