XComp commented on code in PR #22546:
URL: https://github.com/apache/flink/pull/22546#discussion_r1246344004
##########
flink-end-to-end-tests/flink-confluent-schema-registry/pom.xml:
##########
@@ -53,6 +53,11 @@ under the License.
<artifactId>flink-connector-kafka</artifactId>
<version>3.0.0-1.17</version>
</dependency>
+ <dependency>
+ <groupId>org.apache.flink</groupId>
+ <artifactId>flink-shaded-guava</artifactId>
+ <version>30.1.1-jre-16.1</version>
Review Comment:
I agree - we should add a comment here.
##########
pom.xml:
##########
@@ -353,13 +353,13 @@ under the License.
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-shaded-netty</artifactId>
-
<version>4.1.82.Final-${flink.shaded.version}</version>
+
<version>4.1.91.Final-${flink.shaded.version}</version>
Review Comment:
We initially thought of doing the flink-shaded update early in the release
to have enough CI cycles to test the functionality. With the delays we're
getting close to the 1.18 feature freeze which contradicts the initial idea of
the timing. I discussed this with @zentol: Based on past experience, netty was
the source of instabilities. So, if we want to play it safe we could upgrade
all the shaded dependencies except for netty and do the netty upgrade in 1.19
early on.
But on the other note: With 2 weeks until the feature freeze (which might be
postponed anyway) and additional time for bug fixing and release testing, we
still might have enough CI cycles. What's your opinion on that one?
I don't see a reason to not do the upgrade more conservatively (i.e. waiting
with the netty upgrade and doing it in a separate ticket). Chesnay things that
there is still enough time.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
