chia7712 commented on PR #17861:
URL: https://github.com/apache/kafka/pull/17861#issuecomment-2547534981
> Perhaps we could do just one of them to keep maintenance cost at a
reasonable place - the JUnit tests seem easier to configure and debug.
file https://issues.apache.org/jira/brow
ijuma commented on PR #17861:
URL: https://github.com/apache/kafka/pull/17861#issuecomment-2547515965
> Running the clients/streams tests with Java 11 (both system and JUnit)
would be a plus.
Perhaps we could do just one of them to keep maintenance cost at a
reasonable place - the JU
ijuma commented on PR #17861:
URL: https://github.com/apache/kafka/pull/17861#issuecomment-2547512140
Compilation isn't a worry, the JDK does a great job of cross-compiling.
Also, we can use the same version in our tests and the release. The more
interesting question is runtime behavior - t
chia7712 commented on PR #17861:
URL: https://github.com/apache/kafka/pull/17861#issuecomment-2504723014
Maybe we could add a CI job to run clients:check and streams:check using JDK
11? This would not only help us avoid concerns about Java 11 byte-code, but
also be a straightforward task to
mjsax commented on PR #17861:
URL: https://github.com/apache/kafka/pull/17861#issuecomment-2504594900
Yes, it's simpler, but I am still wondering if a "dual setup" would be
better? -- I am not sure if we want to trust JDK17 to produce correct Java-11
byte-code, and thus, it might be better
chia7712 commented on PR #17861:
URL: https://github.com/apache/kafka/pull/17861#issuecomment-2492834691
> Is this PR actually correct?
It seems simpler to use a single JDK version for all E2E tests, so we're
unifying both Vagrant and Dockerfile to use JDK 17 for running both server a
mjsax commented on PR #17861:
URL: https://github.com/apache/kafka/pull/17861#issuecomment-2492174471
The minimum Java version for clients is 11, not 17. 17 is only the minimum
version for brokers.
Is this PR actually correct?
--
This is an automated message from the Apache Git Ser
jolshan commented on PR #17861:
URL: https://github.com/apache/kafka/pull/17861#issuecomment-2489661240
Thanks for closing on this 🚀
--
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
chia7712 merged PR #17861:
URL: https://github.com/apache/kafka/pull/17861
--
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: jira-unsubscr...@kafka.apache
dongnuo123 commented on PR #17861:
URL: https://github.com/apache/kafka/pull/17861#issuecomment-2486587851
A result of testing is
[here](https://confluent-open-source-kafka-branch-builder-system-test-results.s3-us-west-2.amazonaws.com/trunk/2024-11-18--001.f95a8430-046e-4832-b8f6-cf53568f805
dongnuo123 opened a new pull request, #17861:
URL: https://github.com/apache/kafka/pull/17861
We have dropped JDK 8 and now use JDK 11 as the minimum version. This PR
updates the `jdk_major` and `jdk_full` versions used by Vagrant to JDK 17. Also
for zk compatibility with jdk 17, the patch
chia7712 commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2456622247
@cmccabe Are there any concerns with replacing zk 3.4.x jar with zk 3.5 jar
for Kafka versions 2.1, 2.2, and 2.3? With this change, we can simplify the
software versions used in E2E test
chia7712 commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2452731618
> Those old Kafka versions don't support JDK17. You need to run them with
jdk8.
Pardon me, is there any other issue, aside from
[ZOOKEEPER-3779](https://issues.apache.org/jira/bro
cmccabe commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2452655129
> If the ZooKeeper version in the list is the one used by the Kafka release,
it might be slightly different from the version in the source code.
That's fine. The ZooKeeper server 3.
chia7712 commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2452604397
> implement support for multiple JDKs. Probably 8 and 17. Rationale: Kafka
2.1 requires jdk8, and Kafka 4.0 will require jdk17 (for the server, at least).
I prefer using a single J
chia7712 commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2452589480
> So basically I'm proposing:
If the ZooKeeper version in the list is the one used by the Kafka release,
it might be slightly different from the version in the source code.
cmccabe commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2452562898
So basically I'm proposing:
|version|support |zk version|java version|
|---|-|--||
| pre-2.0 | none| n/a | n/a|
|
cmccabe commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2452555088
OK here is what I'm thinking:
* Do not test anything earlier than Kafka 2.1, since
[KIP-896](https://cwiki.apache.org/confluence/display/KAFKA/KIP-896%3A+Remove+old+client+protocol+
cmccabe commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2452524083
I took a look at
[KIP-896](https://cwiki.apache.org/confluence/display/KAFKA/KIP-896%3A+Remove+old+client+protocol+API+versions+in+Kafka+4.0)
and it does specify that we should support Ka
cmccabe commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2452482689
> Do you mean that we don't need to test the new client (4.0) producing and
consuming from older brokers (2.1 to 2.3)?
