Hi Yunfan,
Jobs are supposed to correctly restart from both savepoints and checkpoints
with different parallelisms if only operator states and keyed states are
used. In the cases where there exist unpartitionable states (e.g., those
are produced by the Checkpointed interface), the job will fail to
Hi all!
I'm currently choosing which API to stick with while implementing Flink jobs
(primarily streaming jobs).
Could you please shed a light on in which API new features are implemented
first? Do all Flink features are available in both APIs?
Many thanks in advance.
Best Regards,
Mike Pry
I using 1.2.0 release, so I read the document in
https://ci.apache.org/projects/flink/flink-docs-release-1.2/setup/savepoints.html.
It means it can recover from the savepoint even if I change the parallelism.
How does flink implement it?
For example using rocksdb+hdfs as state backend, flink just
