To put a job on hold when itâs disk usage exceeds its request, add this to the configuration of your execute node
STARTD_JOB_ATTRS = $(STARTD_JOB_ATTRS) RequestDisk
DISK_EXCEEDED = (JobUniverse != 13 && DiskUsage =!= UNDEFINED && DiskUsage > RequestDisk)
use POLICY : WANT_HOLD_IF( DISK_EXCEEDED, 105, disk usage exceeded request_disk )
The choice of 105 for the hold subcode is arbitrary here, you can use a different value if you like.
We ignore VM universe (universe 13) because VM disk usage is handled differently than vanilla universe.
I donât understand what you mean by this question
> Will it include the job output as well?
When a job goes on hold, the output if the job is not returned to the user, is that the question?
What will be the best way to put job on hold when the DiskUsage >= requested disk?
Will it include the job output as well?
The reason is that one user can fill the disk for all running jobs.
Sent from my Mi phone