[
https://issues.apache.org/jira/browse/BEAM-6466?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16747410#comment-16747410
]
Jozef Vilcek commented on BEAM-6466:
------------------------------------
I see checkpoint being finalised in ReadFn processElements after reading from
source enough elements and/or for long enough time
[https://github.com/apache/beam/blob/master/sdks/java/core/src/main/java/org/apache/beam/sdk/io/BoundedReadFromUnboundedSource.java#L203]
Is this guaranteed to be called when all elements downstream are successfully
processed. What are consistency guarantees in face of failures here?
> KafkaIO doesn't commit offsets while being used as bounded source
> -----------------------------------------------------------------
>
> Key: BEAM-6466
> URL: https://issues.apache.org/jira/browse/BEAM-6466
> Project: Beam
> Issue Type: Bug
> Components: io-java-kafka
> Affects Versions: 2.9.0
> Reporter: Alexey Romanenko
> Priority: Major
>
> While using KafkaIO as bounded source (with {{withMaxReadTime()}} or
> {{withMaxNumRecords()}}) it seems doesn't commit offsets all the time.
> See the details in [the
> discussion|https://lists.apache.org/thread.html/bcec8a1fb166029a4adf3f3491c407d49843406020b20f203ec3c2d2@%3Cuser.beam.apache.org%3E]
> on user@list.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
