There are three levels of support we could land on.
1. Flink works with unreserved resources (revert FLINK-7294).
2. Flink works with unreserved resources, and correctly ignores reserved
resources (revert FLINK-7294 and mitigate Fenzo bug).
3. Flink works with unreserved resources and reserved resources.

3 is a moon shot.  Striving for 2.  Fallback on 1.



On Fri, Dec 1, 2017 at 2:10 PM, Aljoscha Krettek <aljos...@apache.org>
wrote:

> Thanks for the update!
>
> Just to be clear, you're proposing going forward with the "simple fix" of
> reverting FLINK-7294?
>
> > On 1. Dec 2017, at 18:39, Eron Wright <eronwri...@gmail.com> wrote:
> >
> > Update on reported Mesos issue (FLINK-8174):
> >
> > TLDR; a PR will be ready within 24 hours that will undo reservation
> support.
> >
> > A couple of months ago, a fix (FLINK-7294) was merged related to how
> Flink
> > accepts Mesos resource offers.  The intention was to allow Flink to make
> > use of so-called +reserved+ resources, a Mesos feature which makes it
> > possible to reserve hosts for use by a specific framework/role.  The fix
> > inadvertently regressed the ability to use +unreserved+ resources.  This
> is
> > a serious regression because unreserved resources are the common case.
> >
> > The simple solution is to revert the earlier fix, deferring support for
> > reservations to another release.   We are spending some time to find a
> fix
> > that works for all scenarios, but seems unlikely at this time.   I am
> > reaching out to the original contributor to get their feedback.
> >
> > In the course of the investigation, a related flaw was discovered in
> Fenzo
> > that causes Flink to misinterpret offers that contain a mix of reserved
> and
> > unreserved resources.   I believe that a small fix is possible purely
> > within Flink; an update to Fenzo does not appear necessary.
> >
> > Going forward, we will contribute an improved integration test suite with
> > which to test Flink under diverse Mesos conditions (e.g. reservations).
> >
> > Thanks,
> > Eron
> >
> > On Thu, Nov 30, 2017 at 9:47 PM, Tzu-Li (Gordon) Tai <
> tzuli...@apache.org>
> > wrote:
> >
> >> Hi,
> >>
> >> I’ve noticed a behavioral regression in the Kafka producer, that should
> >> also be considered a blocker: https://issues.apache.org/
> >> jira/browse/FLINK-8181
> >> There’s already a PR for the issue here: https://github.com/
> >> apache/flink/pull/5108
> >>
> >> Best,
> >> Gordon
> >>
> >> On 30 November 2017 at 5:27:22 PM, Fabian Hueske (fhue...@gmail.com)
> >> wrote:
> >>
> >> I've created a JIRA issue for the the Hadoop 2.9.0 build problem [1].
> >>
> >> Best, Fabian
> >>
> >> [1] https://issues.apache.org/jira/browse/FLINK-8177
> >>
> >> 2017-11-30 4:35 GMT+01:00 Eron Wright <eronwri...@gmail.com>:
> >>
> >>> Unfortunately we've identified a blocker bug for Flink on Mesos -
> >>> FLINK-8174. We'll have a patch ready on Thursday.
> >>>
> >>> Thanks,
> >>> Eron
> >>>
> >>> On Wed, Nov 29, 2017 at 3:40 PM, Eron Wright <eronwri...@gmail.com>
> >> wrote:
> >>>
> >>>> On Dell EMC side, we're testing the RC2 on DCOS 1.10.0. Seeing a
> >>>> potential issue with offer acceptance and we'll update the thread with
> >> a
> >>> +1
> >>>> or with a more concrete issue within 24 hours.
> >>>>
> >>>> Thanks,
> >>>> Eron
> >>>>
> >>>> On Wed, Nov 29, 2017 at 6:54 AM, Chesnay Schepler <ches...@apache.org
> >
> >>>> wrote:
> >>>>
> >>>>> I don't think anyone has taken a look yet, nor was there a discussion
> >> as
> >>>>> to postponing it.
