Dgraph crash loop on aws

We’re using eksctl… 3 m5ad.xlarge instances in us-east-1, one instance in each availability zone. Trying to get dgraph nodes to be 1 per az.