This is a fair question. 2.3 was released in mid-2019, so more
kirktrue commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2452329526
> Looks like we should just convert the streams, connect, and clients tests
to not use ZK 👍
For clients, I have this [massive
PR](https://github.com/apache/kafka/pull/17581) opene
chia7712 commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2451281344
> Please remember that the first production KRaft version with 3.3, and we
do not support upgrades from versions earlier than that. (We will test older
clients, of course, but clients do
bbejeck commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2451044794
> several PRs landed recently:
>
> KAFKA-17609:[3/4]Convert system tests to kraft part 3 (#17327)
> KAFKA-17609:[4/4]Convert system tests to kraft part 4 (#17328)
>KAFKA
cmccabe commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2451017255
> Looks like we should just convert the streams, connect, and clients tests
to not use ZK 👍
several PRs landed recently:
```
KAFKA-17609:[3/4]Convert system tests to kraft
jolshan commented on code in PR #17576:
URL: https://github.com/apache/kafka/pull/17576#discussion_r1825279500
##
tests/setup.py:
##
@@ -51,7 +51,7 @@ def run_tests(self):
license="apache2.0",
packages=find_packages(),
include_package_data=True,
- insta
jolshan commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2451016049
Looks like we should just convert the streams, connect, and clients tests to
not use ZK 👍
--
This is an automated message from the Apache Git Service.
To respond to the message, please
jolshan commented on code in PR #17576:
URL: https://github.com/apache/kafka/pull/17576#discussion_r1825279500
##
tests/setup.py:
##
@@ -51,7 +51,7 @@ def run_tests(self):
license="apache2.0",
packages=find_packages(),
include_package_data=True,
- insta
cmccabe commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2451011828
We are removing zookeeper, so let's not spend time on getting it to run.
Let's just move to jdk11 in the tests for now, and once the removal is fully
done we can go to jdk17 (assumi
chia7712 commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2448053676
> We'll file a PR later,
see #17625
--
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 t
chia7712 commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2442688916
@jolshan We'll file a PR later, and I think we can align the JDK folder
structure with Vagrant by using /opt/jdk/$JDK_MAJOR. This means there will be
two JDK folders: `/opt/jdk/17` and `
jolshan commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2442653084
Awesome! Thanks @chia7712! Is there a way to set a jdk for specific tests? 🤔
Should we somehow try to include in this PR?
--
This is an automated message from the Apache Git Service
chia7712 commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2442661369
> Awesome! Thanks @chia7712! Is there a way to set a jdk for specific tests?
🤔
Should we somehow try to include in this PR?
Maybe we should pre-download different JDK versions a
chia7712 commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2442617253
> any ideas about the ZK issues I saw in 2.0-2.3? I can try to do some more
debugging as well.
It seems Zookeeper client 3.4.x can't run under JDK 17 ...
https://issues.apache.org
jolshan commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2442064649
@chia7712 any ideas about the ZK issues I saw in 2.0-2.3? I can try to do
some more debugging as well.
--
This is an automated message from the Apache Git Service.
To respond to the mes
chia7712 commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2440164738
> sanity_checks.test_performance_services still runs with 0.8.2.2 and fails
I notice another issue and https://issues.apache.org/jira/browse/KAFKA-17883
will fix it
--
This is a
chia7712 commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2439200370
> sanity_checks.test_performance_services still runs with 0.8.2.2 and fails
open https://issues.apache.org/jira/browse/KAFKA-17879 for it
--
This is an automated message from the
jolshan commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2436549619
Digging deeper, looks like an issue with ZK -- the brokers see
`kafka.zookeeper.ZooKeeperClientTimeoutException: Timed out waiting for
connection while in state: CONNECTING`
and on
jolshan commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2436543307
@chia7712 you should be able to see this one:
http://confluent-open-source-kafka-branch-builder-system-test-results.s3-us-west-2.amazonaws.com/trunk/2024-10-23--001.98fe335d-7619-47bc-a9b8
jolshan commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2434331106
Looking at the test results, seems there were issues with tests up to 2.3.1.
@chia7712 are we planning to drop up to that version?
For comparison, with java 11, I only saw issues wi
chia7712 commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-2434805610
> Looking at the test results, seems there were issues with tests up to
2.3.1. @chia7712 are we planning to drop up to that version?
Sorry, I can't access the Confluent link. Could
dongnuo123 commented on PR #17576:
URL: https://github.com/apache/kafka/pull/17576#issuecomment-243486
https://semaphore.ci.confluent.io/jobs/98fe335d-7619-47bc-a9b8-4c8449ec745a
Triggered the task with java 17
--
This is an automated message from the Apache Git Service.
To respond
41 matches
Mail list logo
