Since you are using the simple consumer you will need to handle the OffsetOutOfRange Exception in your code. This happens when your consumer queries for an offset which is no longer persisted in Kafka (The logs have been deleted based on the retention policy). Ideally when this happens, the consumer should issue an OffsetRequest to get the latest/earliest offset currently available on the server. You can then use this in your FetchRequest. Whether you use the latest or earliest depends on your application needs. In essence, you are resetting the consumer to catch up with the server.
On May 24, 2013, at 11:29 AM, Sining Ma <[EMAIL PROTECTED]> wrote:
> We are currently using kafka-0.7.1 right now.
> I have two questions:
> 1. We use SimpleConsumer to aggregate messages to log files and there is no zookeeper. Sometimes we can see kafka.common.OffsetOutOfRangeException.
> And this exception happens when we start our consumer program. We do not know the reason why this happens.
> How can I get a valid latest message offset in kafka-0.7.1 when this exception happens?
> 2. Before we start consumer, we call getOffsetsBefore function to get a list of valid offsets (up to maxSize) before the given time.
> How can we interpret this list?
> For example, this function returns an array [offset1, offset2].
> Does this mean from offset1 to offset2 are valid, and offset2 to current offset are valid? We are confused about the meaning of this array.
> -- Regards
> Sining Ma