Resolved around July 16 at 5:21 AM PDT. Several applications behind CloudFront went down simultaneously across multiple AWS accounts and regions, including Sydney, the UK, Europe, and the US. The applications and databases were healthy when accessed through a VPN that bypassed CloudFront, so the issue appeared to be related to CloudFront. Is anyone else seeing this, particularly with CloudFront VPC Origins?
4 Answers
Services gradually started recovering as AWS rolled out a fix. The incident appeared to be global rather than isolated to one AWS region, and the affected distributions came back online afterward.
It looks like the problem is limited to CloudFront distributions using VPC Origins. Standard CloudFront origins and static sites appear to be working normally. AWS reported increased 5xx errors for customers using VPC Origins connectivity.
We saw the same issue in multiple regions and accounts, including eu-west-1, eu-central-1, us-east-1, and ap-southeast-2. Private ALBs and EKS services connected through VPC Origins were unreachable, while the underlying applications were healthy.
A temporary workaround was to create a public ALB and point the CloudFront distribution directly at it instead of using a VPC Origin. That restored traffic for us, although it may not be suitable for every setup.
Confirmed—only VPC Origins seem to be affected. Distributions using other origin types are still responding normally.