Re: Correlation id

2013-08-19 Thread Tejas Patil
Multiple produce requests are sent asynchronously over the same socket. Suppose you send 2 requests and get back single response, how do you figure out which one it corresponds to of those 2 requests ? Correlation Id helps here. AFAIK, correlation Id is added to produce requests and broker uses the

Re: Kafka startup/restart process

2013-08-19 Thread Tejas Patil
Not that I am some expert on this subject, but I can see that broker logs indicate the shutdown progress: https://github.com/tejasapatil/kafka/blob/0.8.0-beta1-candidate1/core/src/main/scala/kafka/server/KafkaServer.scala#L165 On Mon, Aug 19, 2013 at 10:19 PM, Vadim Keylis wrote: > Neha. Thanks

Re: LeaderNotAvailableException

2013-08-07 Thread Tejas Patil
What is the proper way to shutdown and > allow partition to be serve by other brokers? What is the process to > recover from this? How to avoid this and still have ability to remove > servers from tier? > > Sent from my iPad > > On Aug 6, 2013, at 10:55 PM, Tejas Patil wrot

Re: LeaderNotAvailableException

2013-08-06 Thread Tejas Patil
I assume that "We had 6 kafka in the tier" means that you had 6 kafka brokers. About the exception that you see: I think that the 3 brokers you took down were having the data for [junit2_analytics_data_log,0] and no other live broker has the data for [junit2_analytics_data_log,0]. You could run t

Re: Random Broker/Partition for Producer

2013-07-31 Thread Tejas Patil
Nope. The property "metadata.broker.list" in producer.properties [0] is just used to fetch the metadata about which broker owns which [topic-partition]. Based on this info, the producer data is routed to the relevant leader broker. By default Kafka uses DefaultPartitioner [1] which is used to rand

Re: reprocessing messages in kafka

2013-07-31 Thread Tejas Patil
As @Milind said, it is possible that a consumer consumes the same message more than once. This happens when there is an unclean shutdown of the consumer and it is not able to commit its latest offset to Zookeeper. When the failed consumer comes up, it would fetch the stale offset from zookeeper thu

Loading kafka in Eclipse IDE

2013-07-01 Thread Tejas Patil
Import" -> "Existing projects into workspace" [0] : https://cwiki.apache.org/KAFKA/developer-setup.html [1] : http://scala-ide.org/ [2] : http://kafka.apache.org/code.html [3] : https://github.com/typesafehub/sbteclipse Thanks, Tejas Patil http://www.linkedin.com/in/tejaspatil1