[ 
https://issues.apache.org/jira/browse/IGNITE-4929?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Semen Boikov resolved IGNITE-4929.
----------------------------------
    Resolution: Fixed
      Assignee:     (was: Semen Boikov)

When execute tx on changing topology we already send old value available on 
primary to all backups, so this old value can be used on backups to execute 
'invoke'. So I got rid of code in GridDhtTxPrepareFuture sending to backups 
computed value instead of entry processors.

> onePhase tx for invoke may fail on changing topology
> ----------------------------------------------------
>
>                 Key: IGNITE-4929
>                 URL: https://issues.apache.org/jira/browse/IGNITE-4929
>             Project: Ignite
>          Issue Type: Bug
>          Components: cache
>            Reporter: Semen Boikov
>            Priority: Critical
>             Fix For: 2.0
>
>
> One phase tx for invoke may fail in this scenario:
> - backup did not rebalanced partition yet
> - in this case primary sends on backup new value instead of entry processor, 
> in this case on backup we loose information about value calculated by entry 
> processor
> - primary fails before it receives response from backup, client sends 'check' 
> request on backup to get return value, but return value on backup does not 
> contain expected invoke result
> Added test IgniteOnePhaseCommitInvokeTest.



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

Reply via email to