[ 
https://issues.apache.org/jira/browse/MESOS-2353?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14488069#comment-14488069
 ] 

Benjamin Mahler commented on MESOS-2353:
----------------------------------------

The move assignments are committed, although only offer a minor improvement. 
Let's take a more rigorous approach to this :)

{noformat}
commit 5028490f99d175927a742c14ede5da07bc40c668
Author: Benjamin Mahler <benjamin.mah...@gmail.com>
Date:   Thu Apr 9 12:13:58 2015 -0700

    Used move assignment in the master's state.json endpoint.

    This is a simple attempt at improvement for MESOS-2353 in the
    interim of writing a more comprehensive JSON benchmark.

    Per the discussion on r/31700, moves here avoid the very expensive
    copies of the large JSON objects so long as variant has move assignment
    defined (or generated) and our objects have move generated.

    It appears that variant has move assignment and our json classes can
    have move assignment generated, so this should help.

    Review: https://reviews.apache.org/r/32419
{noformat}

> Improve performance of the master's state.json endpoint for large clusters.
> ---------------------------------------------------------------------------
>
>                 Key: MESOS-2353
>                 URL: https://issues.apache.org/jira/browse/MESOS-2353
>             Project: Mesos
>          Issue Type: Improvement
>          Components: master
>            Reporter: Benjamin Mahler
>            Assignee: Benjamin Mahler
>              Labels: newbie, scalability, twitter
>
> The master's state.json endpoint consistently takes a long time to compute 
> the JSON result, for large clusters:
> {noformat}
> $ time curl -s -o /dev/null localhost:5050/master/state.json
> Mon Jan 26 22:38:50 UTC 2015
> real  0m13.174s
> user  0m0.003s
> sys   0m0.022s
> {noformat}
> This can cause the master to get backlogged if there are many state.json 
> requests in flight.
> Looking at {{perf}} data, it seems most of the time is spent doing memory 
> allocation / de-allocation. This ticket will try to capture any low hanging 
> fruit to speed this up. Possibly we can leverage moves if they are not 
> already being used by the compiler.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to