Hi Xiaolong, For new versions such as flink-1.17, flink sql-gateway supports job management and user can stop/start jobs with savepoint. You can start a job with a given savepoint path as [1] and stop a job with or without savepoint as [2].
[1] https://nightlies.apache.org/flink/flink-docs-master/docs/dev/table/sqlclient/#start-a-sql-job-from-a-savepoint [2] https://nightlies.apache.org/flink/flink-docs-master/docs/dev/table/sqlclient/#terminating-a-job Best, Shammon FY On Tue, Jul 18, 2023 at 9:56 AM Xiaolong Wang <xiaolong.w...@smartnews.com> wrote: > Hi, Shammon, > > I know that the job manager can auto-recover via HA configurations, but > what if I want to upgrade the running Flink SQL submitted by the Flink SQL > gateway ? > > In normal cases, I can use the > >> ./flink run application -s ${SAVEPOINT_PATH} local://${FLINK_JOB_JAR} > > to resume a Flink job from a savepoint/checkpoint. The question is, how to > do so with Flink sql gateway ? What should I fill in the ${FLINK_JOB_JAR} > field ? > > Thanks in advanced. > > On Mon, Jul 17, 2023 at 9:14 AM Shammon FY <zjur...@gmail.com> wrote: > >> Hi Xiaolong, >> >> When a streaming job is submitted via Sql-Gateway, its lifecycle is no >> longer related to Sql Gateway. >> >> Returning to the issue of job recovery, I think if your job cluster is >> configured with HA, jobmanager will recover running streaming jobs from >> their checkpoints after a failover occurs. >> >> Best, >> Shammon FY >> >> >> On Thu, Jul 13, 2023 at 10:22 AM Xiaolong Wang < >> xiaolong.w...@smartnews.com> wrote: >> >>> Hi, >>> >>> I'm currently working on providing a SQL gateway to submit both >>> streaming and batch queries. >>> >>> My question is, if a streaming SQL is submitted and then the jobmanager >>> crashes, is it possible to resume the streaming SQL from the latest >>> checkpoint with the SQL gateway ? >>> >>> >>> >>