This incident has been resolved.

All failover infrastructure is live and job service is restored. We will continue to address the underlying cause.

AWS are experiencing broad capacity constraints around our preferred instance type, we are activating a fallback to secondary instance type to address.

Failover capacity is online and job throughput is increasing.

Errors seem to be caused in part by lack of instance capacity in our primary runner instance pool, we are activating a failover instance pool to unblock jobs.