[
https://issues.apache.org/jira/browse/NIFI-5267?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16788057#comment-16788057
]
Sandish Kumar HN commented on NIFI-5267:
----------------------------------------
[~jasperknulst] Yeah, which one would take into consideration at the time of
replaying, we are trying to keep both initial kafka.offset and kafka.timestamp
in the consumer realm.
> Add Kafka record timestamp to flowfile attributes
> -------------------------------------------------
>
> Key: NIFI-5267
> URL: https://issues.apache.org/jira/browse/NIFI-5267
> Project: Apache NiFi
> Issue Type: Improvement
> Components: Core Framework
> Affects Versions: 1.5.0, 1.6.0
> Reporter: Jasper Knulst
> Assignee: Sandish Kumar HN
> Priority: Minor
> Labels: newbie
> Fix For: 2.0.0
>
>
> The ConsumeKafkaRecord and ConsumeKafka processors (0_10, 0_11 and 1_0) can
> yield 1 flowfile holding many Kafka records. For ConsumeKafka this is
> optional (using demarcator).
> Currently the resulting flowfile already gets an attribute 'kafka.offset'
> which indicates the starting offset (lowest) of any Kafka record within that
> bundle.
> It would be valuable to also have a 'kafka.timestamp' attribute there (also
> only related to the first record of that bundle) to be able to relate all the
> records in the flowfile to the kafka timestamp and be able to replay some
> kafka records based on this timestamp (feature in Kafka > 0.9 where replay by
> offset and by timestamp is now a possibility)
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
