We are still investigating the situation, but possible issue is in linux kernel with Elastic Network Adapter, we temporary downgraded the kernel, there is no way to be sure of this, as it something on the US production is triggering this specific behaviour. We will continue to monitor the servers and try to react quickly.