Roman, thank you for your attention. It looks like you are absolutely right. Thank you very much for helping.
Before submitting a job I do next steps: 1. ./bin/start-cluster.sh 2. ./bin/taskmanager.sh start And in my code there is these line: env.setStateBackend(new RocksDBStateBackend("file:///home/flink/checkpoint-data", true)); So I have a directory 'checkpoint-data' and there I can see chk-x (x=index of checkpointing) folder. I assume it is responsible to store my states as a full snapshot. When I stop the app this chk-x folder is removed. So I cant recover from that point. I added these lines: CheckpointConfig checkpointConfig = env.getCheckpointConfig(); checkpointConfig.enableExternalizedCheckpoints(CheckpointConfig.ExternalizedCheckpointCleanup.RETAIN_ON_CANCELLATION); And now it's works. P.S.: But maybe it's better to use savepoint conceptually (not checkpoint) Thanks again, Yuri L. -- Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/