You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Alexey Goncharuk (JIRA)" <ji...@apache.org> on 2015/04/29 00:19:06 UTC

[jira] [Commented] (IGNITE-424) Update success flag is not checked in atomic cache

    [ https://issues.apache.org/jira/browse/IGNITE-424?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14518218#comment-14518218 ] 

Alexey Goncharuk commented on IGNITE-424:
-----------------------------------------

We need to revisit logic of applying interceptor on near nodes.

I have added test IgniteNearPrimaryNodeChangedRemoveSelfTest which currently fails because valid flag is not taken into account when applying the interceptor; moreover, when flag is false, the value cannot be passed to interceptor because it is out-of-sync with primary node.

Possible solutions:
1) Bring previous value from primary node on prepare step if interceptor is configured
2) Apply interceptor on primary node and send the result back to near node on prepare step. Interceptor should be applied before sending prepare requests to backups.

> Update success flag is not checked in atomic cache
> --------------------------------------------------
>
>                 Key: IGNITE-424
>                 URL: https://issues.apache.org/jira/browse/IGNITE-424
>             Project: Ignite
>          Issue Type: Bug
>          Components: cache
>    Affects Versions: sprint-2
>            Reporter: Valentin Kulichenko
>            Assignee: Anton Vinogradov
>             Fix For: sprint-5
>
>
> It looks suspicious {{GridDhtAtomicCache.updatePartialBatch}} calls interceptor even if update was not successful.
> This code needs to be reviewed, tested and fixed appropriately if needed.



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