[jira] [Commented] (IGNITE-8614) The atomic cache in inconsistent state

["Vladimir Ozerov (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/IGNITE-8614?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16629986#comment-16629986 ] 

Vladimir Ozerov commented on IGNITE-8614:
-----------------------------------------

Moved to 2.8 due to inactivity. Please feel free to move back if you will be able to complete the ticket by AI 2.7 code freeze date, September 30, 2018.

> The atomic cache in inconsistent state
> --------------------------------------
>
>                 Key: IGNITE-8614
>                 URL: https://issues.apache.org/jira/browse/IGNITE-8614
>             Project: Ignite
>          Issue Type: Task
>    Affects Versions: 2.4
>            Reporter: Denis Garus
>            Assignee: Denis Garus
>            Priority: Major
>             Fix For: 2.8
>
>         Attachments: ReproducerAtomicCacheInconsistentStateTest.java
>
>
> Cache configuration:
> CacheMode - PARTITIONED
> Backups - 1
> RebalanceMode - SYNC
> WriteSynchronizationMode - FULL_ASYNC
> There is a situation when the atomic cache becomes inconsistent.
> There are two nodes: Node_1 with primary partition and Node_2 with backup partition.
> 1. Node_3 is started;
> 2. Before a rebalance of the partition from Node_1 to Node_3 is occurred new value is put to the partition on Node_1;
> 3. Node_1 creates a future for a propagating new value to Node_2 and Node_3;
> 4. Node_3 gets new value from Node_1 in the process a rebalance;
> 5. Node_1 is stopped;
> 6. The future from step 3 is failed because of step 5. 
> Now we have Node_3 with the new value and Node_2 with old value.
> We need to find out the way to fix this inconsistent state and implement one.
> The reproducer is in the attach.



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