Re: Flink limitations under Beam

Thanks Aljoscha,How sever of performance difference are we talking about here?  examples of "Iterations/feedback edges" and how they get applied?Thanks so much Aljoscha. From: Aljoscha Krettek To: dev@flink.apache.org; amir bahmanyari Sent:

Re: [VOTE] Release Apache Flink 1.2.1 (RC1)

Yes, aljoscha already opened one against master: https://github.com/apache/flink/pull/3670 On 04.04.2017 17:57, Ted Yu wrote: Should the commits be reverted from master branch as well ? On Tue, Apr 4, 2017 at 4:59 AM, Aljoscha Krettek wrote: The commits around

[jira] [Created] (FLINK-6265) Fix consecutive() for times() pattern.

Kostas Kloudas created FLINK-6265: - Summary: Fix consecutive() for times() pattern. Key: FLINK-6265 URL: https://issues.apache.org/jira/browse/FLINK-6265 Project: Flink Issue Type: Bug

Re: [VOTE] Release Apache Flink 1.2.1 (RC1)

Should the commits be reverted from master branch as well ? On Tue, Apr 4, 2017 at 4:59 AM, Aljoscha Krettek wrote: > The commits around FLINK-5808 have been reverted on release-1.2. > > > On 4. Apr 2017, at 12:16, Stefan Richter > wrote: > > >

[jira] [Created] (FLINK-6264) Kafka consumer fails if can't find leader for partition

Gyula Fora created FLINK-6264: - Summary: Kafka consumer fails if can't find leader for partition Key: FLINK-6264 URL: https://issues.apache.org/jira/browse/FLINK-6264 Project: Flink Issue Type:

[jira] [Created] (FLINK-6263) Leader error in Kafka producer on leader change (broker restart/failrue)

Gyula Fora created FLINK-6263: - Summary: Leader error in Kafka producer on leader change (broker restart/failrue) Key: FLINK-6263 URL: https://issues.apache.org/jira/browse/FLINK-6263 Project: Flink

Re: [VOTE] Release Apache Flink 1.2.1 (RC1)

The commits around FLINK-5808 have been reverted on release-1.2. > On 4. Apr 2017, at 12:16, Stefan Richter wrote: > > I have created a custom build of RocksDB 4.11.2 that fixes a significant > performance problem with append operations. I think this should

Re: [VOTE] Release Apache Flink 1.2.1 (RC1)

I have created a custom build of RocksDB 4.11.2 that fixes a significant performance problem with append operations. I think this should definitely be part of the 1.2.1 release because this is already blocking some users. What is missing is uploading the jar to maven central and a testing run,

Re: [VOTE] Release Apache Flink 1.2.1 (RC1)

Would it be possible to merge also the PR to fix FLINK-6103 ( https://github.com/apache/flink/pull/3598)? On Tue, Apr 4, 2017 at 12:03 PM, Chesnay Schepler wrote: > We can merge the metric changes; I'll rebase the branch and merge them > within the next hours. > > On

Re: [VOTE] Release Apache Flink 1.2.1 (RC1)

We can merge the metric changes; I'll rebase the branch and merge them within the next hours. On 04.04.2017 11:57, Robert Metzger wrote: Thank you for opening a PR for this. Chesnay, do you need more reviews for the metrics changes / backports? Are there any other release blockers for 1.2.1,

Re: [VOTE] Release Apache Flink 1.2.1 (RC1)

Thank you for opening a PR for this. Chesnay, do you need more reviews for the metrics changes / backports? Are there any other release blockers for 1.2.1, or are we good to go? On Mon, Apr 3, 2017 at 6:48 PM, Aljoscha Krettek wrote: > I created a PR for the revert:

[jira] [Created] (FLINK-6261) Add support for TUMBLE, HOP, SESSION to batch SQL

Fabian Hueske created FLINK-6261: Summary: Add support for TUMBLE, HOP, SESSION to batch SQL Key: FLINK-6261 URL: https://issues.apache.org/jira/browse/FLINK-6261 Project: Flink Issue Type:

Re: Flink limitations under Beam

Hi, I think you would mainly lose on performance, since Beam adds an additional layer of abstraction and has some requirements from Runners that slow things down compared to running natively on Flink. I think feature wise, the only thing that you will lose is support for Iterations/feedback

StreamSQL procTime granularity

Hi guys, Based on the discussion about time management in https://github.com/apache/flink/pull/3641 , does it make sense to use nanoTime for procTime semantic aggregation processing? This way we will not have the possibility of events falling in the same "millisecond" processing bucket and

[jira] [Created] (FLINK-6260) Distinct Aggregates for Group By Windows

radu created FLINK-6260: --- Summary: Distinct Aggregates for Group By Windows Key: FLINK-6260 URL: https://issues.apache.org/jira/browse/FLINK-6260 Project: Flink Issue Type: New Feature