In our environment we usually do a canary mechanism for our microservices. we implement 1 service pointed to 2 different deployment config pods. But we found an issue that after we remove one of the DC from the service, we still able to see traffic exist on the removed DC pods. And it happend like forever, the traffic not distributed to the enabled DC even we wait for couple of hours.

The traffic distributed to the enabled DC pods after we restart the pods that call the service. We use openshift 3.6 in our environment.

I attach the flow of the service in this case also.

Badly need help for this issue.