I run:

./aws-windsock-kafka.sh cloud-setup "name=kafka shotover=standard"
./aws-windsock-kafka.sh cloud-run --bench-length-seconds 60 "name=kafka shotover=standard"

The issue occurs when the bench run reaches name=kafka,shotover=standard,size=1B,topology=cluster1:

All throughput stops and then the following bench never starts or completes.

I have ssh'd into the problem docker instance and am running htop, the ssh session dies and I cannot reconnect: (this screenshot did not sort by memory but I imagine shotover is taking up the rest of the RAM)

Another time it died when cleaning up after the previous bench: