[
https://issues.apache.org/jira/browse/FLINK-20252?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17235884#comment-17235884
]
Yuan Mei commented on FLINK-20252:
----------------------------------
sink output:
{code:java}
index:146; key:0; selectedChannel:0; longOrShortString:I am a very long string
to test partial records hohoho hahaha index:147; key:0; selectedChannel:0;
longOrShortString:I am a very long string to test partial records hohoho hahaha
index:148; key:0; selectedChannel:0; longOrShortString:I am a very long string
to test partial records hohoho hahaha index:2658; key:0; selectedChannel:0;
longOrShortString:I am a very long string to test partial records hohoho hahaha
index:2659; key:0; selectedChannel:0; longOrShortString:I am a very long string
to test partial records hohoho hahaha index:2660; key:0; selectedChannel:0;
longOrShortString:I am a very long string to test partial records hohoho
hahaha{code}
> Manual Test for Approximate Local Recovery
> ------------------------------------------
>
> Key: FLINK-20252
> URL: https://issues.apache.org/jira/browse/FLINK-20252
> Project: Flink
> Issue Type: Sub-task
> Reporter: Yuan Mei
> Priority: Critical
> Labels: pull-request-available
> Fix For: 1.12.0
>
>
> This manual test is to verify approximate failover works as expected in
> clusters.
>
> *Test Job Set up:*
> source -> mapper -> printer
> mapper fails after a certain number of records are received.
> link: [https://github.com/apache/flink/pull/14146]
> *Cluster Set up:*
> Test on Yarn clustersm
> add this to flink-conf.yaml
> jobmanager.scheduler.scheduling-strategy: legacy
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
