Can anyone please check this one? Thanks Achintya
-----Original Message----- From: Ghosh, Achintya (Contractor) Sent: Monday, August 08, 2016 9:44 AM To: us...@kafka.apache.org Cc: dev@kafka.apache.org Subject: RE: Kafka consumer getting duplicate message Thank you , Ewen for your response. Actually we are using 1.0.0.M2 Spring Kafka release that uses Kafka 0.9 release. Yes, we see a lot of duplicates and here is our producer and consumer settings in application. We don't see any duplicacy at Producer end I mean if we send 1000 messages to a particular Topic we receive exactly (sometimes less) 1000 messages. But when we consume the message at Consumer level we see a lot of messages with same offset value and same partition , so please let us know what tweaking is needed to avaoid the duplicacy. We have three types of Topics and each topic has 3 replication factors and 10 partitions. Producer Configuration: bootstrap.producer.servers=provisioningservices-aq-dev.g.comcast.net:80 acks=1 retries=3 batch.size=16384 linger.ms=5 buffer.memory=33554432 request.timeout.ms=60000 timeout.ms=60000 key.serializer=org.apache.kafka.common.serialization.StringSerializer value.serializer=com.comcast.provisioning.provisioning_.kafka.CustomMessageSer Consumer Configuration: bootstrap.consumer.servers=provisioningservices-aqr-dev.g.comcast.net:80 group.id=ps-consumer-group enable.auto.commit=false auto.commit.interval.ms=100 session.timeout.ms=15000 key.deserializer=org.apache.kafka.common.serialization.StringDeserializer value.deserializer=com.comcast.provisioning.provisioning_.kafka.CustomMessageDeSer factory.getContainerProperties().setSyncCommits(true); factory.setConcurrency(5); Thanks Achintya -----Original Message----- From: Ewen Cheslack-Postava [mailto:e...@confluent.io] Sent: Saturday, August 06, 2016 1:45 AM To: us...@kafka.apache.org Cc: dev@kafka.apache.org Subject: Re: Kafka consumer getting duplicate message Achintya, 1.0.0.M2 is not an official release, so this version number is not particularly meaningful to people on this list. What platform/distribution are you using and how does this map to actual Apache Kafka releases? In general, it is not possible for any system to guarantee exactly once semantics because those semantics rely on the source and destination systems coordinating -- the source provides some sort of retry semantics, and the destination system needs to do some sort of deduplication or similar to only "deliver" the data one time. That said, duplicates should usually only be generated in the face of failures. If you're seeing a lot of duplicates, that probably means shutdown/failover is not being handled correctly. If you can provide more info about your setup, we might be able to suggest tweaks that will avoid these situations. -Ewen On Fri, Aug 5, 2016 at 8:15 AM, Ghosh, Achintya (Contractor) < achintya_gh...@comcast.com> wrote: > Hi there, > > We are using Kafka 1.0.0.M2 with Spring and we see a lot of duplicate > message is getting received by the Listener onMessage() method . > We configured : > > enable.auto.commit=false > session.timeout.ms=15000 > factory.getContainerProperties().setSyncCommits(true); > factory.setConcurrency(5); > > So what could be the reason to get the duplicate messages? > > Thanks > Achintya > -- Thanks, Ewen
