You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@geode.apache.org by "Bruce Schuchardt (JIRA)" <ji...@apache.org> on 2019/04/12 18:53:00 UTC

[jira] [Reopened] (GEODE-6195) putIfAbsent may get a returned value caused by the same operation due to retry

     [ https://issues.apache.org/jira/browse/GEODE-6195?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Bruce Schuchardt reopened GEODE-6195:
-------------------------------------

Unfortunately the changes made to fix this issue caused inconsistencies in client caches receiving events transmitted from other clusters over WAN gateways.  Servers detecting retried putIfAbsents might tell a client that the operation succeeded but neglect to send it version information about the change.

I'm reopening this ticket and will move the checks to a better place, under the RegionEntry lock.

> putIfAbsent may get a returned value caused by the same operation due to retry
> ------------------------------------------------------------------------------
>
>                 Key: GEODE-6195
>                 URL: https://issues.apache.org/jira/browse/GEODE-6195
>             Project: Geode
>          Issue Type: Bug
>          Components: regions
>    Affects Versions: 1.1.0
>            Reporter: Eric Shu
>            Assignee: Bruce Schuchardt
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: 1.9.0
>
>          Time Spent: 20m
>  Remaining Estimate: 0h
>
> A server which does not store the entry will try the putIfAbsent operation to host member. Currently product treats this as a failed op even though it is due to retried operation. If it does the  operation on behalf of client (and client never failed over to another server), this could leads to client server data inconsistency.



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