The release notes for 2.0 linked to from the blog at

<https://nightlies.apache.org/flink/flink-docs-release-2.0/release-notes/flink-2.0/>

are 404 - is this because they are still being published? /cc @weijie guo

Hi, would like to raise to your attention that with 1.20.1 there was a major bug introduced by https://issues.apache.org/jira/browse/FLINK-35886 which prevents from using custom metrics in watermark generator due to its leading to StackOverflowError. This is due a change in code from:

watermarkGenerator =
                emitProgressiveWatermarks
                        ? watermarkStrategy.createWatermarkGenerator(this::getMetricGroup)
                        : new NoWatermarksGenerator<>();

to

watermarkGenerator =
                emitProgressiveWatermarks
                        ? watermarkStrategy.createWatermarkGenerator(
                                new WatermarkGeneratorSupplier.Context() {
                                    @Override
                                    public MetricGroup getMetricGroup() {
                                        return this.getMetricGroup();
                                    }

                                    @Override
                                    public RelativeClock getInputActivityClock() {
                                        return inputActivityClock;
                                    }
                                })
                        : new NoWatermarksGenerator<>();

Where

this.getMetricGroup

refereed to parent class instance before and now refers to anonymous class instance which leads to infinite recursion. https://issues.apache.org/jira/browse/FLINK-37545

Hello Everyone 👋 , Have anyone used two different Kafka Stream as a soruce in a single Flink application? If so could someone please share any cons that you found? In my use case, I have I want to process events from two different region Kafka cluster, the application itself is critical in terms of latency, so I found three solutions, 1. Use Flink's union stream to use two different Kafka source and process after union-ing it 2. Replicate the events from one region (say X) to other region (say Y) and use Y as source in Flink application 3. Deploy two Flink application so one Flink application can process one region data and the other Flink application can process the other region data - This is hard to maintain for sure For #2 and #3, I can work out pros/cons, but I am curious to know are there any cons/constraints/limitations that I need to be aware of if I choose #1. The doc didn't mention anything explicitly as a drawback, so wondering can some one can tell their experience with #1. Obviously I want to process the events without any latency, time to time both the source might ingest events at various rates.

Hai, When does flink-operator 1.10 submits the flink sessionjob resource? I'm using flink version 1.20 and all batch session job resources are not cleaned up and keep on resubmitting to the Flink Deployments and running multiple times.

Kafka Updates in Flink Connector Kafka Thanks for merging my Kafka 3.9 PR @Arvid Heise. I was wondering about the release versions. The main branch is currently on version

4.0-SNAPSHOT

, but I think if we are moving to the next major release for the Kafka connector it makes sense to also move to the Kafka 4.0 client libraries as they drop support for Kafka <2.1 (a breaking change). I have a PR open for that update. I which case do we want to do a Flink Connector Kafka version 3.5 release which maintains the support for Kafka 2.0 and lower and also supports flink 2.0.0?

Hi everyone. I am new to Flink and open source contributions. I came across a minor issue in the Flink documentation for which I want to open a Jira ticket and find a committer who can review the possible fixes and apply the patch. Is it the right way to go about this? Also, it has been a few days since I requested for the Jira account on the ASF self-serve portal. Who can I reach out to review my Jira account request? Thank you!

Hi, does the flink-kubernetes-operator guarantee support for a minimum number of flink versions? When does it deem a flink version not supported anymore?

Hello, I recently created this issue regarding thread safety issues during jobs submission via REST API (usually happens during job submission by Flink K8S Operator). Would appreciate feedback

Is there a timeline for Apache Pulsar Connector for Flink 1.20 and 2.0 ?

Hello everyone, I've noticed that there is no support for datastream window join in pyflink. I've opened a discussion on the dev mailing list but sadly there has been no response yet from the community. More generally, I am very keen to become more involved in the flink community and more specifically pyflink by doing some tickets and contributing to the project. Let me know if and how I can help Thank you very much

Hi, I'm waiting for Flink Kubernetes Operator v1.12. Does anyone know when it will be available for use?

Hi, I am looking into building Volcano support for Flink natively vs the K8s operator I see this flip got approved and is still open https://cwiki.apache.org/confluence/display/FLINK/FLIP-250%3A+Support+Customized+Kubernetes+Schedulers+Proposal Any idea on the status of this work?

