This can be caused by either log corruption or a bug in kafka that uses an incorrect offset. Could you use the DumpLogSegment tool to see if offset 2983409308 for topic Memcache2Sql partition 0 at broker 1 is valid?
Thanks, Jun On Tue, Jul 10, 2012 at 5:24 AM, <zlai_2...@sina.com> wrote: > Hi,all > There is an exception when consumer get data from one broker. And > can not get new data from this broker. But there are some data on this > broker that are not consumed. The exception is > like:[kafka.consumer.FetcherRunnable$$anonfun$run$3.apply(FetcherRunnable.scala:91)]: > FetchRunnable-0 kafka.consumer.FetcherRunnable - error in FetcherRunnable > for Memcache2Sql:1-0: fetched offset = 2983409308: consumed offset = > 2983409308kafka.common.InvalidMessageSizeException: invalid message size: > -1592784872 only received bytes: 307196 at 2983409308( possible causes (1) > a single message larger than the fetch size; (2) log corruption ) at > kafka.message.ByteBufferMessageSet$$anon$1.makeNextOuter(ByteBufferMessageSet.scala:103) > at > kafka.message.ByteBufferMessageSet$$anon$1.makeNext(ByteBufferMessageSet.scala:138) > at > kafka.message.ByteBufferMessageSet$$anon$1.makeNext(ByteBufferMessageSet.scala:82) > at > kafka.utils.IteratorTemplate.maybeComputeNext(IteratorTemplate.scala:59) > at > kafka.utils.IteratorTemplate.hasNext(IteratorTemplate.scala:51)...... > > > And the kafka.tools.ConsumerOffsetChecker result > is:MemcacheProducerManager,Memcache2Sql,1-0 > (Group,Topic,BrokerId-PartitionId) Owner = null Consumer offset > = 2983409308 = 2,983,409,308 (2.78G) Log size = > 2984795244 = 2,984,795,244 (2.78G) Consumer lag = > 1385936 = 1,385,936 (0.00G) > > What is the reason? How can I resolve it?Thanks.
