You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@flink.apache.org by Gyula Fóra <gy...@apache.org> on 2022/03/11 17:10:43 UTC
[DISCUSS] Preview release for Flink Kubernetes Operator
Hi Team!
I would like to discuss the timeline for the initial preview/milestone
release of the flink-kubernetes-operator
<https://github.com/apache/flink-kubernetes-operator> project.
The last few weeks we have been working very hard with the community to
stabilize the initial feature set and I think we have made great progress.
While we are still far from a production ready-state, a preview release
will give us the opportunity to reach more people and gather much needed
input to take this project to the next level.
There are still a couple missing features that we need to iron out and we
need to make sure we have proper documentation but after that I think it
would be a good time for the preview release.
I propose to aim for the first release candidate around the 25-27th of
March after which we should dedicate a few days for some extensive testing
and bugfixing.
What do you think?
Gyula
Re: [DISCUSS] Preview release for Flink Kubernetes Operator
Posted by Yang Wang <da...@gmail.com>.
The whole release plan sounds good to me. Thanks for driving this.
Best,
Yang
Gyula Fóra <gy...@apache.org> 于2022年3月23日周三 16:22写道：
> Hi Team!
>
> I think we are in good shape to cut the preview release branch in a few
> days and start testing.
> I suggest we aim for a feature freeze on Friday the 25th (except for the
> ongoing session job work which might need 1-2 extra days) and cut the
> release branch on Monday the 28th (and hopefully prepare the first RC).
>
> Until then we should work out the missing pieces around publishing the
> artifacts and aim to wrap up any critical outstanding tickets.
>
> Cheers!
> Gyula
>
> On Tue, Mar 15, 2022 at 3:48 AM Aitozi <gj...@gmail.com> wrote:
>
> > Hi all
> > Thanks for starting the discussion. The first release at the end of
> > March looks good to me, I think the flink-kubernetes-operator is in a
> good
> > shape now. I will try to complete the design to support the session job
> in
> > these two days and will start another discussion before the work.
> > Regarding what has to be improved before release, I think the
> operator
> > metrics are not completed yet, we may have to inject more useful metrics.
> > The feature list
> >
> https://github.com/bgeng777/flink-kubernetes-operator/blob/features/doc/features.md
> posted
> > by @Biao Geng looks good, Maybe we could include this to track the
> > supported features.
> >
> > Best,
> > Aitozi.
> >
> > Gyula Fóra <gy...@gmail.com> 于2022年3月14日周一 16:58写道：
> >
> >> Hi Xintong!
> >>
> >> Thank you for the valuable input, you are completely right we need to
> >> agree
> >> and document these aspects.
> >>
> >> Let me try to address some of the questions and others should chip in
> also
> >> :)
> >>
> >> 1. Version convention:
> >>
> >> I think we should adopt the 3 digit versioning scheme like other flink
> >> projects (I think flink-shaded is a bit of an outlier here).
> >> The preview release should be 0.1.0.
> >>
> >> The supported Flink versions should be documented with the release, each
> >> version of the operator should technically be able to support multiple
> >> Flink versions at the same time.
> >> For the preview release this should be Flink version >= 1.14. We should
> >> later come up with a guarantee that we do not drop flink version support
> >> within the same minor version.
> >>
> >> The only public API here at the moment are the custom resource
> >> definitions.
> >> We agreed to mark them all experimental for the preview release, and I
> >> think for 1.0.0 we should aim for public evolving.
> >> Otherwise we should respect the Flink guarantees here. Practically this
> >> means that for the preview release we do not guarantee anything in terms
> >> of
> >> later compatibility.
> >>
> >> 2. Release process:
> >> I agree that we should follow the official Flink process as much as
> >> possible (and makes sense) with proper voting etc.
> >>
> >> Cheers,
> >> Gyula
> >>
> >>
> >> On Mon, Mar 14, 2022 at 8:32 AM Xintong Song <to...@gmail.com>
> >> wrote:
> >>
> >> > How everyone,
> >> >
> >> > It's great to learn that we are approaching a preview release for
> >> > the flink-kubernetes-operator. Thanks for the efforts.
> >> >
> >> > I have not been involved with any developing efforts in
> >> > flink-kubernetes-operator, thus have no comment on end of March being
> >> the
> >> > targeting date.
> >> >
> >> > However, based on my experiences being one of the Flink release
> >> managers, I
> >> > see a few things that are still missing for creating an official
> >> release.
> >> > (IIUC, the preview release is still an official release, just with
> weak
> >> > functionality and compatibility guarantees.)
> >> >
> >> > 1. The version conventions:
> >> > - How does a flink-kubernetes-operator version look like? E.g., flink
> /
> >> > flink-statefun / flink-ml have three digits x.y.z, while flink-shaded
> >> has
> >> > only two digits x.y.
> >> > - What is the relationship between flink-kubernetes-operator and flink
> >> > versions? E.g., flink-shaded x.* is only designed to support flink
> >> *.x.*.
> >> > - What kind of compatibility guarantees do we provide? E.g., flink
> >> expects
> >> > no Public API compatibility should be broken between minor releases
> (the
> >> > 2nd digit) and no PublicEvolving APIs should be broken between bugfix
> >> > releases (the 3rd digit).
> >> > - What kind of support do we provide for old releases? E.g., flink
> >> provides
> >> > bug fixes for the latest two minor releases (the 2nd digit).
> >> >
> >> > 2. Release process
> >> > You may find the release process for all Flink artifacts in this wiki
> >> page
> >> > [1]. Such a formal documented process would help us to reach consensus
> >> on
> >> > what needs to be done and make sure it complies with the ASF
> regulations
> >> > before creating a release. We probably don't need something as formal
> >> as a
> >> > vote to approve the release process. But we definitely need a formal
> >> vote
> >> > for the flink-kubernetes-operator release, and the release process
> would
> >> > help making sure we are on the same page about what is a releasable
> >> state
> >> > for this artifact.
> >> >
> >> > WDYT?
> >> >
> >> > Thank you~
> >> >
> >> > Xintong Song
> >> >
> >> >
> >> > [1] https://cwiki.apache.org/confluence/display/FLINK/Releasing
> >> >
> >> > On Mon, Mar 14, 2022 at 11:34 AM Biao Geng <bi...@gmail.com>
> wrote:
> >> >
> >> > > Hi there,
> >> > >
> >> > > It is exciting to see the discussion of the release timeline! I
> agree
> >> > that
> >> > > the end of March is a proper date.
> >> > > To make others easier get involved in this discussion, I think we
> may
> >> > need
> >> > > to provide a more straightforward feature list for the preview
> >> release.
> >> > The
> >> > > "Initial Feature Set" in FLIP-212
> >> > > <
> >> > >
> >> >
> >>
> https://cwiki.apache.org/confluence/display/FLINK/FLIP-212:+Introduce+Flink+Kubernetes+Operator
> >> > > >
> >> > > is
> >> > > almost complete. But some new features like webhook based validate
> and
> >> > > flink operator metric are not added and they are only tracked in the
> >> long
> >> > > JIRA list. If we can update the FLIP, it may be more convenient and
> >> can
> >> > > also help us write release notes later. I also created a draft
> >> > > <
> >> > >
> >> >
> >>
> https://github.com/bgeng777/flink-kubernetes-operator/blob/features/doc/features.md
> >> > > >
> >> > > for myself to track completed or in-plan features. Hope it can help.
> >> > >
> >> > > Best,
> >> > > Biao Geng
> >> > >
> >> > > Gyula Fóra <gy...@gmail.com> 于2022年3月14日周一 04:11写道：
> >> > >
> >> > > > @Konstantin: Yes I completely agree that for this release the API
> >> (CRD)
> >> > > > should be marked experimental!
> >> > > > I have opened a ticket to track this:
> >> > > > https://issues.apache.org/jira/browse/FLINK-26620
> >> > > >
> >> > > > @Yang Wang <da...@gmail.com> : I think we still have plenty
> >> of
> >> > > time
> >> > > > to work on features like the session job before the release, would
> >> be
> >> > > nice
> >> > > > to provide a complete story to the users.
