gateway - issue found after long operation in production environment

It takes a very long time (from 1 to 4 hours) for the http-gateway to display and expose the CXF REST endpoints after deploying a new version in fabric (version create, profile import, container upgrade etc).

However, when trying to navigate through Fabric -> Services -> APIs on Howtio web console, those endpoints had been displayed correctly already for a while.

Tried to check the fabric registry folder "/fabric/registry/clusters/apis" using the command:
zk:list -r /fabric/registry/clusters/apis
There were a lot of entries (over 36620 entries) and it looks like that some of old endpoints were never cleaned up. And some entries were strange as they had extra "$" character and "/" character missing, like:
/fabric/registry/clusters/apis/rest/$NavigationResource1364924834
I'd rather expect something like:
/fabric/registry/clusters/apis/rest/NavigationResource/1364924834

As recommended in this KCS article:
https://access.redhat.com/solutions/3539351
The property "immediateUpdate" had been set to "true" on "gateway-http" profile "io.fabric8.gateway.http.mapping-apis" PID.