You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@bookkeeper.apache.org by "Jia Zhai (JIRA)" <ji...@apache.org> on 2017/07/28 04:40:02 UTC

[jira] [Resolved] (BOOKKEEPER-1104) BookieInitializationTest.testWithDiskFullAndAbilityToCreateNewIndexFile testcase is unreliable

     [ https://issues.apache.org/jira/browse/BOOKKEEPER-1104?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jia Zhai resolved BOOKKEEPER-1104.
----------------------------------
    Resolution: Fixed

Issue resolved by merging pull request 310
            [https://github.com/apache/bookkeeper/pull/310]

            {noformat}
            commit 0ccbc0e9c3e6a82d5e44b8347cc6567e043f253f
Author:     Samuel Just <sj...@salesforce.com>
AuthorDate: Fri Jul 28 12:39:03 2017 +0800
Commit:     jiazhai <zh...@live.com>
CommitDate: Fri Jul 28 12:39:03 2017 +0800

    BOOKKEEPER-1104: Fix testWithDiskFullAndAbilityToCreateNewIndexFile
    
    The usage threshhold was chosen to be very close to the actual disk
    usage at test time.  This made the test unnecessarily fragile in the
    case that there other things concurrently using the disk.  Since we
    aren't really testing that here, simply set the threshhold to be really
    low.
    
    Signed-off-by: Samuel Just <sjustsalesforce.com>
    
    Author: Samuel Just <sj...@salesforce.com>
    
    Reviewers: Jia Zhai <None>, Matteo Merli <None>
    
    This closes #310 from athanatos/forupstream/BOOKKEEPER-1104

            {noformat}
            

> BookieInitializationTest.testWithDiskFullAndAbilityToCreateNewIndexFile testcase is unreliable
> ----------------------------------------------------------------------------------------------
>
>                 Key: BOOKKEEPER-1104
>                 URL: https://issues.apache.org/jira/browse/BOOKKEEPER-1104
>             Project: Bookkeeper
>          Issue Type: Bug
>            Reporter: Samuel Just
>            Assignee: Samuel Just
>            Priority: Minor
>             Fix For: 4.5.0
>
>
> The bug is that the test sets the full threshhold really close the actual usage (though just below), and if the disk is otherwise in use, the usage can fall below that threshhold during the test.  Instead, just set it to .001 since that's not what we're testing here anyway.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)