> >> > > >
> >> > > > Gyula
> >> > > >
> >> > > > On Sun, Mar 13, 2022 at 5:17 PM Konstantin Knauf <
> knaufk@apache.org
> >> >
> >> > > > wrote:
> >> > > >
> >> > > > > Hi everyone,
> >> > > > >
> >> > > > > can we mark all the APIs as experimental/alpha so that it is
> clear
> >> > that
> >> > > > > these can be broken in future releases for now? I think this
> >> would be
> >> > > > very
> >> > > > > important given the early stage of the project. We want to be
> >> able to
> >> > > > > address shortcomings without worrying too much about backwards
> >> > > > > compatibility at this stage, I believe.
> >> > > > >
> >> > > > > Cheers,
> >> > > > >
> >> > > > > Konstantin
> >> > > > >
> >> > > > > On Sun, Mar 13, 2022 at 7:48 AM Yang Wang <
> danrtsey.wy@gmail.com>
> >> > > wrote:
> >> > > > >
> >> > > > > > Thanks Gyula for starting this discussion.
> >> > > > > >
> >> > > > > > Given that the core functionality is closing to stable, I am
> in
> >> > favor
> >> > > > of
> >> > > > > > having the MVP release at the end of March.
> >> > > > > > The first release will help us to collect more feedbacks from
> >> the
> >> > > > users.
> >> > > > > > Also it is a good chance to let the users know that the
> >> community
> >> > is
> >> > > > > trying
> >> > > > > > to maintain an official Kubernetes operator :)
> >> > > > > > I hope that the companies could build their own production
> >> > streaming
> >> > > > > > platform on top of the flink-kubernetes-operator in the
> future.
> >> > > > > >
> >> > > > > > FYI: @Wenjun Min is still working hard on supporting the
> Session
> >> > Job
> >> > > in
> >> > > > > > Flink Kubernetes operator, It will be great if we could
> include
> >> it
> >> > in
> >> > > > the
> >> > > > > > first release.
> >> > > > > > And I believe we have enough time.
> >> > > > > >
> >> > > > > > Moreover, I agree with you that we need to invest more time in
> >> the
> >> > > > > > documentation, e2e tests, helm install optimization, logging,
> >> > > > > > etc. before the release.
> >> > > > > >
> >> > > > > >
> >> > > > > > Best,
> >> > > > > > Yang
> >> > > > > >
> >> > > > > >
> >> > > > > > Gyula Fóra <gy...@apache.org> 于2022年3月12日周六 01:10写道：
> >> > > > > >
> >> > > > > > > Hi Team!
> >> > > > > > >
> >> > > > > > > I would like to discuss the timeline for the initial
> >> > > > preview/milestone
> >> > > > > > > release of the flink-kubernetes-operator
> >> > > > > > > <https://github.com/apache/flink-kubernetes-operator>
> >> project.
> >> > > > > > >
> >> > > > > > > The last few weeks we have been working very hard with the
> >> > > community
> >> > > > to
> >> > > > > > > stabilize the initial feature set and I think we have made
> >> great
> >> > > > > > progress.
> >> > > > > > > While we are still far from a production ready-state, a
> >> preview
> >> > > > release
> >> > > > > > > will give us the opportunity to reach more people and gather
> >> much
> >> > > > > needed
> >> > > > > > > input to take this project to the next level.
> >> > > > > > >
> >> > > > > > > There are still a couple missing features that we need to
> iron
> >> > out
> >> > > > and
> >> > > > > we
> >> > > > > > > need to make sure we have proper documentation but after
> that
> >> I
> >> > > think
> >> > > > > it
> >> > > > > > > would be a good time for the preview release.
> >> > > > > > >
> >> > > > > > > I propose to aim for the first release candidate around the
> >> > 25-27th
> >> > > > of
> >> > > > > > > March after which we should dedicate a few days for some
> >> > extensive
> >> > > > > > testing
> >> > > > > > > and bugfixing.
> >> > > > > > >
> >> > > > > > > What do you think?
> >> > > > > > >
> >> > > > > > > Gyula
> >> > > > > > >
> >> > > > > >
> >> > > > >
> >> > > > >
> >> > > > > --
> >> > > > >
> >> > > > > Konstantin Knauf
> >> > > > >
> >> > > > > https://twitter.com/snntrable
> >> > > > >
> >> > > > > https://github.com/knaufk
> >> > > > >
> >> > > >
> >> > >
> >> >
> >>
> >
>
Re: [DISCUSS] Preview release for Flink Kubernetes Operator
Posted by Gyula Fóra <gy...@apache.org>.
Hi Team!
I think we are in good shape to cut the preview release branch in a few
days and start testing.
I suggest we aim for a feature freeze on Friday the 25th (except for the
ongoing session job work which might need 1-2 extra days) and cut the
release branch on Monday the 28th (and hopefully prepare the first RC).
Until then we should work out the missing pieces around publishing the
artifacts and aim to wrap up any critical outstanding tickets.
Cheers!
Gyula
On Tue, Mar 15, 2022 at 3:48 AM Aitozi <gj...@gmail.com> wrote:
> Hi all
> Thanks for starting the discussion. The first release at the end of
> March looks good to me, I think the flink-kubernetes-operator is in a good
> shape now. I will try to complete the design to support the session job in
> these two days and will start another discussion before the work.
> Regarding what has to be improved before release, I think the operator
> metrics are not completed yet, we may have to inject more useful metrics.
> The feature list
> https://github.com/bgeng777/flink-kubernetes-operator/blob/features/doc/features.md posted
> by @Biao Geng looks good, Maybe we could include this to track the
> supported features.
>
> Best,
> Aitozi.
>
> Gyula Fóra <gy...@gmail.com> 于2022年3月14日周一 16:58写道：
>
>> Hi Xintong!
>>
>> Thank you for the valuable input, you are completely right we need to
>> agree
>> and document these aspects.
>>
>> Let me try to address some of the questions and others should chip in also
>> :)
>>
>> 1. Version convention:
>>
>> I think we should adopt the 3 digit versioning scheme like other flink
>> projects (I think flink-shaded is a bit of an outlier here).
>> The preview release should be 0.1.0.
>>
>> The supported Flink versions should be documented with the release, each
>> version of the operator should technically be able to support multiple
>> Flink versions at the same time.
>> For the preview release this should be Flink version >= 1.14. We should
>> later come up with a guarantee that we do not drop flink version support
>> within the same minor version.
>>
>> The only public API here at the moment are the custom resource
>> definitions.
>> We agreed to mark them all experimental for the preview release, and I
>> think for 1.0.0 we should aim for public evolving.
>> Otherwise we should respect the Flink guarantees here. Practically this
>> means that for the preview release we do not guarantee anything in terms
>> of
>> later compatibility.
>>
>> 2. Release process:
>> I agree that we should follow the official Flink process as much as
>> possible (and makes sense) with proper voting etc.
>>
>> Cheers,
>> Gyula
>>
>>
>> On Mon, Mar 14, 2022 at 8:32 AM Xintong Song <to...@gmail.com>
>> wrote:
>>
>> > How everyone,
>> >
>> > It's great to learn that we are approaching a preview release for
>> > the flink-kubernetes-operator. Thanks for the efforts.
>> >
>> > I have not been involved with any developing efforts in
>> > flink-kubernetes-operator, thus have no comment on end of March being
>> the
>> > targeting date.
>> >
>> > However, based on my experiences being one of the Flink release
>> managers, I
>> > see a few things that are still missing for creating an official
>> release.
>> > (IIUC, the preview release is still an official release, just with weak
>> > functionality and compatibility guarantees.)
>> >
>> > 1. The version conventions:
>> > - How does a flink-kubernetes-operator version look like? E.g., flink /
>> > flink-statefun / flink-ml have three digits x.y.z, while flink-shaded
>> has
>> > only two digits x.y.
>> > - What is the relationship between flink-kubernetes-operator and flink
>> > versions? E.g., flink-shaded x.* is only designed to support flink
>> *.x.*.
>> > - What kind of compatibility guarantees do we provide? E.g., flink
>> expects
>> > no Public API compatibility should be broken between minor releases (the
>> > 2nd digit) and no PublicEvolving APIs should be broken between bugfix
>> > releases (the 3rd digit).
>> > - What kind of support do we provide for old releases? E.g., flink
>> provides
>> > bug fixes for the latest two minor releases (the 2nd digit).
