[ 
https://issues.apache.org/jira/browse/SAMZA-1768?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16560316#comment-16560316
 ] 

ASF GitHub Bot commented on SAMZA-1768:
---------------------------------------

GitHub user xinyuiscool opened a pull request:

    https://github.com/apache/samza/pull/588

    SAMZA-1768: Handle corrupted OFFSET file

    This patch addresses the following tickets:
    
    SAMZA-1778: SIGSEGV when reading properties (metrics) on a closed RocksDB 
store
    SAMZA-1777: Logged store OFFSET file write during flush should be atomic
    SAMZA-1768: Handle corrupted OFFSET file elegantly

You can merge this pull request into a Git repository by running:

    $ git pull https://github.com/xinyuiscool/samza SAMZA-1768

Alternatively you can review and apply these changes as the patch at:

    https://github.com/apache/samza/pull/588.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

    This closes #588
    
----
commit 27e24e8238b39f0cf927d34fbada8f6e8d22e2c0
Author: xinyuiscool <xiliu@...>
Date:   2018-07-26T18:14:45Z

    SAMZA-1768: Handle corrupted OFFSET file elegantly

commit 853f48ab9909fcad410f651d03f2f98507e6222d
Author: xinyuiscool <xiliu@...>
Date:   2018-07-26T18:31:50Z

    Minor changes

----


> Handle corrupted OFFSET file elegantly
> --------------------------------------
>
>                 Key: SAMZA-1768
>                 URL: https://issues.apache.org/jira/browse/SAMZA-1768
>             Project: Samza
>          Issue Type: Bug
>            Reporter: Xinyu Liu
>            Assignee: Xinyu Liu
>            Priority: Major
>
> TaskStorageManager.readOffsetFile() has a bug that if the OFFSET file is 
> corrupted, it will throw exception and shut down the container. If host 
> affinity is turned on, the container won't be able start up again since it 
> will read the same corrupted OFFSET file every time, until it was manually 
> removed. Since we cannot recover in this case, we should catch the exception 
> and return null and let the store bootstrap from changelog.



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

Reply via email to