We've had chef run failures during a large scale-up event in our autoscaling group. Each instance of the autoscaling group that comes online adds itself to the same DNS CNAME in route53; however, because of the large scale-up event, multiple instances attempt to update the record exceeding the AWS rate limit. #change_record has a rescue to prevent a fail_on_error when this is false, but it only rescues Aws::Route53::Errors::ServiceError. The AWS ruby sdk documents the rate throttling exception as Aws::Route53::Types::ThrottlingException. Thus, the chef run fails even when fail_on_error is set to false (default value) for the route53_record resource.
#change_record should rescue Aws::Route53::Types::ThrottlingException since the last record update will be able to capture all of the records that need to be added.
We've had chef run failures during a large scale-up event in our autoscaling group. Each instance of the autoscaling group that comes online adds itself to the same DNS CNAME in route53; however, because of the large scale-up event, multiple instances attempt to update the record exceeding the AWS rate limit.
#change_recordhas a rescue to prevent afail_on_errorwhen this isfalse, but it only rescues Aws::Route53::Errors::ServiceError. The AWS ruby sdk documents the rate throttling exception as Aws::Route53::Types::ThrottlingException. Thus, the chef run fails even whenfail_on_erroris set to false (default value) for theroute53_recordresource.#change_recordshould rescueAws::Route53::Types::ThrottlingExceptionsince the last record update will be able to capture all of the records that need to be added.