You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Nikolay Tikhonov (JIRA)" <ji...@apache.org> on 2016/07/25 18:12:20 UTC

[jira] [Commented] (IGNITE-3164) Add an option to send resulting value instead of entry processor in transactional cache

    [ https://issues.apache.org/jira/browse/IGNITE-3164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15392416#comment-15392416 ] 

Nikolay Tikhonov commented on IGNITE-3164:
------------------------------------------

Implemented that for {{pessemistic}} transactions {{entry processors}} called when processing {{GridNearLockRequest}} request on primary nodes (results of invocation and new value for this key send together with {{GridNearLockResponce}}). For {{optimisic}} transactions, added new message {{GridNearInvokeResponce}} which in case invoke replaces a regular {{GridNearLockRequest}}  and contains entry processor with arguments. In case when it invoked several times for the same key then only first invocation will be called on primary node and subsequent calls will be on client node.

> Add an option to send resulting value instead of entry processor in transactional cache
> ---------------------------------------------------------------------------------------
>
>                 Key: IGNITE-3164
>                 URL: https://issues.apache.org/jira/browse/IGNITE-3164
>             Project: Ignite
>          Issue Type: Improvement
>          Components: cache
>    Affects Versions: 1.5.0.final
>            Reporter: Valentin Kulichenko
>            Assignee: Nikolay Tikhonov
>              Labels: important
>             Fix For: 1.7
>
>
> In some cases user can't guarantee that EP behaves consistently on all nodes. In transactional cache this can lead to inconsistent cache, because we invoke EP on both primary and backup nodes.
> We can add {{withSendValueToBackup()}} flag (naming is arguable) which will override current default behavior.
> We also need to update documentation, especially provide the explanation of EP behavior in atomic and transactional caches.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)