,Hi
.I'm trying to get a flow working with checkpoints and restart the flow when an error occurs somewhere in the flow
When I run the flow the first time (setting the runId), it seems to work, creating temporary data for the checkpoint. I then simulate an error by throwing exception at the end of the flow, which I assume would require the preceding steps (and checkpoints) to complete first
:Once the job fails, I re-run it with the same runId. However, I get the error
Caused by: org.apache.hadoop.mapred.FileAlreadyExistsException: Output directory hdfs://vbox.localdomain:8020/tmp/hadoop-moranparbi1/cea-flow/restart/checkpoint already exists
Seemingly indicating that the flow isn't reusing that checkpoint data, but trying to overwrite it with the new run
?Am I missing something
Thanks