>> >
>> > 2. Release process
>> > You may find the release process for all Flink artifacts in this wiki
>> page
>> > [1]. Such a formal documented process would help us to reach consensus
>> on
>> > what needs to be done and make sure it complies with the ASF regulations
>> > before creating a release. We probably don't need something as formal
>> as a
>> > vote to approve the release process. But we definitely need a formal
>> vote
>> > for the flink-kubernetes-operator release, and the release process would
>> > help making sure we are on the same page about what is a releasable
>> state
>> > for this artifact.
>> >
>> > WDYT?
>> >
>> > Thank you~
>> >
>> > Xintong Song
>> >
>> >
>> > [1] https://cwiki.apache.org/confluence/display/FLINK/Releasing
>> >
>> > On Mon, Mar 14, 2022 at 11:34 AM Biao Geng <bi...@gmail.com> wrote:
>> >
>> > > Hi there,
>> > >
>> > > It is exciting to see the discussion of the release timeline! I agree
>> > that
>> > > the end of March is a proper date.
>> > > To make others easier get involved in this discussion, I think we may
>> > need
>> > > to provide a more straightforward feature list for the preview
>> release.
>> > The
>> > > "Initial Feature Set" in FLIP-212
>> > > <
>> > >
>> >
>> https://cwiki.apache.org/confluence/display/FLINK/FLIP-212:+Introduce+Flink+Kubernetes+Operator
>> > > >
>> > > is
>> > > almost complete. But some new features like webhook based validate and
>> > > flink operator metric are not added and they are only tracked in the
>> long
>> > > JIRA list. If we can update the FLIP, it may be more convenient and
>> can
>> > > also help us write release notes later. I also created a draft
>> > > <
>> > >
>> >
>> https://github.com/bgeng777/flink-kubernetes-operator/blob/features/doc/features.md
>> > > >
>> > > for myself to track completed or in-plan features. Hope it can help.
>> > >
>> > > Best,
>> > > Biao Geng
>> > >
>> > > Gyula Fóra <gy...@gmail.com> 于2022年3月14日周一 04:11写道：
>> > >
>> > > > @Konstantin: Yes I completely agree that for this release the API
>> (CRD)
>> > > > should be marked experimental!
>> > > > I have opened a ticket to track this:
>> > > > https://issues.apache.org/jira/browse/FLINK-26620
>> > > >
>> > > > @Yang Wang <da...@gmail.com> : I think we still have plenty
>> of
>> > > time
>> > > > to work on features like the session job before the release, would
>> be
>> > > nice
>> > > > to provide a complete story to the users.
>> > > >
>> > > > Gyula
>> > > >
>> > > > On Sun, Mar 13, 2022 at 5:17 PM Konstantin Knauf <knaufk@apache.org
>> >
>> > > > wrote:
>> > > >
>> > > > > Hi everyone,
>> > > > >
>> > > > > can we mark all the APIs as experimental/alpha so that it is clear
>> > that
>> > > > > these can be broken in future releases for now? I think this
>> would be
>> > > > very
>> > > > > important given the early stage of the project. We want to be
>> able to
>> > > > > address shortcomings without worrying too much about backwards
>> > > > > compatibility at this stage, I believe.
>> > > > >
>> > > > > Cheers,
>> > > > >
>> > > > > Konstantin
>> > > > >
>> > > > > On Sun, Mar 13, 2022 at 7:48 AM Yang Wang <da...@gmail.com>
>> > > wrote:
>> > > > >
>> > > > > > Thanks Gyula for starting this discussion.
>> > > > > >
>> > > > > > Given that the core functionality is closing to stable, I am in
>> > favor
>> > > > of
>> > > > > > having the MVP release at the end of March.
>> > > > > > The first release will help us to collect more feedbacks from
>> the
>> > > > users.
>> > > > > > Also it is a good chance to let the users know that the
>> community
>> > is
>> > > > > trying
>> > > > > > to maintain an official Kubernetes operator :)
>> > > > > > I hope that the companies could build their own production
>> > streaming
>> > > > > > platform on top of the flink-kubernetes-operator in the future.
>> > > > > >
>> > > > > > FYI: @Wenjun Min is still working hard on supporting the Session
>> > Job
>> > > in
>> > > > > > Flink Kubernetes operator, It will be great if we could include
>> it
>> > in
>> > > > the
>> > > > > > first release.
>> > > > > > And I believe we have enough time.
>> > > > > >
>> > > > > > Moreover, I agree with you that we need to invest more time in
>> the
>> > > > > > documentation, e2e tests, helm install optimization, logging,
>> > > > > > etc. before the release.
>> > > > > >
>> > > > > >
>> > > > > > Best,
>> > > > > > Yang
>> > > > > >
>> > > > > >
>> > > > > > Gyula Fóra <gy...@apache.org> 于2022年3月12日周六 01:10写道：
>> > > > > >
>> > > > > > > Hi Team!
>> > > > > > >
>> > > > > > > I would like to discuss the timeline for the initial
>> > > > preview/milestone
>> > > > > > > release of the flink-kubernetes-operator
>> > > > > > > <https://github.com/apache/flink-kubernetes-operator>
>> project.
>> > > > > > >
>> > > > > > > The last few weeks we have been working very hard with the
>> > > community
>> > > > to
>> > > > > > > stabilize the initial feature set and I think we have made
>> great
>> > > > > > progress.
>> > > > > > > While we are still far from a production ready-state, a
>> preview
>> > > > release
>> > > > > > > will give us the opportunity to reach more people and gather
>> much
>> > > > > needed
>> > > > > > > input to take this project to the next level.
>> > > > > > >
>> > > > > > > There are still a couple missing features that we need to iron
>> > out
>> > > > and
>> > > > > we
>> > > > > > > need to make sure we have proper documentation but after that
>> I
>> > > think
>> > > > > it
>> > > > > > > would be a good time for the preview release.
>> > > > > > >
>> > > > > > > I propose to aim for the first release candidate around the
>> > 25-27th
>> > > > of
>> > > > > > > March after which we should dedicate a few days for some
>> > extensive
>> > > > > > testing
>> > > > > > > and bugfixing.
>> > > > > > >
>> > > > > > > What do you think?
>> > > > > > >
>> > > > > > > Gyula
>> > > > > > >
>> > > > > >
>> > > > >
>> > > > >
>> > > > > --
>> > > > >
>> > > > > Konstantin Knauf
>> > > > >
>> > > > > https://twitter.com/snntrable
>> > > > >
>> > > > > https://github.com/knaufk
>> > > > >
>> > > >
>> > >
>> >
>>
>
Re: [DISCUSS] Preview release for Flink Kubernetes Operator
Posted by Aitozi <gj...@gmail.com>.
Hi all
Thanks for starting the discussion. The first release at the end of
March looks good to me, I think the flink-kubernetes-operator is in a good
shape now. I will try to complete the design to support the session job in
these two days and will start another discussion before the work.
Regarding what has to be improved before release, I think the operator
metrics are not completed yet, we may have to inject more useful metrics.
The feature list
https://github.com/bgeng777/flink-kubernetes-operator/blob/features/doc/features.md
posted
by @Biao Geng looks good, Maybe we could include this to track the
supported features.
Best,
Aitozi.
Gyula Fóra <gy...@gmail.com> 于2022年3月14日周一 16:58写道：
> Hi Xintong!
>
> Thank you for the valuable input, you are completely right we need to agree
> and document these aspects.
>
> Let me try to address some of the questions and others should chip in also
> :)
>
> 1. Version convention:
>
> I think we should adopt the 3 digit versioning scheme like other flink
> projects (I think flink-shaded is a bit of an outlier here).
> The preview release should be 0.1.0.
>
> The supported Flink versions should be documented with the release, each
> version of the operator should technically be able to support multiple
> Flink versions at the same time.
> For the preview release this should be Flink version >= 1.14. We should
> later come up with a guarantee that we do not drop flink version support
> within the same minor version.
>
> The only public API here at the moment are the custom resource definitions.
> We agreed to mark them all experimental for the preview release, and I
> think for 1.0.0 we should aim for public evolving.
> Otherwise we should respect the Flink guarantees here. Practically this
> means that for the preview release we do not guarantee anything in terms of
> later compatibility.
>
> 2. Release process:
> I agree that we should follow the official Flink process as much as
> possible (and makes sense) with proper voting etc.
