[
https://issues.apache.org/jira/browse/FLINK-41209?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ramin Gharib updated FLINK-41209:
---------------------------------
Release Note: Materialized tables in continuous refresh mode can now be
created, suspended, resumed, refreshed, altered and dropped from any
TableEnvironment, not only through the SQL Gateway. The refresh job runs on a
session cluster: set 'execution.target' to 'remote', 'yarn-session' or
'kubernetes-session'. CREATE MATERIALIZED TABLE in continuous mode now fails on
a TableEnvironment without such a target. Before, it only wrote the catalog.
Full refresh mode still requires the SQL Gateway.
> Run and control materialized table refresh jobs from any TableEnvironment
> -------------------------------------------------------------------------
>
> Key: FLINK-41209
> URL: https://issues.apache.org/jira/browse/FLINK-41209
> Project: Flink
> Issue Type: Sub-task
> Components: Client / Job Submission, Table SQL / API, Table SQL /
> Gateway, Table SQL / Planner
> Reporter: Timo Theusner
> Assignee: Timo Theusner
> Priority: Major
> Labels: pull-request-available
> Fix For: 2.5.0
>
>
> Only the SQL Gateway's {{MaterializedTableManager}} starts and controls
> refresh jobs, through Flink's cluster client in {{flink-clients}}, which no
> table module depends on. So {{executeSql()}} materialized-table DDL only
> writes the catalog and leaves the table {{INITIALIZING}}, while {{SUSPEND}},
> {{RESUME}} and {{REFRESH}} throw {{UnsupportedOperationException}}. Move the
> logic into a planner {{MaterializedTableExecutor}}, used by the SQL Gateway
> and every {{TableEnvironment}}, which reaches existing refresh jobs through a
> new internal {{PipelineExecutor}} extension in {{flink-core}}.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
