Add fail_if_missing to ValohaiDownloadExecutionOperator

[arimbr]

With this code, If no output is downloaded it fails by default. It will be clearer to avoid triggering downstream tasks in case the filename did not match.

I tested locally. For example if we now set output_name='model.pbt' instead of output_name='model.pb' we get:

[2019-11-13 16:43:08,861] {base_task_runner.py:101} INFO - Job 135: Subtask download_model airflow.exceptions.AirflowException: Failed to find any output for task_id: train_model, output_name: model.pbt, output_name_pattern: None
[2019-11-13 16:43:13,166] {logging_mixin.py:95} INFO - [2019-11-13 16:43:13,165] {jobs.py:2562} INFO - Task exited with return code 1

[ppierrep]

Should it be nice and doable to inform all inputs name when a mismatch happen ?

[arimbr]

@ppierrep That's a good point! I think it is already handled as for each ignored output we get a log:

'Ignore ouput name: ...

Can you confirm that this is the case or did I misread something?

[jdiprizio]

@ppierrep That's a good point! I think it is already handled as for each ignored output we get a log:

'Ignore ouput name: ...

Can you confirm that this is the case or did I misread something?

Seems like this is indeed covered by existing logging mechanisms.

[ppierrep]

Seems like this is indeed covered by existing logging mechanisms.

Oh my bad !