>
> Cheers,
> Gyula
>
>
> On Mon, Mar 14, 2022 at 8:32 AM Xintong Song <to...@gmail.com>
> wrote:
>
> > How everyone,
> >
> > It's great to learn that we are approaching a preview release for
> > the flink-kubernetes-operator. Thanks for the efforts.
> >
> > I have not been involved with any developing efforts in
> > flink-kubernetes-operator, thus have no comment on end of March being the
> > targeting date.
> >
> > However, based on my experiences being one of the Flink release
> managers, I
> > see a few things that are still missing for creating an official release.
> > (IIUC, the preview release is still an official release, just with weak
> > functionality and compatibility guarantees.)
> >
> > 1. The version conventions:
> > - How does a flink-kubernetes-operator version look like? E.g., flink /
> > flink-statefun / flink-ml have three digits x.y.z, while flink-shaded has
> > only two digits x.y.
> > - What is the relationship between flink-kubernetes-operator and flink
> > versions? E.g., flink-shaded x.* is only designed to support flink *.x.*.
> > - What kind of compatibility guarantees do we provide? E.g., flink
> expects
> > no Public API compatibility should be broken between minor releases (the
> > 2nd digit) and no PublicEvolving APIs should be broken between bugfix
> > releases (the 3rd digit).
> > - What kind of support do we provide for old releases? E.g., flink
> provides
> > bug fixes for the latest two minor releases (the 2nd digit).
> >
> > 2. Release process
> > You may find the release process for all Flink artifacts in this wiki
> page
> > [1]. Such a formal documented process would help us to reach consensus on
> > what needs to be done and make sure it complies with the ASF regulations
> > before creating a release. We probably don't need something as formal as
> a
> > vote to approve the release process. But we definitely need a formal vote
> > for the flink-kubernetes-operator release, and the release process would
> > help making sure we are on the same page about what is a releasable state
> > for this artifact.
> >
> > WDYT?
> >
> > Thank you~
> >
> > Xintong Song
> >
> >
> > [1] https://cwiki.apache.org/confluence/display/FLINK/Releasing
> >
> > On Mon, Mar 14, 2022 at 11:34 AM Biao Geng <bi...@gmail.com> wrote:
> >
> > > Hi there,
> > >
> > > It is exciting to see the discussion of the release timeline! I agree
> > that
> > > the end of March is a proper date.
> > > To make others easier get involved in this discussion, I think we may
> > need
> > > to provide a more straightforward feature list for the preview release.
> > The
> > > "Initial Feature Set" in FLIP-212
> > > <
> > >
> >
> https://cwiki.apache.org/confluence/display/FLINK/FLIP-212:+Introduce+Flink+Kubernetes+Operator
> > > >
> > > is
> > > almost complete. But some new features like webhook based validate and
> > > flink operator metric are not added and they are only tracked in the
> long
> > > JIRA list. If we can update the FLIP, it may be more convenient and can
> > > also help us write release notes later. I also created a draft
> > > <
> > >
> >
> https://github.com/bgeng777/flink-kubernetes-operator/blob/features/doc/features.md
> > > >
> > > for myself to track completed or in-plan features. Hope it can help.
> > >
> > > Best,
> > > Biao Geng
> > >
> > > Gyula Fóra <gy...@gmail.com> 于2022年3月14日周一 04:11写道：
> > >
> > > > @Konstantin: Yes I completely agree that for this release the API
> (CRD)
> > > > should be marked experimental!
> > > > I have opened a ticket to track this:
> > > > https://issues.apache.org/jira/browse/FLINK-26620
> > > >
> > > > @Yang Wang <da...@gmail.com> : I think we still have plenty of
> > > time
> > > > to work on features like the session job before the release, would be
> > > nice
> > > > to provide a complete story to the users.
> > > >
> > > > Gyula
> > > >
> > > > On Sun, Mar 13, 2022 at 5:17 PM Konstantin Knauf <kn...@apache.org>
> > > > wrote:
> > > >
> > > > > Hi everyone,
> > > > >
> > > > > can we mark all the APIs as experimental/alpha so that it is clear
> > that
> > > > > these can be broken in future releases for now? I think this would
> be
> > > > very
> > > > > important given the early stage of the project. We want to be able
> to
> > > > > address shortcomings without worrying too much about backwards
> > > > > compatibility at this stage, I believe.
> > > > >
> > > > > Cheers,
> > > > >
> > > > > Konstantin
> > > > >
> > > > > On Sun, Mar 13, 2022 at 7:48 AM Yang Wang <da...@gmail.com>
> > > wrote:
> > > > >
> > > > > > Thanks Gyula for starting this discussion.
> > > > > >
> > > > > > Given that the core functionality is closing to stable, I am in
> > favor
> > > > of
> > > > > > having the MVP release at the end of March.
> > > > > > The first release will help us to collect more feedbacks from the
> > > > users.
> > > > > > Also it is a good chance to let the users know that the community
> > is
> > > > > trying
> > > > > > to maintain an official Kubernetes operator :)
> > > > > > I hope that the companies could build their own production
> > streaming
> > > > > > platform on top of the flink-kubernetes-operator in the future.
> > > > > >
> > > > > > FYI: @Wenjun Min is still working hard on supporting the Session
> > Job
> > > in
> > > > > > Flink Kubernetes operator, It will be great if we could include
> it
> > in
> > > > the
> > > > > > first release.
> > > > > > And I believe we have enough time.
> > > > > >
> > > > > > Moreover, I agree with you that we need to invest more time in
> the
> > > > > > documentation, e2e tests, helm install optimization, logging,
> > > > > > etc. before the release.
> > > > > >
> > > > > >
> > > > > > Best,
> > > > > > Yang
> > > > > >
> > > > > >
> > > > > > Gyula Fóra <gy...@apache.org> 于2022年3月12日周六 01:10写道：
> > > > > >
> > > > > > > Hi Team!
> > > > > > >
> > > > > > > I would like to discuss the timeline for the initial
> > > > preview/milestone
> > > > > > > release of the flink-kubernetes-operator
> > > > > > > <https://github.com/apache/flink-kubernetes-operator> project.
> > > > > > >
> > > > > > > The last few weeks we have been working very hard with the
> > > community
> > > > to
> > > > > > > stabilize the initial feature set and I think we have made
> great
> > > > > > progress.
> > > > > > > While we are still far from a production ready-state, a preview
> > > > release
> > > > > > > will give us the opportunity to reach more people and gather
> much
> > > > > needed
> > > > > > > input to take this project to the next level.
> > > > > > >
> > > > > > > There are still a couple missing features that we need to iron
> > out
> > > > and
> > > > > we
> > > > > > > need to make sure we have proper documentation but after that I
> > > think
> > > > > it
> > > > > > > would be a good time for the preview release.
> > > > > > >
> > > > > > > I propose to aim for the first release candidate around the
> > 25-27th
> > > > of
> > > > > > > March after which we should dedicate a few days for some
> > extensive
> > > > > > testing
> > > > > > > and bugfixing.
> > > > > > >
> > > > > > > What do you think?
> > > > > > >
> > > > > > > Gyula
> > > > > > >
> > > > > >
> > > > >
> > > > >
> > > > > --
> > > > >
> > > > > Konstantin Knauf
> > > > >
> > > > > https://twitter.com/snntrable
> > > > >
> > > > > https://github.com/knaufk
> > > > >
> > > >
> > >
> >
>
Re: [DISCUSS] Preview release for Flink Kubernetes Operator
Posted by Xintong Song <to...@gmail.com>.
Hi Gyula,
Thanks for the quick response. And your answers to the questions sound good
to me.
I'd suggest first creating a "Creating a Flink Kubernetes Operator Release"
page and a "Verifying a Flink Kubernetes Operator Release" page on the
wiki[1], just like other projects. We can then discuss / comment around the
docs and keep refining them.
Thank you~
Xintong Song
[1] https://cwiki.apache.org/confluence/display/FLINK/Releasing
On Mon, Mar 14, 2022 at 4:52 PM Gyula Fóra <gy...@gmail.com> wrote:
> Hi Xintong!
>
> Thank you for the valuable input, you are completely right we need to agree
> and document these aspects.
>
> Let me try to address some of the questions and others should chip in also
> :)
>
> 1. Version convention:
>
> I think we should adopt the 3 digit versioning scheme like other flink
> projects (I think flink-shaded is a bit of an outlier here).
