Platform LSF Running Jobs Version 6.2

Chapter 2
Working with Jobs
Running Jobs with Platform LSF
33
Modifying Running Jobs
Modifying resource reservation
A job is usually submitted with a resource reservation for the maximum amount
required. Use
bmod -R to modify the resource reservation for a running job. This
command is usually used to decrease the reservation, allowing other jobs access to the
resource.
The following example sets the resource reservation for job 101 to 25MB of memory
and 50 MB of swap space:
% bmod -R "rusage[mem=25:swp=50]" 101
By default, you can modify resource reservation for running jobs. Set
LSB_MOD_ALL_JOBS in
lsf.conf to modify additional job options.
See “Reserving Resources for Jobs” on page 46 for more details.
Modifying other job options
If LSB_MOD_ALL_JOBS is specified in lsf.conf, the job owner or the LSF
administrator can use the
bmod command to modify the following job options for
running jobs:
◆ CPU limit (-c [hour:]minute[/host_name | /host_model]|-cn)
◆ Memory limit (-M mem_limit | -Mn)
◆ Run limit (-W run_limit[/host_name | /host_model]|-Wn)
◆ Standard output file name (-o output_file | -on)
◆ Standard error file name (-e error_file | -en)
◆ Rerunnable jobs (-r | -rn)
In addition to resource reservation, these are the only
bmod options that are valid for
running jobs. You cannot make any other modifications after a job has been dispatched.
An error message is issued and the modification fails if these options are used on
running jobs in combination with other
bmod options.
Modifying resource limits for running jobs
The new resource limits cannot exceed the resource limits defined in the queue.
To modify the CPU limit of running jobs, LSB_JOB_CPULIMIT=Y must be defined
in
lsf.conf.
To modify the memory limit of running jobs, LSB_JOB_MEMLIMIT=Y must be
defined in
lsf.conf.
Limitations
Modifying remote running jobs in a MultiCluster environment is not supported.
To modify the name of job error file for a running job, you must use
bsub -e or
bmod -e to specify an error file before the job starts running.
For more information
See Administering Platform LSF for more information about job output files, using
job-level resource limits, and submitting rerunnable jobs.