> >>>>>
> >>>>> It just slipped through the cracks i guess...
> >>>>>
> >>>>>
> >>>>> On 29.11.2017 15:47, Gyula Fóra wrote:
> >>>>>
> >>>>>> Hi guys,
> >>>>>> I ran into this again while playing with savepoint/restore
> >> parallelism:
> >>>>>>
> >>>>>> https://issues.apache.org/jira/browse/FLINK-7595
> >>>>>> https://github.com/apache/flink/pull/4651
> >>>>>>
> >>>>>> Anyone has some idea about the status of this PR or were we planning
> >> to
> >>>>>> postpone this to 1.5?
> >>>>>>
> >>>>>> Thanks,
> >>>>>> Gyula
> >>>>>>
> >>>>>>
> >>>>>> Fabian Hueske <fhue...@gmail.com> ezt írta (időpont: 2017. nov.
> 29.,
> >>>>>> Sze,
> >>>>>> 13:10):
> >>>>>>
> >>>>>> OK, the situation is the following:
> >>>>>>>
> >>>>>>> The test class (org.apache.flink.yarn.UtilsTest) implements a
> >> Hadoop
> >>>>>>> interface (Container) that was extended in Hadoop 2.9.0 by a getter
> >>> and
> >>>>>>> setter.
> >>>>>>> By adding the methods, we can compile Flink for Hadoop 2.9.0.
> >> However,
> >>>>>>> the
> >>>>>>> getter/setter add a dependency on a class that was also added in
> >>> Hadoop
> >>>>>>> 2.9.0.
> >>>>>>> Therefore, the implementation is not backwards compatible with
> >> Hadoop
> >>>>>>> versions < 2.9.0.
> >>>>>>>
> >>>>>>> Not sure how we can fix the problem. We would need two version of
> >> the
> >>>>>>> class
> >>>>>>> that are chosen based on the Hadoop version. Do we have something
> >> like
> >>>>>>> that
> >>>>>>> somewhere else?
> >>>>>>>
> >>>>>>> Since this is only a problem in a test class, Flink 1.4.0 might
> >> still
> >>>>>>> work
> >>>>>>> very well with Hadoop 2.9.0.
> >>>>>>> However, this has not been tested AFAIK.
> >>>>>>>
> >>>>>>> Cheers, Fabian
> >>>>>>>
> >>>>>>> 2017-11-29 12:47 GMT+01:00 Fabian Hueske <fhue...@gmail.com>:
> >>>>>>>
> >>>>>>> I just tried to build the release-1.4 branch for Hadoop 2.9.0
> >>> (released
> >>>>>>>> a
> >>>>>>>> few days ago) and got a compilation failure in a test class.
> >>>>>>>>
> >>>>>>>> Right now, I'm assessing how much we need to fix to support Hadoop
> >>>>>>>> 2.9.0.
> >>>>>>>> I'll report later.
> >>>>>>>>
> >>>>>>>> Best, Fabian
> >>>>>>>>
> >>>>>>>> 2017-11-29 11:16 GMT+01:00 Aljoscha Krettek <aljos...@apache.org
> >:
> >>>>>>>>
> >>>>>>>> Agreed, this is a regression compared to the previous
> >> functionality.
> >>> I
> >>>>>>>>> updated the issue to "Blocker".
> >>>>>>>>>
> >>>>>>>>> On 29. Nov 2017, at 10:01, Gyula Fóra <gyula.f...@gmail.com>
> >> wrote:
> >>>>>>>>>>
> >>>>>>>>>> Hi all,
> >>>>>>>>>>
> >>>>>>>>>> I have found the following issue:
> >>>>>>>>>> https://issues.apache.org/jira/browse/FLINK-8165
> >>>>>>>>>>
> >>>>>>>>>> I would say this is a blocker (I personally pass the
> >> ParameterTool
> >>>>>>>>>> all
> >>>>>>>>>>
> >>>>>>>>> over
> >>>>>>>>>
> >>>>>>>>>> the place in my production apps), but a pretty trivial issue to
> >>> fix,
> >>>>>>>>>>
> >>>>>>>>> we
> >>>>>>>
> >>>>>>>> can
> >>>>>>>>>
> >>>>>>>>>> wait a little to find other potential problems.
