[
https://issues.apache.org/jira/browse/FLINK-7498?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Piotr Nowojski updated FLINK-7498:
----------------------------------
Description: Make sure that state fields are coupled together between
checkpoints. This way all opened transactions by one operator instance will be
restored to same operator after restoring from checkpoint. (was: Currently
when using TwoPhaseCommitSinkFunction, if there is some intermittent failure in
"beginTransaction", not only the snapshot that triggered this call fail, but
also any subsequent write requests will fail also, rendering such sink unusable
until application restart.
This issue is in code that hasn't been released yet.)
> Bind together state fields of TwoPhaseCommitSinkFunction
> --------------------------------------------------------
>
> Key: FLINK-7498
> URL: https://issues.apache.org/jira/browse/FLINK-7498
> Project: Flink
> Issue Type: Bug
> Components: Streaming
> Affects Versions: 1.4.0
> Reporter: Piotr Nowojski
> Assignee: Piotr Nowojski
> Fix For: 1.4.0
>
>
> Make sure that state fields are coupled together between checkpoints. This
> way all opened transactions by one operator instance will be restored to same
> operator after restoring from checkpoint.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
