geode-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Bruce Schuchardt (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (GEODE-2718) inconsistent reads during network partition
Date Thu, 30 Mar 2017 14:46:42 GMT

    [ https://issues.apache.org/jira/browse/GEODE-2718?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15949218#comment-15949218
] 

Bruce Schuchardt commented on GEODE-2718:
-----------------------------------------

I was talking with [~dschneider] about this ticket and he pointed out that the bucket advisors
don't really need to see the view change.  They just need to notice that data has been lost
.

I think the correct thing to do is take advantage of the redundancy zone (rack aware) feature
of Geode to make sure you don't lose data in the event of a network partition.  That will
cause buckets to be backed up on a different rack.

> inconsistent reads during network partition
> -------------------------------------------
>
>                 Key: GEODE-2718
>                 URL: https://issues.apache.org/jira/browse/GEODE-2718
>             Project: Geode
>          Issue Type: Bug
>          Components: membership, regions
>            Reporter: Swapnil Bawaskar
>
> non-persistent partitioned regions exhibit the following behavior:
>  In the event of a network partition where all copies of a bucket end up on the losing
side, a get() operation returns a null. A null can signify that the key did not exist, so
we should be more explicit about the network partition and throw an error like PartitionOfflineException
(which is thrown if the region was persistent).



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Mime
View raw message