> >>>>>>>>>>
> >>>>>>>>>> I can submit a fix in a little bit.
> >>>>>>>>>>
> >>>>>>>>>> Cheers,
> >>>>>>>>>> Gyula
> >>>>>>>>>>
> >>>>>>>>>> Tzu-Li (Gordon) Tai <tzuli...@apache.org> ezt írta (időpont:
> >> 2017.
> >>>>>>>>>>
> >>>>>>>>> nov.
> >>>>>>>
> >>>>>>>> 29., Sze, 9:23):
> >>>>>>>>>>
> >>>>>>>>>> +1
> >>>>>>>>>>>
> >>>>>>>>>>> Verified:
> >>>>>>>>>>> - No missing release Maven artifacts
> >>>>>>>>>>> - Staged Apache source & binary convenience releases looks good
> >>>>>>>>>>> - NOTICE / LICENSE is correct, README is sane
> >>>>>>>>>>> - Built from source (macOS, Scala 2.11, Hadoop-free & Hadoop
> >> 2.8)
> >>>>>>>>>>> - Cluster testing on AWS EMR (see release-testing-doc for
> >>>>>>>>>>>
> >>>>>>>>>> configuration
> >>>>>>>
> >>>>>>>> details)
> >>>>>>>>>>> - Tested Kinesis / Elasticsearch connector (no dependency
> >> clashes
> >>> on
> >>>>>>>>>>> cluster execution, works locally in IDE)
> >>>>>>>>>>>
> >>>>>>>>>>> Thanks a lot for managing the release Aljoscha!
> >>>>>>>>>>>
> >>>>>>>>>>> Cheers,
> >>>>>>>>>>> Gordon
> >>>>>>>>>>>
> >>>>>>>>>>> On 28 November 2017 at 8:32:42 PM, Stefan Richter (
> >>>>>>>>>>> s.rich...@data-artisans.com) wrote:
> >>>>>>>>>>>
> >>>>>>>>>>> +1 (non-binding)
> >>>>>>>>>>>
> >>>>>>>>>>> I tested Flink in a cluster setup on Google Cloud,
> YARN-per-job,
> >>>>>>>>>>>
> >>>>>>>>>> checked
> >>>>>>>>>
> >>>>>>>>>> that for all backends that HA, recovery, at-least-once,
> >> end-to-end
> >>>>>>>>>>>
> >>>>>>>>>> exactly
> >>>>>>>>>
> >>>>>>>>>> once (with Kafka11 Producer), savepoints, externalized
> >> checkpoints,
> >>>>>>>>>>>
> >>>>>>>>>> and
> >>>>>>>
> >>>>>>>> rescaling work correctly.
> >>>>>>>>>>>
> >>>>>>>>>>> Am 28.11.2017 um 11:47 schrieb Aljoscha Krettek <
> >>>>>>>>>>>>
> >>>>>>>>>>> aljos...@apache.org
> >>>>>>>
> >>>>>>>> :
> >>>>>>>>>>
> >>>>>>>>>>> +1
> >>>>>>>>>>>>
> >>>>>>>>>>>> Verified:
> >>>>>>>>>>>> - NOTICE and LICENSE are correct
> >>>>>>>>>>>> - source doesn't contain binaries
> >>>>>>>>>>>> - verified signatures
> >>>>>>>>>>>> - verified hashes
> >>>>>>>>>>>> - cluster testing on AWS and Cloudera VM (with Kerberos) (see
> >>>>>>>>>>>>
> >>>>>>>>>>> release-testing doc)
> >>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>> On 28. Nov 2017, at 11:20, Aljoscha Krettek <
> >> aljos...@apache.org
> >>>>
> >>>>>>>>>>>>>
> >>>>>>>>>>>> wrote:
> >>>>>>>>>>>
> >>>>>>>>>>>> Phew, thanks for the update!
