Thanks, I did some stress-testing and infra is scalable enough but I forgot about the AWS quotas, my bad. Quota increase requested and servers are killed off so hopefully "soon" the issue will go away.