> The preview release should be 0.1.0.
>
> The supported Flink versions should be documented with the release, each
> version of the operator should technically be able to support multiple
> Flink versions at the same time.
> For the preview release this should be Flink version >= 1.14. We should
> later come up with a guarantee that we do not drop flink version support
> within the same minor version.
>
> The only public API here at the moment are the custom resource definitions.
> We agreed to mark them all experimental for the preview release, and I
> think for 1.0.0 we should aim for public evolving.
> Otherwise we should respect the Flink guarantees here. Practically this
> means that for the preview release we do not guarantee anything in terms of
> later compatibility.
>
> 2. Release process:
> I agree that we should follow the official Flink process as much as
> possible (and makes sense) with proper voting etc.
>
> Cheers,
> Gyula
>
>
> On Mon, Mar 14, 2022 at 8:32 AM Xintong Song <to...@gmail.com>
> wrote:
>
> > How everyone,
> >
> > It's great to learn that we are approaching a preview release for
> > the flink-kubernetes-operator. Thanks for the efforts.
> >
> > I have not been involved with any developing efforts in
> > flink-kubernetes-operator, thus have no comment on end of March being the
> > targeting date.
> >
> > However, based on my experiences being one of the Flink release
> managers, I
> > see a few things that are still missing for creating an official release.
> > (IIUC, the preview release is still an official release, just with weak
> > functionality and compatibility guarantees.)
> >
> > 1. The version conventions:
> > - How does a flink-kubernetes-operator version look like? E.g., flink /
> > flink-statefun / flink-ml have three digits x.y.z, while flink-shaded has
> > only two digits x.y.
> > - What is the relationship between flink-kubernetes-operator and flink
> > versions? E.g., flink-shaded x.* is only designed to support flink *.x.*.
> > - What kind of compatibility guarantees do we provide? E.g., flink
> expects
> > no Public API compatibility should be broken between minor releases (the
> > 2nd digit) and no PublicEvolving APIs should be broken between bugfix
> > releases (the 3rd digit).
> > - What kind of support do we provide for old releases? E.g., flink
> provides
> > bug fixes for the latest two minor releases (the 2nd digit).
> >
> > 2. Release process
> > You may find the release process for all Flink artifacts in this wiki
> page
> > [1]. Such a formal documented process would help us to reach consensus on
> > what needs to be done and make sure it complies with the ASF regulations
> > before creating a release. We probably don't need something as formal as
> a
> > vote to approve the release process. But we definitely need a formal vote
> > for the flink-kubernetes-operator release, and the release process would
> > help making sure we are on the same page about what is a releasable state
> > for this artifact.
> >
> > WDYT?
> >
> > Thank you~
> >
> > Xintong Song
> >
> >
> > [1] https://cwiki.apache.org/confluence/display/FLINK/Releasing
> >
> > On Mon, Mar 14, 2022 at 11:34 AM Biao Geng <bi...@gmail.com> wrote:
> >
> > > Hi there,
> > >
> > > It is exciting to see the discussion of the release timeline! I agree
> > that
> > > the end of March is a proper date.
> > > To make others easier get involved in this discussion, I think we may
> > need
> > > to provide a more straightforward feature list for the preview release.
> > The
> > > "Initial Feature Set" in FLIP-212
> > > <
> > >
> >
> https://cwiki.apache.org/confluence/display/FLINK/FLIP-212:+Introduce+Flink+Kubernetes+Operator
> > > >
> > > is
> > > almost complete. But some new features like webhook based validate and
> > > flink operator metric are not added and they are only tracked in the
> long
> > > JIRA list. If we can update the FLIP, it may be more convenient and can
> > > also help us write release notes later. I also created a draft
> > > <
> > >
> >
> https://github.com/bgeng777/flink-kubernetes-operator/blob/features/doc/features.md
> > > >
> > > for myself to track completed or in-plan features. Hope it can help.
> > >
> > > Best,
> > > Biao Geng
> > >
> > > Gyula Fóra <gy...@gmail.com> 于2022年3月14日周一 04:11写道：
> > >
> > > > @Konstantin: Yes I completely agree that for this release the API
> (CRD)
> > > > should be marked experimental!
> > > > I have opened a ticket to track this:
> > > > https://issues.apache.org/jira/browse/FLINK-26620
> > > >
> > > > @Yang Wang <da...@gmail.com> : I think we still have plenty of
> > > time
> > > > to work on features like the session job before the release, would be
> > > nice
> > > > to provide a complete story to the users.
> > > >
> > > > Gyula
> > > >
> > > > On Sun, Mar 13, 2022 at 5:17 PM Konstantin Knauf <kn...@apache.org>
> > > > wrote:
> > > >
> > > > > Hi everyone,
> > > > >
> > > > > can we mark all the APIs as experimental/alpha so that it is clear
> > that
> > > > > these can be broken in future releases for now? I think this would
> be
> > > > very
> > > > > important given the early stage of the project. We want to be able
> to
> > > > > address shortcomings without worrying too much about backwards
> > > > > compatibility at this stage, I believe.
> > > > >
> > > > > Cheers,
> > > > >
> > > > > Konstantin
> > > > >
> > > > > On Sun, Mar 13, 2022 at 7:48 AM Yang Wang <da...@gmail.com>
> > > wrote:
> > > > >
> > > > > > Thanks Gyula for starting this discussion.
> > > > > >
> > > > > > Given that the core functionality is closing to stable, I am in
> > favor
> > > > of
> > > > > > having the MVP release at the end of March.
> > > > > > The first release will help us to collect more feedbacks from the
> > > > users.
> > > > > > Also it is a good chance to let the users know that the community
> > is
> > > > > trying
> > > > > > to maintain an official Kubernetes operator :)
> > > > > > I hope that the companies could build their own production
> > streaming
> > > > > > platform on top of the flink-kubernetes-operator in the future.
> > > > > >
> > > > > > FYI: @Wenjun Min is still working hard on supporting the Session
> > Job
> > > in
> > > > > > Flink Kubernetes operator, It will be great if we could include
> it
> > in
> > > > the
> > > > > > first release.
> > > > > > And I believe we have enough time.
> > > > > >
> > > > > > Moreover, I agree with you that we need to invest more time in
> the
> > > > > > documentation, e2e tests, helm install optimization, logging,
> > > > > > etc. before the release.
> > > > > >
> > > > > >
> > > > > > Best,
> > > > > > Yang
> > > > > >
> > > > > >
> > > > > > Gyula Fóra <gy...@apache.org> 于2022年3月12日周六 01:10写道：
> > > > > >
> > > > > > > Hi Team!
> > > > > > >
> > > > > > > I would like to discuss the timeline for the initial
> > > > preview/milestone
> > > > > > > release of the flink-kubernetes-operator
> > > > > > > <https://github.com/apache/flink-kubernetes-operator> project.
> > > > > > >
> > > > > > > The last few weeks we have been working very hard with the
> > > community
> > > > to
> > > > > > > stabilize the initial feature set and I think we have made
> great
> > > > > > progress.
> > > > > > > While we are still far from a production ready-state, a preview
> > > > release
> > > > > > > will give us the opportunity to reach more people and gather
> much
> > > > > needed
> > > > > > > input to take this project to the next level.
> > > > > > >
> > > > > > > There are still a couple missing features that we need to iron
> > out
> > > > and
> > > > > we
> > > > > > > need to make sure we have proper documentation but after that I
> > > think
> > > > > it
> > > > > > > would be a good time for the preview release.
> > > > > > >
> > > > > > > I propose to aim for the first release candidate around the
> > 25-27th
> > > > of
> > > > > > > March after which we should dedicate a few days for some
> > extensive
> > > > > > testing
> > > > > > > and bugfixing.
> > > > > > >
> > > > > > > What do you think?
> > > > > > >
> > > > > > > Gyula
> > > > > > >
> > > > > >
> > > > >
> > > > >
> > > > > --
> > > > >
> > > > > Konstantin Knauf
> > > > >
> > > > > https://twitter.com/snntrable
> > > > >
> > > > > https://github.com/knaufk
> > > > >
> > > >
> > >
> >
>
Re: [DISCUSS] Preview release for Flink Kubernetes Operator
Posted by Gyula Fóra <gy...@gmail.com>.
Hi Xintong!
