[
https://issues.apache.org/jira/browse/FLINK-16487?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17053975#comment-17053975
]
Zhimin Li edited comment on FLINK-16487 at 3/7/20, 10:35 AM:
-------------------------------------------------------------
p1.When creating the sink a addr must be specified.need to implements
KuduSinkFunction
p2. The sink uses a metadataUpdateInterval to determine and update metadata.
Every input element can be regarded as the row of base table,the base table
contains some columns.
1. user can create some tables in Kudu as need, set the targetTable in
Kudu,the targetTable should not be null.
2. user can add columns or delete columns by requirement.
but the columns of to add or to delete should be in the base table,the sink
will determine lateast metadata periodically and update the metadata.
p3.If case of a failure, and in order to guarantee exactly-once semantics, the
sink should flush before checkpoint occurred.
was (Author: forward):
<p>When creating the sink a \{@code addr} must be specified.
need to implements \{@link KuduSinkFunction <T>} and set target table in
\{@link KuduSinkFunction<T>}.
<p>The sink uses a \{@code metadataUpdateInterval} to determine and update
metadata.
Every input element can be regarded as the row of base table,the base table
contains some columns.
1. user can create some tables in Kudu as need, set the targetTable in Kudu,the
targetTable should not be null,\{@code setTargetTable}
2. user can add columns or delete columns by requirement,
but the columns of to add or to delete should be in the base table,the sink
will determine lateast metadata periodically
and update the metadata {@code targetTables)
<p>If case of a failure, and in order to guarantee exactly-once semantics, the
sink should \{@code flush()} before
checkpoint occurred.
> Add flink-connector-kudu
> ------------------------
>
> Key: FLINK-16487
> URL: https://issues.apache.org/jira/browse/FLINK-16487
> Project: Flink
> Issue Type: New Feature
> Components: flink-contrib
> Affects Versions: 1.11.0
> Reporter: Zhimin Li
> Priority: Minor
> Attachments: kudu_sink_process.pdf
>
>
> Kudu needs to be used in many flow processing scenarios. Our company's
> production environment is also in use. I want to contribute my connector kudu
> to Flink to help more people use it. Here are my design documents. Welcome to
> discuss with you and improve my ideas
> 许多流处理场景下需要使用到kudu，我们公司生产环境也在使用，我想将我的connector-kudu贡献到flink，帮助更多人使用，以下是我的设计文档，欢迎和大家一起讨论，改进我的想法
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
