Hi Stig, I don't have this kind of errors normally. It just occured at the rolling restart of kafka.
Also the __consumer_offsets configuration is: Topic:__consumer_offsets PartitionCount:50 ReplicationFactor:3 Configs:segment.bytes=104857600,cleanup.policy=compact,compression.type=producer Topic: __consumer_offsets Partition: 0 Leader: 0 Replicas: 2,1,0 Isr: 0,1 ... The fact that the replication factor is 3 even if there is only two broker can cause an issue ? On Mon, Aug 21, 2017 at 6:51 PM, Stig Rohde Døssing <s...@apache.org> wrote: > The spout will reemit some messages if it fails to commit offsets to > Kafka. Are these CommitFailedExceptions occuring in your logs normally? > > Also since the spout stores offsets in Kafka, you may want to check the > replication factor on that topic by running `./kafka-topics.sh --zookeeper > localhost:2181 --describe --topic __consumer_offsets` in one of your Kafka > /bin directories. > > 2017-08-21 17:17 GMT+02:00 Elyahou Ittah <elyaho...@fiverr.com>: > >> The config is the default one, I just set the bootstrap server. >> >> Kafka version is 0.11 >> >> Storm-kafka-client is 1.1.0 >> >> On Mon, Aug 21, 2017 at 5:48 PM, Stig Rohde Døssing <s...@apache.org> >> wrote: >> >>> Hi Elyahou, >>> >>> Could you post your spout configuration, Kafka version and >>> storm-kafka-client version? The logs imply that your spouts are not polling >>> often enough. >>> >>> 2017-08-21 <20%2017%2008%2021> 9:44 GMT+02:00 Elyahou Ittah < >>> elyaho...@fiverr.com>: >>> >>>> I noticed that storm kafka spout reconsume all kafka message after a >>>> rolling restart of kafka cluster. >>>> >>>> This issue occured only with kafkaSpout consumer and not for my other >>>> consumers (ruby based using the kafka consumer API like kafkaSpout) >>>> >>>> Attached logs of the spout. >>>> >>>> Do you know what can cause this kind of behavior ? >>>> >>> >>> >> >