You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@aurora.apache.org by "Maxim Khutornenko (JIRA)" <ji...@apache.org> on 2015/04/08 23:43:12 UTC

[jira] [Created] (AURORA-1263) Fetching update details for a large job update takes seconds in the UI

Maxim Khutornenko created AURORA-1263:
-----------------------------------------

             Summary: Fetching update details for a large job update takes seconds in the UI
                 Key: AURORA-1263
                 URL: https://issues.apache.org/jira/browse/AURORA-1263
             Project: Aurora
          Issue Type: Bug
          Components: Scheduler
            Reporter: Maxim Khutornenko
            Assignee: Maxim Khutornenko


Fetching job update details for a large job (e.g. one with 4000 instances) takes around 30 seconds on average. Investigation shows the majority of that time is spent in SQL/MyBatis. Due to the way we select and map JobUpdateDetails data as a single joined statement, the event rows multiply the end resultset upwards of 300K rows for a 4K instance job. That resultset includes serialized TaskConfigs (twice) and draws significant CPU and RAM during mapping.

Refactoring JobUpdateDetails mapping statement to avoid joining against event tables will fix the problem.



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