Thank you for the valuable input, you are completely right we need to agree
and document these aspects.
Let me try to address some of the questions and others should chip in also
:)
1. Version convention:
I think we should adopt the 3 digit versioning scheme like other flink
projects (I think flink-shaded is a bit of an outlier here).
The preview release should be 0.1.0.
The supported Flink versions should be documented with the release, each
version of the operator should technically be able to support multiple
Flink versions at the same time.
For the preview release this should be Flink version >= 1.14. We should
later come up with a guarantee that we do not drop flink version support
within the same minor version.
The only public API here at the moment are the custom resource definitions.
We agreed to mark them all experimental for the preview release, and I
think for 1.0.0 we should aim for public evolving.
Otherwise we should respect the Flink guarantees here. Practically this
means that for the preview release we do not guarantee anything in terms of
later compatibility.
2. Release process:
I agree that we should follow the official Flink process as much as
possible (and makes sense) with proper voting etc.
Cheers,
Gyula
On Mon, Mar 14, 2022 at 8:32 AM Xintong Song <to...@gmail.com> wrote:
> How everyone,
>
> It's great to learn that we are approaching a preview release for
> the flink-kubernetes-operator. Thanks for the efforts.
>
> I have not been involved with any developing efforts in
> flink-kubernetes-operator, thus have no comment on end of March being the
> targeting date.
>
> However, based on my experiences being one of the Flink release managers, I
> see a few things that are still missing for creating an official release.
> (IIUC, the preview release is still an official release, just with weak
> functionality and compatibility guarantees.)
>
> 1. The version conventions:
> - How does a flink-kubernetes-operator version look like? E.g., flink /
> flink-statefun / flink-ml have three digits x.y.z, while flink-shaded has
> only two digits x.y.
> - What is the relationship between flink-kubernetes-operator and flink
> versions? E.g., flink-shaded x.* is only designed to support flink *.x.*.
> - What kind of compatibility guarantees do we provide? E.g., flink expects
> no Public API compatibility should be broken between minor releases (the
> 2nd digit) and no PublicEvolving APIs should be broken between bugfix
> releases (the 3rd digit).
> - What kind of support do we provide for old releases? E.g., flink provides
> bug fixes for the latest two minor releases (the 2nd digit).
>
> 2. Release process
> You may find the release process for all Flink artifacts in this wiki page
> [1]. Such a formal documented process would help us to reach consensus on
> what needs to be done and make sure it complies with the ASF regulations
> before creating a release. We probably don't need something as formal as a
> vote to approve the release process. But we definitely need a formal vote
> for the flink-kubernetes-operator release, and the release process would
> help making sure we are on the same page about what is a releasable state
> for this artifact.
>
> WDYT?
>
> Thank you~
>
> Xintong Song
>
>
> [1] https://cwiki.apache.org/confluence/display/FLINK/Releasing
>
> On Mon, Mar 14, 2022 at 11:34 AM Biao Geng <bi...@gmail.com> wrote:
>
> > Hi there,
> >
> > It is exciting to see the discussion of the release timeline! I agree
> that
> > the end of March is a proper date.
> > To make others easier get involved in this discussion, I think we may
> need
> > to provide a more straightforward feature list for the preview release.
> The
> > "Initial Feature Set" in FLIP-212
> > <
> >
> https://cwiki.apache.org/confluence/display/FLINK/FLIP-212:+Introduce+Flink+Kubernetes+Operator
> > >
> > is
> > almost complete. But some new features like webhook based validate and
> > flink operator metric are not added and they are only tracked in the long
> > JIRA list. If we can update the FLIP, it may be more convenient and can
> > also help us write release notes later. I also created a draft
> > <
> >
> https://github.com/bgeng777/flink-kubernetes-operator/blob/features/doc/features.md
> > >
> > for myself to track completed or in-plan features. Hope it can help.
> >
> > Best,
> > Biao Geng
> >
> > Gyula Fóra <gy...@gmail.com> 于2022年3月14日周一 04:11写道：
> >
> > > @Konstantin: Yes I completely agree that for this release the API (CRD)
> > > should be marked experimental!
> > > I have opened a ticket to track this:
> > > https://issues.apache.org/jira/browse/FLINK-26620
> > >
> > > @Yang Wang <da...@gmail.com> : I think we still have plenty of
> > time
> > > to work on features like the session job before the release, would be
> > nice
> > > to provide a complete story to the users.
> > >
> > > Gyula
> > >
> > > On Sun, Mar 13, 2022 at 5:17 PM Konstantin Knauf <kn...@apache.org>
> > > wrote:
> > >
> > > > Hi everyone,
> > > >
> > > > can we mark all the APIs as experimental/alpha so that it is clear
> that
> > > > these can be broken in future releases for now? I think this would be
> > > very
> > > > important given the early stage of the project. We want to be able to
> > > > address shortcomings without worrying too much about backwards
> > > > compatibility at this stage, I believe.
> > > >
> > > > Cheers,
> > > >
> > > > Konstantin
> > > >
> > > > On Sun, Mar 13, 2022 at 7:48 AM Yang Wang <da...@gmail.com>
> > wrote:
> > > >
> > > > > Thanks Gyula for starting this discussion.
> > > > >
> > > > > Given that the core functionality is closing to stable, I am in
> favor
> > > of
> > > > > having the MVP release at the end of March.
> > > > > The first release will help us to collect more feedbacks from the
> > > users.
> > > > > Also it is a good chance to let the users know that the community
> is
> > > > trying
> > > > > to maintain an official Kubernetes operator :)
> > > > > I hope that the companies could build their own production
> streaming
> > > > > platform on top of the flink-kubernetes-operator in the future.
> > > > >
> > > > > FYI: @Wenjun Min is still working hard on supporting the Session
> Job
> > in
> > > > > Flink Kubernetes operator, It will be great if we could include it
> in
> > > the
> > > > > first release.
> > > > > And I believe we have enough time.
> > > > >
> > > > > Moreover, I agree with you that we need to invest more time in the
> > > > > documentation, e2e tests, helm install optimization, logging,
> > > > > etc. before the release.
> > > > >
> > > > >
> > > > > Best,
> > > > > Yang
> > > > >
> > > > >
> > > > > Gyula Fóra <gy...@apache.org> 于2022年3月12日周六 01:10写道：
> > > > >
> > > > > > Hi Team!
> > > > > >
> > > > > > I would like to discuss the timeline for the initial
> > > preview/milestone
> > > > > > release of the flink-kubernetes-operator
> > > > > > <https://github.com/apache/flink-kubernetes-operator> project.
> > > > > >
> > > > > > The last few weeks we have been working very hard with the
> > community
> > > to
> > > > > > stabilize the initial feature set and I think we have made great
> > > > > progress.
> > > > > > While we are still far from a production ready-state, a preview
> > > release
> > > > > > will give us the opportunity to reach more people and gather much
> > > > needed
> > > > > > input to take this project to the next level.
> > > > > >
> > > > > > There are still a couple missing features that we need to iron
> out
> > > and
> > > > we
> > > > > > need to make sure we have proper documentation but after that I
> > think
> > > > it
> > > > > > would be a good time for the preview release.
> > > > > >
> > > > > > I propose to aim for the first release candidate around the
> 25-27th
> > > of
> > > > > > March after which we should dedicate a few days for some
> extensive
> > > > > testing
> > > > > > and bugfixing.
> > > > > >
> > > > > > What do you think?
> > > > > >
> > > > > > Gyula
> > > > > >
> > > > >
> > > >
> > > >
> > > > --
> > > >
> > > > Konstantin Knauf
> > > >
> > > > https://twitter.com/snntrable
> > > >
> > > > https://github.com/knaufk
> > > >
> > >
> >
>
Re: [DISCUSS] Preview release for Flink Kubernetes Operator
Posted by Xintong Song <to...@gmail.com>.
How everyone,
It's great to learn that we are approaching a preview release for
the flink-kubernetes-operator. Thanks for the efforts.
I have not been involved with any developing efforts in
flink-kubernetes-operator, thus have no comment on end of March being the
targeting date.
However, based on my experiences being one of the Flink release managers, I
see a few things that are still missing for creating an official release.
(IIUC, the preview release is still an official release, just with weak
functionality and compatibility guarantees.)
