The disk space in all nodemanagers will be less and it will cause yarn to stop executing all jobs

To handle this situation, use the following script:

#! /bin/bash

### To be run by admins

echo "Started!!!"


for i in $(seq 0 7); 


do yes | ssh wn${i}-infowo -t 


sudo rm -rf /mnt/resource/hadoop/yarn/local/usercache /mnt/resource/hadoop/yarn/local/filecache; 


done


Thanks

Sripada