[jira] [Commented] (ASTERIXDB-2269) All active nodes are included in job resources calculation

["ASF subversion and git services (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ASTERIXDB-2269?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16342422#comment-16342422 ] 

ASF subversion and git services commented on ASTERIXDB-2269:
------------------------------------------------------------

Commit a05657664cb34fc79f4b94757e38718a689806f7 in asterixdb's branch refs/heads/master from [~mhubail]
[ https://git-wip-us.apache.org/repos/asf?p=asterixdb.git;h=a056576 ]

[ASTERIXDB-2269][RT] Use Job Locations To Estimate Resources

- user model changes: no
- storage format changes: no
- interface changes: yes
    - INodeJobTracker (+) getJobParticipatingNodes

Details:
- Use job locations to calculate the job's required
  resources rather than all cluster locations.
- Add test case.

Change-Id: Iecd8e234aa52a9f324e64044e01477fb12dc14e6
Reviewed-on: https://asterix-gerrit.ics.uci.edu/2330
Sonar-Qube: Jenkins <je...@fulliautomatix.ics.uci.edu>
Tested-by: Jenkins <je...@fulliautomatix.ics.uci.edu>
Integration-Tests: Jenkins <je...@fulliautomatix.ics.uci.edu>
Contrib: Jenkins <je...@fulliautomatix.ics.uci.edu>
Reviewed-by: Michael Blow <mb...@apache.org>


> All active nodes are included in job resources calculation
> ----------------------------------------------------------
>
>                 Key: ASTERIXDB-2269
>                 URL: https://issues.apache.org/jira/browse/ASTERIXDB-2269
>             Project: Apache AsterixDB
>          Issue Type: Bug
>          Components: RT - Runtime
>            Reporter: Murtadha Hubail
>            Assignee: Murtadha Hubail
>            Priority: Major
>
> Currently, when a query comes, all active nodes are assumed to be participants in the query execution and therefore the query required resources is calculated accordingly. However, this isn't the case when a dataset is partitioned in a node group that doesn't include all nodes (e.g. the metadata datasets). This will result in an overestimation in resources required and will cause the query to fail with (HYR0009: Job requirement exceeds capacity) when one of the none participating nodes leaves the cluster when it shouldn't fail.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)