Broken pipe errors are happening when Kubernetes is restarting the container, so you need to look deeper into logs to find the culprit, but seems like you identified it correctly. A port never becoming ACTIVE is obviously a Neutron issue. We reported and got fixed several problems like that in the course of Rocky-Train cycles. What OpenStack version do you use there?
Broken pipe errors are happening when Kubernetes is restarting the container, so you need to look deeper into logs to find the culprit, but seems like you identified it correctly. A port never becoming ACTIVE is obviously a Neutron issue. We reported and got fixed several problems like that in the course of Rocky-Train cycles. What OpenStack version do you use there?