Same issue here. I use on-demand instances though, so it's not spot related. I checked logs in Cloudwatch and there is nothing in log groups for the instance. Any other troubleshooting tips? I know before we had an issue where operation was canceled due to resource exhaustion but I updated ec2 size to xlarge which should be overkill for what it does.

There seems to be a bug with pagenation in the scale-down function. Seems occuring with ephemeral runners and a high load.