root
/
kafka
mirror of https://github.com/apache/kafka.git
1
0
Fork 0
Code Issues Actions 2 Packages Projects Releases Wiki Activity

MINOR:Upgrade guide updates for KIP-479 (#7550) 

Reviewers: A. Sophie Blee-Goldmann <sophie@confluent.io>, John Roesler <john@confluent.io>, Matthias J. Sax <matthias@confluent.io>
This commit is contained in:
Bill Bejeck 2019-10-21 20:27:23 -04:00 committed by Matthias J. Sax
parent 7bef73d724
commit c7a3d1b52d
1 changed files with 19 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
docs/streams/upgrade-guide.html
View File

@ -77,7 +77,7 @@
<!-- Placeholder KIP-213 -->
<p>
In the 2.4 release, you now can name all operators in a Kafka Streams DSL topology via
<a href="https://cwiki.apache.org/confluence/display/KAFKA/KIP-307%3A+Allow+to+define+custom+processor+names+with+KStreams+DSL">KIP-307</p>.
<a href="https://cwiki.apache.org/confluence/display/KAFKA/KIP-307%3A+Allow+to+define+custom+processor+names+with+KStreams+DSL">KIP-307</a>.
Giving your operators meaningful names makes it easier to understand the topology
description (<code>Topology#describe()#toString()</code>) and
understand the full context of what your Kafka Streams application is doing.
@ -89,7 +89,24 @@
either <code>Joined</code> or the new <code>StreamJoined</code> object.
</p>
<!-- Placeholder KIP-479 -->
<p>
Before the 2.4.0 version of Kafa Streams, users of the DSL could not name the state stores involved in a stream-stream join.
If users changed their topology and added a operator before the
join, the internal names of the state stores would shift, requiring an application reset when redeploying.
In the 2.4.0 release, Kafka Streams adds the <code>StreamJoined</code>
class, which gives users the ability to name the join processor, repartition topic(s) (if a repartition is required),
and the state stores involved in the join. Also, by naming the state stores, the changelog topics
backing the state stores are named as well. It's important to note that naming the stores
<strong>will not</strong> make them queryable via Interactive Queries.
<br/>
Another feature delivered by <code>StreamJoined</code> is that you can now configure the type of state store used in the join.
You can elect to use in-memory stores or custom state stores for a stream-stream join. Note that the provided stores
will not be available for querying via Interactive Queries. With the addition
of <code>StreamJoined</code>, stream-stream join operations
using <code>Joined</code> have been deprecated. Please switch over to stream-stream join methods using the
new overloaded methods. You can get more details from
<a href="https://cwiki.apache.org/confluence/display/KAFKA/KIP-479%3A+Add+StreamJoined+config+object+to+Join">KIP-479</a>.
</p>
<p>
With the introduction of incremental cooperative rebalancing, Streams no longer requires all tasks be revoked at the beginning of a rebalance. Instead, at the completion of the rebalance only those tasks which are to be migrated to another consumer