Hi,

I have run into a job failure because of disk space. I noticed that in case
of multi stage job (e.g M-R-R-R) the intermediate output data from all the
stages are not deleted until the whole job is complete. Is there any
configuration that will help deletion of the intermediate data if we see
some preconfigured number of child level is already complete. I know we
keep that for failure recovery but in case of M-R-R-R dag, when we are
processing the last level we don't need the output of M stage.

I am using tez 0.7

Regards,
Abhishek Das

Reply via email to