Due to the workflow being used for the batch-jobs container (i.e. standard WM image that pulls from GitHub at runtime) it likely won't be feasible to incorporate catalog creation into the CI process.
In lieu, the simplest route infrastructure-wise will likely be to manage catalog generation through the API itself. When a catalog is requested, the API could use this workflow:
- If available, reference catalog from API in-memory store
- If not in memory, reference catalog from S3 store (and save in-memory)
- If catalog reference older than 1 hour (?), update S3/memory catalog based on current GitHub catalog
Also, add an endpoint that can be used to force the API to update based on the current GitHub catalog. Could possibly tie this to dev/admin permissions?
Due to the workflow being used for the batch-jobs container (i.e. standard WM image that pulls from GitHub at runtime) it likely won't be feasible to incorporate catalog creation into the CI process.
In lieu, the simplest route infrastructure-wise will likely be to manage catalog generation through the API itself. When a catalog is requested, the API could use this workflow:
Also, add an endpoint that can be used to force the API to update based on the current GitHub catalog. Could possibly tie this to dev/admin permissions?