1. The version conventions:
- How does a flink-kubernetes-operator version look like? E.g., flink /
flink-statefun / flink-ml have three digits x.y.z, while flink-shaded has
only two digits x.y.
- What is the relationship between flink-kubernetes-operator and flink
versions? E.g., flink-shaded x.* is only designed to support flink *.x.*.
- What kind of compatibility guarantees do we provide? E.g., flink expects
no Public API compatibility should be broken between minor releases (the
2nd digit) and no PublicEvolving APIs should be broken between bugfix
releases (the 3rd digit).
- What kind of support do we provide for old releases? E.g., flink provides
bug fixes for the latest two minor releases (the 2nd digit).
2. Release process
You may find the release process for all Flink artifacts in this wiki page
[1]. Such a formal documented process would help us to reach consensus on
what needs to be done and make sure it complies with the ASF regulations
before creating a release. We probably don't need something as formal as a
vote to approve the release process. But we definitely need a formal vote
for the flink-kubernetes-operator release, and the release process would
help making sure we are on the same page about what is a releasable state
for this artifact.
WDYT?
Thank you~
Xintong Song
[1] https://cwiki.apache.org/confluence/display/FLINK/Releasing
On Mon, Mar 14, 2022 at 11:34 AM Biao Geng <bi...@gmail.com> wrote:
> Hi there,
>
> It is exciting to see the discussion of the release timeline! I agree that
> the end of March is a proper date.
> To make others easier get involved in this discussion, I think we may need
> to provide a more straightforward feature list for the preview release. The
> "Initial Feature Set" in FLIP-212
> <
> https://cwiki.apache.org/confluence/display/FLINK/FLIP-212:+Introduce+Flink+Kubernetes+Operator
> >
> is
> almost complete. But some new features like webhook based validate and
> flink operator metric are not added and they are only tracked in the long
> JIRA list. If we can update the FLIP, it may be more convenient and can
> also help us write release notes later. I also created a draft
> <
> https://github.com/bgeng777/flink-kubernetes-operator/blob/features/doc/features.md
> >
> for myself to track completed or in-plan features. Hope it can help.
>
> Best,
> Biao Geng
>
> Gyula Fóra <gy...@gmail.com> 于2022年3月14日周一 04:11写道：
>
> > @Konstantin: Yes I completely agree that for this release the API (CRD)
> > should be marked experimental!
> > I have opened a ticket to track this:
> > https://issues.apache.org/jira/browse/FLINK-26620
> >
> > @Yang Wang <da...@gmail.com> : I think we still have plenty of
> time
> > to work on features like the session job before the release, would be
> nice
> > to provide a complete story to the users.
> >
> > Gyula
> >
> > On Sun, Mar 13, 2022 at 5:17 PM Konstantin Knauf <kn...@apache.org>
> > wrote:
> >
> > > Hi everyone,
> > >
> > > can we mark all the APIs as experimental/alpha so that it is clear that
> > > these can be broken in future releases for now? I think this would be
> > very
> > > important given the early stage of the project. We want to be able to
> > > address shortcomings without worrying too much about backwards
> > > compatibility at this stage, I believe.
> > >
> > > Cheers,
> > >
> > > Konstantin
> > >
> > > On Sun, Mar 13, 2022 at 7:48 AM Yang Wang <da...@gmail.com>
> wrote:
> > >
> > > > Thanks Gyula for starting this discussion.
> > > >
> > > > Given that the core functionality is closing to stable, I am in favor
> > of
> > > > having the MVP release at the end of March.
> > > > The first release will help us to collect more feedbacks from the
> > users.
> > > > Also it is a good chance to let the users know that the community is
> > > trying
> > > > to maintain an official Kubernetes operator :)
> > > > I hope that the companies could build their own production streaming
> > > > platform on top of the flink-kubernetes-operator in the future.
> > > >
> > > > FYI: @Wenjun Min is still working hard on supporting the Session Job
> in
> > > > Flink Kubernetes operator, It will be great if we could include it in
> > the
> > > > first release.
> > > > And I believe we have enough time.
> > > >
> > > > Moreover, I agree with you that we need to invest more time in the
> > > > documentation, e2e tests, helm install optimization, logging,
> > > > etc. before the release.
> > > >
> > > >
> > > > Best,
> > > > Yang
> > > >
> > > >
> > > > Gyula Fóra <gy...@apache.org> 于2022年3月12日周六 01:10写道：
> > > >
> > > > > Hi Team!
> > > > >
> > > > > I would like to discuss the timeline for the initial
> > preview/milestone
> > > > > release of the flink-kubernetes-operator
> > > > > <https://github.com/apache/flink-kubernetes-operator> project.
> > > > >
> > > > > The last few weeks we have been working very hard with the
> community
> > to
> > > > > stabilize the initial feature set and I think we have made great
> > > > progress.
> > > > > While we are still far from a production ready-state, a preview
> > release
> > > > > will give us the opportunity to reach more people and gather much
> > > needed
> > > > > input to take this project to the next level.
> > > > >
> > > > > There are still a couple missing features that we need to iron out
> > and
> > > we
> > > > > need to make sure we have proper documentation but after that I
> think
> > > it
> > > > > would be a good time for the preview release.
> > > > >
> > > > > I propose to aim for the first release candidate around the 25-27th
> > of
> > > > > March after which we should dedicate a few days for some extensive
> > > > testing
> > > > > and bugfixing.
> > > > >
> > > > > What do you think?
> > > > >
> > > > > Gyula
> > > > >
> > > >
> > >
> > >
> > > --
> > >
> > > Konstantin Knauf
> > >
> > > https://twitter.com/snntrable
> > >
> > > https://github.com/knaufk
> > >
> >
>
Re: [DISCUSS] Preview release for Flink Kubernetes Operator
Posted by Biao Geng <bi...@gmail.com>.
Hi there,
It is exciting to see the discussion of the release timeline! I agree that
the end of March is a proper date.
To make others easier get involved in this discussion, I think we may need
to provide a more straightforward feature list for the preview release. The
"Initial Feature Set" in FLIP-212
<https://cwiki.apache.org/confluence/display/FLINK/FLIP-212:+Introduce+Flink+Kubernetes+Operator>
is
almost complete. But some new features like webhook based validate and
flink operator metric are not added and they are only tracked in the long
JIRA list. If we can update the FLIP, it may be more convenient and can
also help us write release notes later. I also created a draft
<https://github.com/bgeng777/flink-kubernetes-operator/blob/features/doc/features.md>
for myself to track completed or in-plan features. Hope it can help.
Best,
Biao Geng
Gyula Fóra <gy...@gmail.com> 于2022年3月14日周一 04:11写道：
> @Konstantin: Yes I completely agree that for this release the API (CRD)
> should be marked experimental!
> I have opened a ticket to track this:
> https://issues.apache.org/jira/browse/FLINK-26620
>
> @Yang Wang <da...@gmail.com> : I think we still have plenty of time
> to work on features like the session job before the release, would be nice
> to provide a complete story to the users.
>
> Gyula
>
> On Sun, Mar 13, 2022 at 5:17 PM Konstantin Knauf <kn...@apache.org>
> wrote:
>
> > Hi everyone,
> >
> > can we mark all the APIs as experimental/alpha so that it is clear that
> > these can be broken in future releases for now? I think this would be
> very
> > important given the early stage of the project. We want to be able to
> > address shortcomings without worrying too much about backwards
> > compatibility at this stage, I believe.
> >
> > Cheers,
> >
> > Konstantin
> >
> > On Sun, Mar 13, 2022 at 7:48 AM Yang Wang <da...@gmail.com> wrote:
> >
> > > Thanks Gyula for starting this discussion.
> > >
> > > Given that the core functionality is closing to stable, I am in favor
> of
> > > having the MVP release at the end of March.
> > > The first release will help us to collect more feedbacks from the
> users.
> > > Also it is a good chance to let the users know that the community is
> > trying
> > > to maintain an official Kubernetes operator :)
> > > I hope that the companies could build their own production streaming
> > > platform on top of the flink-kubernetes-operator in the future.
> > >
> > > FYI: @Wenjun Min is still working hard on supporting the Session Job in
> > > Flink Kubernetes operator, It will be great if we could include it in
> the
> > > first release.
> > > And I believe we have enough time.
> > >
> > > Moreover, I agree with you that we need to invest more time in the
> > > documentation, e2e tests, helm install optimization, logging,
> > > etc. before the release.
