phoenix-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Monani Mihir (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (PHOENIX-5218) When INDEX_FAILURE_HANDLING_REBUILD_PERIOD is set, Phoenix Client can still transition Index from INACTIVE to ACTIVE
Date Fri, 29 Mar 2019 18:22:00 GMT

    [ https://issues.apache.org/jira/browse/PHOENIX-5218?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16805295#comment-16805295
] 

Monani Mihir commented on PHOENIX-5218:
---------------------------------------

Hi [~gjacoby] ,
{quote}Can we have logic that either synchronously unassigns, or waits on the unassign being
complete?
{quote}
I have used sleep and async region un-assign to simulate index write failure( and then succeeds
in retry) while region moves.

I agree that sleep can make test flapper but i am not sure how to generate index write failure
in test case which can succeed in retries. 

 

I have uploaded new patch [^PHOENIX-5218-4.x-HBase-1.3.02.patch]  and removed commented
statement from another test case

 

 

> When INDEX_FAILURE_HANDLING_REBUILD_PERIOD is set, Phoenix Client can still transition
Index from INACTIVE to ACTIVE
> --------------------------------------------------------------------------------------------------------------------
>
>                 Key: PHOENIX-5218
>                 URL: https://issues.apache.org/jira/browse/PHOENIX-5218
>             Project: Phoenix
>          Issue Type: Bug
>    Affects Versions: 4.15.0, 4.14.1, 5.1.0
>            Reporter: Monani Mihir
>            Assignee: Monani Mihir
>            Priority: Critical
>         Attachments: PHOENIX-5218-4.x-HBase-1.3.01.patch, PHOENIX-5218-4.x-HBase-1.3.02.patch
>
>
> With PHOENIX-5094, we were handling index transition from INACTIVE to ACTIVE state
> While debugging test cases failures for PHOENIX-5194 for PartialIndexRebuilderIT, two
test cases 1) testIndexWriteFailureDisablingIndex 2) testIndexWriteFailureLeavingIndexActive
fails.
> For this IT, INDEX_FAILURE_HANDLING_REBUILD_PERIOD is set to 50sec and in above mentioned
test cases round robin Partial Index rebuilder is being used instead of building entire index
from INDEX_DISABLE_TIMESTAMP. 



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

Mime
View raw message