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> On 28. Nov 2017, at 11:19, Gyula Fóra <gyf...@apache.org>
> >>> wrote:
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> Ok seems like I had to remove the snappy jar as it was
> >>> corrupted
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>> (makes
> >>>>>>>>>
> >>>>>>>>>> total sense) :P
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> Gyula Fóra <gyf...@apache.org> ezt írta (időpont: 2017.
> nov.
> >>>>>>>>>>>>>> 28.,
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>> K,
> >>>>>>>>>
> >>>>>>>>>> 11:13):
> >>>>>>>>>>>
> >>>>>>>>>>>> Hi Aljoscha,
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Thanks for the release candidate. I am having a hard time
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>> building
> >>>>>>>
> >>>>>>>> the rc,
> >>>>>>>>>>>
> >>>>>>>>>>>> I seem to get this error no matter what I do:
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> [ERROR] Failed to execute goal
> >>>>>>>>>>>>>>> org.apache.maven.plugins:maven-shade-plugin:2.4.1:shade
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>> (shade-hadoop) on
> >>>>>>>>>>>
> >>>>>>>>>>>> project flink-shaded-hadoop2-uber: Error creating shaded jar:
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>> invalid
> >>>>>>>>>
> >>>>>>>>>> LOC
> >>>>>>>>>>>
> >>>>>>>>>>>> header (bad signature) -> [Help 1]
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> (Apache Maven 3.3.9)
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Any idea what I am missing?
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Thanks,
> >>>>>>>>>>>>>>> Gyula
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Aljoscha Krettek <aljos...@apache.org> ezt írta (időpont:
> >>> 2017.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>> nov.
> >>>>>>>>>
> >>>>>>>>>> 27.,
> >>>>>>>>>>>
> >>>>>>>>>>>> H, 19:35):
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> Hi everyone,
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> Please review and vote on release candidate #2 for the
> >>> version
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> 1.4.0, as
> >>>>>>>>>>>
> >>>>>>>>>>>> follows:
> >>>>>>>>>>>>>>>> [ ] +1, Approve the release
> >>>>>>>>>>>>>>>> [ ] -1, Do not approve the release (please provide
> specific
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> comments)
> >>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>>>>>>> The complete staging area is available for your review,
> >> which
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> includes:
> >>>>>>>>>>>
> >>>>>>>>>>>> * JIRA release notes [1],
> >>>>>>>>>>>>>>>> * the official Apache source release and binary
> convenience
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> releases
> >>>>>>>>>
> >>>>>>>>>> to
> >>>>>>>>>>>
> >>>>>>>>>>>> be deployed to dist.apache.org[2], which are signed with the
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> key
> >>>>>>>
> >>>>>>>> with
> >>>>>>>>>>>
> >>>>>>>>>>>> fingerprint F2A67A8047499BBB3908D17AA8F4FD97121D7293 [3],
> >>>>>>>>>>>>>>>> * all artifacts to be deployed to the Maven Central
> >>> Repository
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> [4],
> >>>>>>>>>
> >>>>>>>>>> * source code tag "release-1.4.0-rc1" [5],
> >>>>>>>>>>>>>>>> * website pull request listing the new release [6].
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> Please have a careful look at the website PR because I
> >>> changed
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> some
> >>>>>>>>>
> >>>>>>>>>> wording and we're now also releasing a binary without Hadoop
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> dependencies.
> >>>>>>>>>>>
> >>>>>>>>>>>> Please use this document for coordinating testing efforts: [7]
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> The only change between RC1 and this RC2 is that the
> source
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> release
> >>>>>>>>>
> >>>>>>>>>> package does not include the erroneously included binary Ruby
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> dependencies
> >>>>>>>>>>>
> >>>>>>>>>>>> of the documentation anymore. Because of this I would like to
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> propose a
> >>>>>>>>>>>
> >>>>>>>>>>>> shorter voting time and close the vote around the time that
> RC1
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> would have
> >>>>>>>>>>>
> >>>>>>>>>>>> closed. This would mean closing by end of Wednesday. Please
> let
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> me
> >>>>>>>
> >>>>>>>> know if
> >>>>>>>>>>>
> >>>>>>>>>>>> you disagree with this. The vote is adopted by majority
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> approval,
> >>>>>>>
> >>>>>>>> with at
> >>>>>>>>>>>
> >>>>>>>>>>>> least 3 PMC affirmative votes.
