Open amrc-benmorrow opened 8 months ago
It probably makes more sense to store these in the Config Store now?
Yes, I agree schemas should end up in the ConfigDB. However we need to master that information somewhere, and I'm not sure stuffing them all into the ACS repo is a good idea. ACS works perfectly well with a completely different set of schemas, if someone wants to do that.
Maybe we leave them in their own repo, and rather than mirroring we have that repo build another image like the edge-helm image which pushes the preset schemas into the ConfigDB? We could also let people supply their own service-setup images to preload their own schemas or whatever else.
For private schemas a service pulling from AMRC github (or other corporate version control) and updating the ConfigDB would be useful. We might be able to get Flux to help here?
There is quite a lot to be said for pulling the schemas repo on-prem, rather than contacting public Github directly. This would need these actions:
schemas
repo.clusters.js
in service-setup. This needs moving into a new file and the result passing into bothclusters.js
andmanager.js
so the Manager can be informed of the repo UUID.~We need to decide soon if this is practical for v3.0.0.~