Does anyone know a good way to observe & measure the impact of this? I have a small fleet of Linux servers behind NLBs with cross-AZ load balancing turned on, and they have some significant collection of OS-level metrics via the Prometheus node_exporter.

Would incidents of this be captured by the "node_netstat_TcpExt_TCPAbortOnData" metric ("connections reset due to unexpected data")?