I configured a cluster of 4 API Gateway (2 of them are in DB replica and the other 2 are processing nodes). With this kind of configuration, when I restarted the Master APIGW, I cannot connect to the Policy Manager using the IP address (or hostname) of the others APIGW. Instead, if I simply stop the ssg service I can connect to the Policy Manager of the others APIGW.
I was wondering if that "behavior" is correct or I was wrong in some steps during cluster configuration.
Have a look at idea add real loadbalancing features as I believe this explains, what you are seeing.
so you are telling me that actually I correctly configured the cluster of API Gateway and that the "behavior" I described is a precise choice of product design? Did I rightly understand?
If I understand correct, what you are describing. Yes.