Hi WeiWei, All,

I create a JIRA: YARN-8670. And add some new ideas.
If anyone who have doubt or advice, please feel free to comment.

Thanks

Best Regards
Sichen Zhao

________________________________________
From: Weiwei Yang <cheersy...@hotmail.com>
Sent: Thursday, August 16, 2018 1:18:15 PM
To: Sichen Zhao; Sunil G; Yufei Gu
Cc: Daniel Templeton; Hadoop Common; Wangda Tan
Subject: Re: YARN SLS improving idea

Hi Sichen

Sure, please go ahead creating JIRA to capture your ideas, we can move 
discussion to smaller groups there.
At meanwhile, I am interested how you plan to support a large number of 
“specific” resource requests, look forward to seeing more details.
Looks like 2.b depends on YARN-3409 but it is not yet merged, maybe that can be 
another JIRA.

Thanks!
--
Weiwei

On 8/16/18, 12:06 PM, "Sichen Zhao" <zsc19940...@outlook.com> wrote:

    Hi all,
    Thanks for your advise.
    1. I will create JIRA once the details are identified.
    2. I read the patch YARN-8007. the scheduling request already done, but 
there are some difference between the patch and my ideas:
        a) the patch use the SYNTH as schedulingrequest input. But SYNTH can't 
create a large number of specific resource requests. I wanna create a new
            schedulingrequest input format for the more authentic input.
        b) the patch doesn't add the attribute for the Node. I think this can 
be done.

    Best Regards
    Sichen Zhao

    ________________________________________
    From: Weiwei Yang <cheersy...@hotmail.com>
    Sent: Tuesday, August 14, 2018 8:54:55 AM
    To: Sunil G; Yufei Gu
    Cc: Daniel Templeton; zsc19940...@outlook.com; Hadoop Common; Wangda Tan
    Subject: Re: YARN SLS improving idea

    Hi Sichen

    Thanks for proposing this, this for sure a nice enhancement for SLS.
    For SLS improvements, feel free to create JIRA tickets under 
https://issues.apache.org/jira/browse/YARN-5065, that’s the umbrella we track 
all related items.
    For scheduling requests support, there is an existing JIRA and an existing 
patch in https://issues.apache.org/jira/browse/YARN-8007, would you like to 
take a look?

    Thanks
    --
    Weiwei

    From: Sunil G <sun...@apache.org>
    Date: Tuesday, August 14, 2018 at 1:31 AM
    To: Yufei Gu <flyrain...@gmail.com>
    Cc: Daniel Templeton <dan...@cloudera.com>, YANG WEIWEI 
<cheersy...@hotmail.com>, "zsc19940...@outlook.com" <zsc19940...@outlook.com>, 
Hadoop Common <common-dev@hadoop.apache.org>, Wangda Tan <wheele...@gmail.com>
    Subject: Re: YARN SLS improving idea

    Hi Sichen

    1. Add input support for the scheduling request format.
    Yes. I suppose this change is in SLS end (client to generate requests)
    2. Add support for scheduling request resource format in NMSim.
    Makes sense.
    3. Adding scheduling request support for the Capacity Scheduler(maybe it is 
already done in current version).
    This support is already there.

    Adding to this, major challenge is to specify constrains per app level and 
verifying results. I also suggest to cross check the scheduler invariants check 
support added as per YARN-6547 and see how we can incorporate same to predict 
o/p

    - Sunil


    On Mon, Aug 13, 2018 at 10:39 PM Yufei Gu 
<flyrain...@gmail.com<mailto:flyrain...@gmail.com>> wrote:
    +YANG WEIWEI<mailto:cheersy...@hotmail.com>

    Make sense to me from SLS perspective, but I am not familiar with Placement 
Constraints. Add WeiWei.

    Best,

    Yufei

    `This is not a contribution`


    On Sat, Aug 11, 2018 at 8:57 AM Daniel Templeton 
<dan...@cloudera.com<mailto:dan...@cloudera.com>> wrote:
    Yufei, Wangda, Sunil, any comments?

    Daniel

    On 8/11/18 8:48 AM, Sichen Zhao wrote:
    > Hi,
    > Is there anyone who can reply my ideas?
    >
    > Best Regards
    > Sichen Zhao
    >
    > ________________________________________
    > From: Sichen Zhao 
<zsc19940...@outlook.com<mailto:zsc19940...@outlook.com>>
    > Sent: Friday, August 10, 2018 11:10
    > To: Hadoop Common
    > Subject: YARN SLS improving idea
    >
    > Hi,
    > I am a developer from AliBaBa China, i recently used SLS for scheduling 
simulation, SLS currently supports multidimensional resource input(CPU, mem , 
other resources: disk). But SLS can't take scheduling request, which is 
currently widely used in YARN, as input, so Placement Constraints and 
attributes are not supported.
    >
    > So what i wanna improve the SLS: Add scheduling emulation for scheduling 
request resource format.
    >
    > The specific work is as follows:
    > 1. Add input support for the scheduling request format.
    > 2. Add support for scheduling request resource format in NMSim.
    > 3. Adding scheduling request support for the Capacity Scheduler(maybe it 
is already done in current version).
    >
    > What do you think about my ideas?
    >
    >
    > Best Regards
    > Sichen Zhao
    >
    > ---------------------------------------------------------------------
    > To unsubscribe, e-mail: 
common-dev-unsubscr...@hadoop.apache.org<mailto:common-dev-unsubscr...@hadoop.apache.org>
    > For additional commands, e-mail: 
common-dev-h...@hadoop.apache.org<mailto:common-dev-h...@hadoop.apache.org>
    >
    >
    > ---------------------------------------------------------------------
    > To unsubscribe, e-mail: 
common-dev-unsubscr...@hadoop.apache.org<mailto:common-dev-unsubscr...@hadoop.apache.org>
    > For additional commands, e-mail: 
common-dev-h...@hadoop.apache.org<mailto:common-dev-h...@hadoop.apache.org>
    >



---------------------------------------------------------------------
To unsubscribe, e-mail: common-dev-unsubscr...@hadoop.apache.org
For additional commands, e-mail: common-dev-h...@hadoop.apache.org

Reply via email to