Hi all, I've raised a PR to address an issue in KinesisSource which causes high CPU utilisation and throttling when reading from idle shards. Would appreciate review from committer. Thank you! https://github.com/apache/flink-connector-aws/pull/195

👋 hey all was hoping to get a review on this pr. and is it possible to backport it to 1.20? https://github.com/apache/flink/pull/26301

Hi all, I am interested in the current setup and infrastructure used to build and test PRs, nightly and release candidates as well as performance tests. Specifically, what is the best resource for me to start looking into this? I am looking into possibility of contributing additional coverage and as well as infrastructures to run these. Thank you!

Hi, any tentative date to release flink-connector-kafka 4.0.0 lib which supports flink 2.0, thanks

Hello I am new here and having committed to other Apache projects in the past, I plan to contribute with good first issues (and probably with the flink-iceberg interaction given my experience with the latter)

What is the process of adding test cases for a class in Flink? Do I need to create a JIRA ticket for it??

Hi All, I have a question from security standpoint on flink's docker image. If you see the dockerfile of flink's docker image, we are adding gosu: https://github.com/apache/flink-docker/blob/6e226503dbb228467905c70ccfb6f33f4c676872/1.20/scala_2.12-java17-ubuntu/Dockerfile#L27-L44 This is later used to switch to the user

flink

in the entrypoint script: https://github.com/apache/flink-docker/blob/6e226503dbb228467905c70ccfb6f33f4c676872/1.20/scala_2.12-java17-ubuntu/docker-entrypoint.sh#L37. Gosu itself is tagged by popular scanners as being vulnerable due to outdated golang usage (though author claims that gosu itself is not). My question is, why do we need gosu to switch to

flink

user and not use

USER flink:flink

or something similar in

Dockerfile

? This way we could avoid depending on gosu, and present a cleaner scan result.

Hello everyone, I'm the RM of release-2.1, I'm here to ask for help, currently there is a test case on the master branch that fails a lot, is there anyone who is familiar with this test and can help to pinpoint the root cause, thank you very much. • https://issues.apache.org/jira/browse/FLINK-37701 As a side note, this case has been failing since release-2.0.

Hey - we've run into a schema registry registration issue which has been resolved here , but it not merged yet. I'd like to understand whether we can assist in getting the change vetted to allow it to be added to the main project.

Hi All, yesterday I've pushed to flink-web repo because we're releasing the 1.12.0 operator. Seems like the blog post is not appearing on https://flink.apache.org/posts/. Any idea what could be wrong?

does anyone know the status of support for Flink 2.0 in the Iceberg connector ? There's only 1.20 on Maven, but I've seen activity here: https://github.com/apache/iceberg/tree/main/flink/v2.0

hi all, i have raised a PR : https://github.com/apache/flink-shaded/pull/147 to update the curator library. Would appreciate review from the flink community members. Thank you 🙂

Hey folks, I have a PR for updating the Flink Connector Kafka to the 3.9.1 Kafka client version. Would someone with the power be able to kick off the CI? Maybe @Martijn Visser or @Arvid Heise?

Does flink 1.18 have basic support for Java 21 ? I am trying to place which version of flink I need to upgrade to use Java 21. From the docs it looks like flink 2.0 has full support, but I might not need the full support. Please clarify. Thanks

👋 We're looking for Flink Connector Kafka maintainers to look at an internal change we've merged for handling Projection Pushdown for Kafka https://github.com/apache/flink-connector-kafka/pull/174 This helps us handle non-uniform streams (CDC), particularly with Flink SQL and the Flink Catalog.

hi all, RM for Flink 1.19.3 and 1.20.2 here, both release requires binding votes, if any PMC has some time, pls vote: • 1.19 thread: https://lists.apache.org/thread/bw2s9j76k3glt16365ls4hbqbh820qd8 • 1.20 thread: https://lists.apache.org/thread/xwsmpvj6t0jnhkyfhpqd4m79hnyssjyy

Does anyone know what the release schedule looks like for the 1.19.3 and 1.20.2 releases? When will the official releases be available?

Hey, opened a ticket and made a pr a few months ago was hoping to get some eyes on to merge this one soon.