Curious why Maesh needs to live outside of the user's namespace? I'm naive when it comes to Kubernetes DNS resolution outside of routing applications, so let me know if I'm missing something fairly obvious.
Would love to use this in our multi-tenant cluster's namespace, and we still could if we need the second namespace, but this will slow down development.
Maesh also ignores its own namespace, and does not create endpoints for services in its namespace.
Maesh ignoring user services definitely doesn't sound like desirable behavior Thanks for clarifying.
Follow up question, couldn't this be solved with labels or selectors? I'm assuming you're saying: Maesh creates its own services for .maesh endpoints. It then uses its own namespace to ignore these, since we know anything in that namespace is a .maesh endpoint.
Couldn't you instead label all these endpoints with something like type=maesh and from there ignore those services? Another method could be to label with with id=<uuid4>, and have maesh keep track of all endpoint IDs. Whenever generating the .maesh endpoint you could check the intersection of the set of IDs against the service you're currently looking at to verify you aren't touching a maesh service.