> > >
> > >
> > > Best,
> > > Yang
> > >
> > >
> > > Gyula Fóra <gy...@apache.org> 于2022年3月12日周六 01:10写道：
> > >
> > > > Hi Team!
> > > >
> > > > I would like to discuss the timeline for the initial
> preview/milestone
> > > > release of the flink-kubernetes-operator
> > > > <https://github.com/apache/flink-kubernetes-operator> project.
> > > >
> > > > The last few weeks we have been working very hard with the community
> to
> > > > stabilize the initial feature set and I think we have made great
> > > progress.
> > > > While we are still far from a production ready-state, a preview
> release
> > > > will give us the opportunity to reach more people and gather much
> > needed
> > > > input to take this project to the next level.
> > > >
> > > > There are still a couple missing features that we need to iron out
> and
> > we
> > > > need to make sure we have proper documentation but after that I think
> > it
> > > > would be a good time for the preview release.
> > > >
> > > > I propose to aim for the first release candidate around the 25-27th
> of
> > > > March after which we should dedicate a few days for some extensive
> > > testing
> > > > and bugfixing.
> > > >
> > > > What do you think?
> > > >
> > > > Gyula
> > > >
> > >
> >
> >
> > --
> >
> > Konstantin Knauf
> >
> > https://twitter.com/snntrable
> >
> > https://github.com/knaufk
> >
>
Re: [DISCUSS] Preview release for Flink Kubernetes Operator
Posted by Gyula Fóra <gy...@gmail.com>.
@Konstantin: Yes I completely agree that for this release the API (CRD)
should be marked experimental!
I have opened a ticket to track this:
https://issues.apache.org/jira/browse/FLINK-26620
@Yang Wang <da...@gmail.com> : I think we still have plenty of time
to work on features like the session job before the release, would be nice
to provide a complete story to the users.
Gyula
On Sun, Mar 13, 2022 at 5:17 PM Konstantin Knauf <kn...@apache.org> wrote:
> Hi everyone,
>
> can we mark all the APIs as experimental/alpha so that it is clear that
> these can be broken in future releases for now? I think this would be very
> important given the early stage of the project. We want to be able to
> address shortcomings without worrying too much about backwards
> compatibility at this stage, I believe.
>
> Cheers,
>
> Konstantin
>
> On Sun, Mar 13, 2022 at 7:48 AM Yang Wang <da...@gmail.com> wrote:
>
> > Thanks Gyula for starting this discussion.
> >
> > Given that the core functionality is closing to stable, I am in favor of
> > having the MVP release at the end of March.
> > The first release will help us to collect more feedbacks from the users.
> > Also it is a good chance to let the users know that the community is
> trying
> > to maintain an official Kubernetes operator :)
> > I hope that the companies could build their own production streaming
> > platform on top of the flink-kubernetes-operator in the future.
> >
> > FYI: @Wenjun Min is still working hard on supporting the Session Job in
> > Flink Kubernetes operator, It will be great if we could include it in the
> > first release.
> > And I believe we have enough time.
> >
> > Moreover, I agree with you that we need to invest more time in the
> > documentation, e2e tests, helm install optimization, logging,
> > etc. before the release.
> >
> >
> > Best,
> > Yang
> >
> >
> > Gyula Fóra <gy...@apache.org> 于2022年3月12日周六 01:10写道：
> >
> > > Hi Team!
> > >
> > > I would like to discuss the timeline for the initial preview/milestone
> > > release of the flink-kubernetes-operator
> > > <https://github.com/apache/flink-kubernetes-operator> project.
> > >
> > > The last few weeks we have been working very hard with the community to
> > > stabilize the initial feature set and I think we have made great
> > progress.
> > > While we are still far from a production ready-state, a preview release
> > > will give us the opportunity to reach more people and gather much
> needed
> > > input to take this project to the next level.
> > >
> > > There are still a couple missing features that we need to iron out and
> we
> > > need to make sure we have proper documentation but after that I think
> it
> > > would be a good time for the preview release.
> > >
> > > I propose to aim for the first release candidate around the 25-27th of
> > > March after which we should dedicate a few days for some extensive
> > testing
> > > and bugfixing.
> > >
> > > What do you think?
> > >
> > > Gyula
> > >
> >
>
>
> --
>
> Konstantin Knauf
>
> https://twitter.com/snntrable
>
> https://github.com/knaufk
>
Re: [DISCUSS] Preview release for Flink Kubernetes Operator
Posted by Konstantin Knauf <kn...@apache.org>.
Hi everyone,
can we mark all the APIs as experimental/alpha so that it is clear that
these can be broken in future releases for now? I think this would be very
important given the early stage of the project. We want to be able to
address shortcomings without worrying too much about backwards
compatibility at this stage, I believe.
Cheers,
Konstantin
On Sun, Mar 13, 2022 at 7:48 AM Yang Wang <da...@gmail.com> wrote:
> Thanks Gyula for starting this discussion.
>
> Given that the core functionality is closing to stable, I am in favor of
> having the MVP release at the end of March.
> The first release will help us to collect more feedbacks from the users.
> Also it is a good chance to let the users know that the community is trying
> to maintain an official Kubernetes operator :)
> I hope that the companies could build their own production streaming
> platform on top of the flink-kubernetes-operator in the future.
>
> FYI: @Wenjun Min is still working hard on supporting the Session Job in
> Flink Kubernetes operator, It will be great if we could include it in the
> first release.
> And I believe we have enough time.
>
> Moreover, I agree with you that we need to invest more time in the
> documentation, e2e tests, helm install optimization, logging,
> etc. before the release.
>
>
> Best,
> Yang
>
>
> Gyula Fóra <gy...@apache.org> 于2022年3月12日周六 01:10写道：
>
> > Hi Team!
> >
> > I would like to discuss the timeline for the initial preview/milestone
> > release of the flink-kubernetes-operator
> > <https://github.com/apache/flink-kubernetes-operator> project.
> >
> > The last few weeks we have been working very hard with the community to
> > stabilize the initial feature set and I think we have made great
> progress.
> > While we are still far from a production ready-state, a preview release
> > will give us the opportunity to reach more people and gather much needed
> > input to take this project to the next level.
> >
> > There are still a couple missing features that we need to iron out and we
> > need to make sure we have proper documentation but after that I think it
> > would be a good time for the preview release.
> >
> > I propose to aim for the first release candidate around the 25-27th of
> > March after which we should dedicate a few days for some extensive
> testing
> > and bugfixing.
> >
> > What do you think?
> >
> > Gyula
> >
>
--
Konstantin Knauf
https://twitter.com/snntrable
https://github.com/knaufk
Re: [DISCUSS] Preview release for Flink Kubernetes Operator
Posted by Yang Wang <da...@gmail.com>.
Thanks Gyula for starting this discussion.
Given that the core functionality is closing to stable, I am in favor of
having the MVP release at the end of March.
The first release will help us to collect more feedbacks from the users.
Also it is a good chance to let the users know that the community is trying
to maintain an official Kubernetes operator :)
I hope that the companies could build their own production streaming
platform on top of the flink-kubernetes-operator in the future.
FYI: @Wenjun Min is still working hard on supporting the Session Job in
Flink Kubernetes operator, It will be great if we could include it in the
first release.
And I believe we have enough time.
Moreover, I agree with you that we need to invest more time in the
documentation, e2e tests, helm install optimization, logging,
etc. before the release.
Best,
Yang
Gyula Fóra <gy...@apache.org> 于2022年3月12日周六 01:10写道：
> Hi Team!
>
> I would like to discuss the timeline for the initial preview/milestone
> release of the flink-kubernetes-operator
> <https://github.com/apache/flink-kubernetes-operator> project.
>
> The last few weeks we have been working very hard with the community to
> stabilize the initial feature set and I think we have made great progress.
> While we are still far from a production ready-state, a preview release
> will give us the opportunity to reach more people and gather much needed
> input to take this project to the next level.
>
> There are still a couple missing features that we need to iron out and we
> need to make sure we have proper documentation but after that I think it
> would be a good time for the preview release.
>
> I propose to aim for the first release candidate around the 25-27th of
> March after which we should dedicate a few days for some extensive testing
> and bugfixing.
>
> What do you think?
>
> Gyula
>