A proposal for a new scheduler

[Dominic Kim <st...@gmail.com>]

Dear whiskers.

I opened a PR to propose a new scheduler.

Please refer to this PR: https://github.com/apache/openwhisk/pull/4921

You may find these useful:
https://cwiki.apache.org/confluence/display/OPENWHISK/New+architecture+proposal
It is about design considerations and general architecture.

Since it's a bit big module, we would incrementally merge small PRs.
So we need to allow some incomplete modules at some point.
For example, some modules may not have their counterparts, and some modules
may include dummy codes as there is no relevant module yet, and so on.

But we will do our best to modularize them and add as many test codes as
possible.

Please look into it and share your thoughts.
Any comments would be appreciated.

Thank you.
Best regards.
Dominic

Re: A proposal for a new scheduler

[Dominic Kim <st...@gmail.com>]

Dear Whiskers.

I would merge the PR* in 24 hours based on the silent consent.
Please share your opinion if you have any.

Thanks in advance.
-dom

* [1] Add a proposal for a new scheduler:
https://github.com/apache/openwhisk/pull/4921

2020년 6월 30일 (화) 오후 11:49, Rodric Rabbah <ro...@gmail.com>님이 작성:

> Hi Dominic.
>
> I have read the POEM and provided some suggestions and posed some questions
> in the PR.
>
> In general, I support this direction. In fact, I think it will work well
> with the POEM I have been (slowly) working on and want to submit for
> containerless functions in support of edge use cases and using isolates.
>
> This is a substantial innovation to the project - thank you and
> @ningyougang,
> @keonhee, @jiangpengcheng, @upgle for contributing this work to the
> project.
>
> -r
>
>
> On Mon, Jun 29, 2020 at 11:42 PM Dominic Kim <st...@gmail.com> wrote:
>
> > Dear whiskers.
> >
> > I want to see if there is anyone with comments.
> > As per our guideline[1], I am supposed to proceed with a vote to let my
> PR
> > be included and start development.
> >
> > Before opening a vote, please share your opinion.
> > As discussed, we would incrementally merge small pieces of PRs while
> > allowing some transient incomplete codes such as some modules without
> > counterparts, some modules with dummy codes.
> > Even if this kind of code would be complete as more PRs are merged, I
> hope
> > we can form a consensus about this in advance.
> >
> >
> >
> > Thank you
> > Best regards
> > Dominic
> >
> > * [1] POEM guideline:
> >
> >
> https://github.com/apache/openwhisk/blob/master/proposals/POEM-1-proposal-for-openwhisk-enhancements.md#procedures
> >
>

Re: A proposal for a new scheduler

[Rodric Rabbah <ro...@gmail.com>]

Hi Dominic.

I have read the POEM and provided some suggestions and posed some questions
in the PR.

In general, I support this direction. In fact, I think it will work well
with the POEM I have been (slowly) working on and want to submit for
containerless functions in support of edge use cases and using isolates.

This is a substantial innovation to the project - thank you and @ningyougang,
@keonhee, @jiangpengcheng, @upgle for contributing this work to the project.

-r


On Mon, Jun 29, 2020 at 11:42 PM Dominic Kim <st...@gmail.com> wrote:

> Dear whiskers.
>
> I want to see if there is anyone with comments.
> As per our guideline[1], I am supposed to proceed with a vote to let my PR
> be included and start development.
>
> Before opening a vote, please share your opinion.
> As discussed, we would incrementally merge small pieces of PRs while
> allowing some transient incomplete codes such as some modules without
> counterparts, some modules with dummy codes.
> Even if this kind of code would be complete as more PRs are merged, I hope
> we can form a consensus about this in advance.
>
>
>
> Thank you
> Best regards
> Dominic
>
> * [1] POEM guideline:
>
> https://github.com/apache/openwhisk/blob/master/proposals/POEM-1-proposal-for-openwhisk-enhancements.md#procedures
>

Re: A proposal for a new scheduler

[Dominic Kim <st...@gmail.com>]

Dear whiskers.

I want to see if there is anyone with comments.
As per our guideline[1], I am supposed to proceed with a vote to let my PR
be included and start development.

Before opening a vote, please share your opinion.
As discussed, we would incrementally merge small pieces of PRs while
allowing some transient incomplete codes such as some modules without
counterparts, some modules with dummy codes.
Even if this kind of code would be complete as more PRs are merged, I hope
we can form a consensus about this in advance.



Thank you
Best regards
Dominic

* [1] POEM guideline:
https://github.com/apache/openwhisk/blob/master/proposals/POEM-1-proposal-for-openwhisk-enhancements.md#procedures