Warn messages repeatedly logged when polling_frequency is higher than 20.

[ajoliveira]

Using version 3.0.3 with Logstash 5.4 on RedHat Linux, setting polling_frequency => 30, results in the following WARN message being repeatedly logged:

[2017-05-04T22:05:23,818][WARN ][logstash.inputs.sqs ] Aws::SQS::Errors::ServiceError ... retrying SQS request with exponential backoff {:queue=>"SQS_QUEUE_NAME", :sleep_time=>1, :error=>#<Aws::SQS::Errors::InvalidParameterValue: Value 30 for parameter WaitTimeSeconds is invalid. Reason: Must be >= 0 and <= 20, if provided.>}

It appears that the desired value may not actually be getting applied and is reset to 20. Configuration is similar to following:

input {
  sqs {
    queue => "SQS_QUEUE_NAME"
    region => "us-east-1"
    access_key_id => "SOME_KEYID"
    secret_access_key => "SOME_SECRET"
    polling_frequency => 30
  }
}