[jira] [Commented] (GEODE-6899) retried client should set last try's version tag if found

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

    [ https://issues.apache.org/jira/browse/GEODE-6899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16869970#comment-16869970 ] 

ASF subversion and git services commented on GEODE-6899:
--------------------------------------------------------

Commit 0011e1405caa93c0fafc1a87b15c5fd58679f56d in geode's branch refs/heads/feature/GEODE-6899 from zhouxh
[ https://gitbox.apache.org/repos/asf?p=geode.git;h=0011e14 ]

GEODE-6899: retried client should set last try's version tag if found

        Co-authored-by: Xiaojian Zhou <gz...@pivotal.io>
        Co-authored-by: Eric Shu <es...@pivotal.io>


> retried client should set last try's version tag if found
> ---------------------------------------------------------
>
>                 Key: GEODE-6899
>                 URL: https://issues.apache.org/jira/browse/GEODE-6899
>             Project: Geode
>          Issue Type: Bug
>            Reporter: xiaojian zhou
>            Priority: Major
>
> client does a put to serverA with replicated region, serverA distributed to B and C, before the distribution arrived at C, A is killed. Then client could retry to C. C noticed this is a retry operation, it will search for previous try's version tag. 
> The found tag should be set into the current event. Interesting thing is:
> I found other operations, such as PutIfAbsent and create, they both did that. But replace (i.e. put) did not.  
> Another issue is: GEODE-6802 introduce a synchronizeIfNotScheduled(). But there could be a race that membershipListener is also scheduling. The fix is to pause 1 second before calling the newly introduced synchronizeIfNotScheduled(). 



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