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> Thanks,
> >>>>>>>>>>>>>>>> Your friendly Release Manager
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> [1]
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> https://issues.apache.org/jira/secure/ReleaseNote.jspa?
> >> proje
> >>>>>>>>>>>
> >>>>>>>>>> ctId=12315522&version=12340533
> >>>>>>>>>
> >>>>>>>>>> [2] http://people.apache.org/~aljoscha/flink-1.4.0-rc2/
> >>>>>>>>>>>>>>>> [3] https://dist.apache.org/repos/dist/release/flink/KEYS
> >>>>>>>>>>>>>>>> [4]
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> https://repository.apache.org/
> >> content/repositories/orgapache
> >>>>>>> flink-1140
> >>>>>>>
> >>>>>>>> [5]
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> https://git-wip-us.apache.org/
> >> repos/asf?p=flink.git;a=tag;h=
> >>>>>>>>>>>
> >>>>>>>>>> ea751b7b23b23446ed3fcdeed564bbe8bf4adf9c
> >>>>>>>>>
> >>>>>>>>>> [6] https://github.com/apache/flink-web/pull/95
> >>>>>>>>>>>>>>>> [7]
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> https://docs.google.com/document/d/
> >> 1HqYyrNoMSXwo8zBpZj7s39Uz
> >>>>>>>>>>>
> >>>>>>>>>> UdlFcFO8TRpHNZ_cl44/edit?usp=sharing
> >>>>>>>>>
> >>>>>>>>>> Pro-tip: you can create a settings.xml file with these contents:
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> <settings>
> >>>>>>>>>>>>>>>> <activeProfiles>
> >>>>>>>>>>>>>>>> <activeProfile>flink-1.4.0</activeProfile>
> >>>>>>>>>>>>>>>> </activeProfiles>
> >>>>>>>>>>>>>>>> <profiles>
> >>>>>>>>>>>>>>>> <profile>
> >>>>>>>>>>>>>>>> <id>flink-1.4.0</id>
> >>>>>>>>>>>>>>>> <repositories>
> >>>>>>>>>>>>>>>> <repository>
> >>>>>>>>>>>>>>>> <id>flink-1.4.0</id>
> >>>>>>>>>>>>>>>> <url>
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> https://repository.apache.org/
> >> content/repositories/orgapache
> >>>>>>>>>>>
> >>>>>>>>>> flink-1140/
> >>>>>>>>>
> >>>>>>>>>> </url>
> >>>>>>>>>>>>>>>> </repository>
> >>>>>>>>>>>>>>>> <repository>
> >>>>>>>>>>>>>>>> <id>archetype</id>
> >>>>>>>>>>>>>>>> <url>
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> https://repository.apache.org/
> >> content/repositories/orgapache
> >>>>>>>>>>>
> >>>>>>>>>> flink-1140/
> >>>>>>>>>
> >>>>>>>>>> </url>
> >>>>>>>>>>>>>>>> </repository>
> >>>>>>>>>>>>>>>> </repositories>
> >>>>>>>>>>>>>>>> </profile>
> >>>>>>>>>>>>>>>> </profiles>
> >>>>>>>>>>>>>>>> </settings>
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>> And reference that in you maven commands via --settings
> >>>>>>>>>>>>>>>> path/to/settings.xml. This is useful for creating a
> >>> quickstart
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> based
> >>>>>>>>>
> >>>>>>>>>> on the
> >>>>>>>>>>>
> >>>>>>>>>>>> staged release and for building against the staged jars.
> >>>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>
> >>>>>
> >>>>
> >>>
> >>
>
>

Reply via email to