[ https://issues.apache.org/jira/browse/BEAM-2534?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16067106#comment-16067106 ]
ASF GitHub Bot commented on BEAM-2534: -------------------------------------- GitHub user rangadi opened a pull request: https://github.com/apache/beam/pull/3461 [BEAM-2534] Handle offset gaps in Kafka messages. KafkaIO logged a warning when there is a gap in offstes for messages. Kafka also supports log compaction for topics some of the messages are deleted leading to gaps in offsets. Log compaction removes earlier messages for key. This PR removes the log and accounts for offset gaps in backlog estimate. You can merge this pull request into a Git repository by running: $ git pull https://github.com/rangadi/incubator-beam handle_offset_gaps Alternatively you can review and apply these changes as the patch at: https://github.com/apache/beam/pull/3461.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #3461 ---- commit 3874bbaa6c88ace25adc132a9f785fb8ace4b68a Author: Raghu Angadi <rang...@google.com> Date: 2017-06-28T19:07:06Z Handle offset gaps in Kafka messages. KafkaIO logged a warning when there is a gap in offstes for messages. Kafka also support 'KV' store style topics where some of the messages are deleted leading gaps in offsets. This PR removes the log and accounts for offset gaps in backlog estimate. ---- > KafkaIO should allow gaps in message offsets > -------------------------------------------- > > Key: BEAM-2534 > URL: https://issues.apache.org/jira/browse/BEAM-2534 > Project: Beam > Issue Type: Bug > Components: sdk-java-extensions > Affects Versions: 2.0.0 > Reporter: Raghu Angadi > Assignee: Raghu Angadi > Priority: Minor > Fix For: 2.1.0 > > > KafkaIO reader logs a warning when it notices gaps in offsets for messages. > While such gaps are not expected for normal Kafka topics, there could be gaps > when log compaction is enabled (which deletes older messages for a key). > This warning log is not very useful. Also we should take such gaps while > estimating backlog. -- This message was sent by Atlassian JIRA (v6.4.14#64029)