[
https://issues.apache.org/jira/browse/BEAM-223?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15647844#comment-15647844
]
ASF GitHub Bot commented on BEAM-223:
-------------------------------------
GitHub user rangadi opened a pull request:
https://github.com/apache/incubator-beam/pull/1312
[BEAM-223] Use Avro serializer for Kafka checkpoint mark
This is more portable. It would also make it extendable even across job
updates in future.
+R: @dhalperi, @amitsela
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/rangadi/incubator-beam avro
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/incubator-beam/pull/1312.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 #1312
----
commit f1adb9d118de18b329717889cb48869b44b88281
Author: Raghu Angadi <[email protected]>
Date: 2016-11-08T15:08:32Z
Use Avro serializer for Kafka checkpoint mark.
This is more partable.
----
> KafkaIO: don't use SerializableCoder
> ------------------------------------
>
> Key: BEAM-223
> URL: https://issues.apache.org/jira/browse/BEAM-223
> Project: Beam
> Issue Type: Bug
> Components: sdk-java-extensions
> Reporter: Daniel Halperin
> Assignee: Raghu Angadi
>
> Reuven says:
> {quote}
> I noticed that we're using SerializableCoder for the checkpoint mark in
> KafkaIO. This is generally highly discouraged in streaming pipelines.
> Partially because it's inefficient, but more importantly because Java
> serialization is not guaranteed to be stable. If a user updates their
> pipeline, the new pipeline may not be able to decode the existing checkpoint
> marks; this will either cause exceptions to be thrown, or data loss.
> {quote}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
