[
https://issues.apache.org/jira/browse/FLINK-23870?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Anton Kalashnikov updated FLINK-23870:
--------------------------------------
Description:
Test for https://issues.apache.org/jira/browse/FLINK-22684:
# configure the external checkpoint address (storage)
# enable unaligned checkpoint always (not the default aligned or alternative)
# wait for the checkpoint then stop the job
# restore from the last checkpoint - it should be restored on the last
state(optional step)
# set the
`execution.checkpointing.recover-without-channel-state.checkpoint-id` to the
last checkpoint
# restore from the last checkpoint - in-flight data would be ignored so it
would be restored a little less data than expected in the usual case.
was:
Test for https://issues.apache.org/jira/browse/FLINK-22684:
# configure the externalize checkpoint for unaligned checkpoint
# wait for the checkpoint then stop the job
# restore from the last checkpoint - it should be restored on the last
state(optional step)
# set the
`execution.checkpointing.recover-without-channel-state.checkpoint-id` to the
last checkpoint
# restore from the last checkpoint - in-flight data would be ignored so it
would be restored a little less data than expected in the usual case.
> Test the ability for ignoring in-flight data on recovery
> --------------------------------------------------------
>
> Key: FLINK-23870
> URL: https://issues.apache.org/jira/browse/FLINK-23870
> Project: Flink
> Issue Type: Technical Debt
> Reporter: Anton Kalashnikov
> Priority: Major
>
> Test for https://issues.apache.org/jira/browse/FLINK-22684:
> # configure the external checkpoint address (storage)
> # enable unaligned checkpoint always (not the default aligned or alternative)
> # wait for the checkpoint then stop the job
> # restore from the last checkpoint - it should be restored on the last
> state(optional step)
> # set the
> `execution.checkpointing.recover-without-channel-state.checkpoint-id` to the
> last checkpoint
> # restore from the last checkpoint - in-flight data would be ignored so it
> would be restored a little less data than expected in the usual case.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
