[
https://issues.apache.org/jira/browse/FLINK-41390?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Timo Theusner updated FLINK-41390:
----------------------------------
Description: When a materialized table statement starts a continuous
refresh job and the catalog then fails to record it, the job keeps running
unrecorded. On CREATE the job is orphaned. On RESUME, convert-to-MT and the
ALTER restart, a retry or the ALTER rollback starts a second job writing to the
same sink. Cancel such jobs, keep the original error when the cleanup fails,
and name the savepoint when SUSPEND cannot record it. (was: When {{CREATE
MATERIALIZED TABLE}} submits a continuous refresh job but the catalog can't
record it, the table is dropped while the job keeps running unrecorded. If that
compensation drop fails, the original error is lost, and a {{SUSPEND}} whose
catalog write fails after the stop loses the savepoint path. Cancel the
unrecorded job, keep the original error and name the savepoint, for the SQL
Gateway and every {{TableEnvironment.}})
> Cancel materialized table refresh jobs the catalog could not record
> -------------------------------------------------------------------
>
> Key: FLINK-41390
> URL: https://issues.apache.org/jira/browse/FLINK-41390
> Project: Flink
> Issue Type: Sub-task
> Components: Table SQL / Planner
> Reporter: Timo Theusner
> Assignee: Timo Theusner
> Priority: Major
> Labels: pull-request-available
>
> When a materialized table statement starts a continuous refresh job and the
> catalog then fails to record it, the job keeps running unrecorded. On CREATE
> the job is orphaned. On RESUME, convert-to-MT and the ALTER restart, a retry
> or the ALTER rollback starts a second job writing to the same sink. Cancel
> such jobs, keep the original error when the cleanup fails, and name the
> savepoint when SUSPEND cannot record it.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
