Cleaning the disks of the batch farm machines
Problem
After a crash of a job on the batchfarm, there might be files left over on the /tmp directory of the execution host. Especialy HLD files need a lot of diskspace (2GB).
Normaly a job should clean up the disk after succesfull execution.
Solution
The
clean_main_batch.sh script will submit a job to each of the batch farm machines that checks for files and directories of the user that executes the script. All files and directories that are older then 30 minutes are removed.
Last change by
WikiGuest