Yup, this got fixed in 0.10.2 https://issues.apache.org/jira/browse/KAFKA-2000 <https://issues.apache.org/jira/browse/KAFKA-2000>
-James > On Feb 23, 2017, at 11:10 AM, Jeff Widman <j...@netskope.com> wrote: > > The topic deletion only triggers tombstone on brokers >= 0.10.2, correct? I > thought there was an outstanding bug report for this in lower versions... > > On Wed, Feb 22, 2017 at 6:17 PM, Hans Jespersen <h...@confluent.io> wrote: > >> The __consumer_offsets topic should also get a tombstone message as soon as >> a topic is deleted. >> >> -hans >> >> /** >> * Hans Jespersen, Principal Systems Engineer, Confluent Inc. >> * h...@confluent.io (650)924-2670 >> */ >> >> On Wed, Feb 22, 2017 at 5:59 PM, Jun MA <mj.saber1...@gmail.com> wrote: >> >>> Hi Todd, >>> >>> Thank you so much for your reply. I assume that the broker will produce >>> the tombstone to __consumer_offsets topic when the offset expires? I’m >>> curious how broker notices the offset expires? Does it store the offset >>> message in memory and periodically check if someone expires? >>> >>> Thanks, >>> Jun >>> >>>> On Feb 22, 2017, at 4:37 PM, Todd Palino <tpal...@gmail.com> wrote: >>>> >>>> __consumer_offsets is a log-compacted topic, and a NULL body indicates >> a >>>> delete tombstone. So it means to delete the entry that matches the key >>>> (group, topic, partition tuple). >>>> >>>> -Todd >>>> >>>> >>>> >>>> On Wed, Feb 22, 2017 at 3:50 PM, Jun MA <mj.saber1...@gmail.com> >> wrote: >>>> >>>>> Hi guys, >>>>> >>>>> I’m trying to consume from __consumer_offsets topic to get exact >>> committed >>>>> offset of each consumer. Normally I can see messages like: >>>>> >>>>> [eds-els-recopp-jenkins-01-5651,eds-incre-staging-1,0]::[ >>>>> OffsetMetadata[29791925,NO_METADATA],CommitTime >>>>> 1487090167367,ExpirationTime 1487176567367], >>>>> >>>>> which make sense to me. But sometimes I see messages like: >>>>> >>>>> [eds-elssearchindex-curiosity-stg-10892,eds-incre-v2- >>> staging-els,0]::NULL. >>>>> >>>>> Can someone explains what is NULL means here and why a NULL value get >>>>> published to __consumer_offsets? >>>>> >>>>> We’re running kafka 0.9.0.1 and we use org.apache.kafka.common. >>>>> serialization.ByteArrayDeserializer and GroupMetadataManager. >>> OffsetsMessageFormatter >>>>> to parse the message. >>>>> >>>>> Thanks, >>>>> Jun >>>> >>>> >>>> >>>> >>>> -- >>>> *Todd Palino* >>>> Staff Site Reliability Engineer >>>> Data Infrastructure Streaming >>>> >>>> >>>> >>>> linkedin.com/in/toddpalino >>> >>> >>