Graham, If you have a collection of data that should always be sent and consumed together and in-order, why not send it using a single Kafka message? Or is the payload really huge?
- Niek On Sat, Jul 7, 2012 at 10:18 AM, graham sanderson <gra...@vast.com> wrote: > 1) I would like to guarantee that a group of messages are always delivered in > their entirety together (because there is contextual information in messages > which precede other messages). I'm a little confused by the use of the term > "nested message sets" since I don't really see much in the code (though II > don't really know Scala) - perhaps this refers to the fact that you can have > a set of messages within a message set file on disk. Anyway, I was curious > (and I'm using the Java api now, but may move to the Scala later) what I need > to do to guarantee N messages are sent and delivered as a single message set; > is a single ProducerData with a List of messages always sent as a single > message set? does compression need to be turned on? how does this affect > network limits etc. (i.e. does the entire message set have to fit). I'm also > assuming that once I have my message set containing all my messages it will > be discarded in its entirety. > > 2) Related to 1) from the consumer side, can I tell the boundaries of a > message set (perhaps not required for me), but nevertheless I do want to make > sure I receive the entire set in one go (again do I have to set network > limits accordingly). The docs say that the entire message set is always > delivered to the client when compressed, but I'm not sure if it can be > subdivided if not compressed. Note I'm happy to stick with compression if > required. > > 3) So I'm using the ZookeeperConsumerConnector, since I don't want to manage > finding the brokers myself, however I was wondering if there are any plans to > decouple the consumer offset tracking from the former. One of my use cases is > that I'll have a lot of ad-hoc one off consumers that simply read a subset of > data until they die - from looking at ConsoleConsumer, there is currently a > hack to simply delete the zookeeper info after the fact to get around this. > > Thanks, > > Graham.