Understand the story for multiple namespaces and a single or multiple kore deployment about keda

[ahmelsayed]

Do you mean that Kore has to be in the same namespace as the deployment for custom metrics to work?

I was thinking of the following scenarios:

• I want to deploy kore to manage ALL ScaledObjects in all namespaces
• I want kore to only manage the namespace it's deployed in.
• I want to deploy kore to manage 1 or more namespaces

For example I could be:

• an operator (or a dev with a cluster) and just want all my users to be able to deploy functions.
• a user for a cluster. The cluster operator created a namespace for me and I have full permissions just on that namespace.
• An operator with a very large cluster, and I want to partition my cluster to 2 or more kore instances.

[yaron2]

We should be pursuing option 1, IMHO

[jeffhollan]

This appears to work fine today. I always deploy keda in it's own namespace and my functions in default and everything is happy