[
https://issues.apache.org/jira/browse/IGNITE-6756?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Tim Onyschak updated IGNITE-6756:
---------------------------------
Attachment: GridCacheNearClientHitTest.java
Test used to prove.
> Primary Node Change causes invalidation of GridNearCacheEntry topVer
> --------------------------------------------------------------------
>
> Key: IGNITE-6756
> URL: https://issues.apache.org/jira/browse/IGNITE-6756
> Project: Ignite
> Issue Type: Bug
> Security Level: Public(Viewable by anyone)
> Components: general
> Affects Versions: 2.1, 2.2, 2.3
> Reporter: Tim Onyschak
> Attachments: GridCacheNearClientHitTest.java
>
>
> When using a near cache after a cache exists in a cluster, their appears to
> be a bug which causes the GridNearCacheEntry.topVer to be set to NONE on a
> check of the primaryNode after a topology change when the
> !nodeId.equals(primary.id(). This will prevent the topVer from being updated
> to the latest, which then cause the GridNearCacheEntry.valid to return false
> and force a hard hit to the cluster.
> Steps to reproduce:
> # Create 2 node
> # create cache on 1 node
> # populate cache 1
> # create client
> # create nearCache of original cache
> # warm nearCache
> # Cause topolgy change
> Expected:
> Topology change cause 1 hard hit to cluster to make sure value value is up to
> date (guessing on this)
> All future request will be retrived from near cache
> Actual
> All request are to cluster since the GridNearCacheEntry.valid method return
> false.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
