sometime in the future, depending on resource availability and

 

batch system scheduling policies.

 

Batch job submissions typically provide instructions on I/O

 

management, such as files from which to read input and

 

filenames to collect output.

 

By default, LSF jobs are batch jobs. The output is e-mailed to

 

the user, which requires that e-mail be set up properly. SLURM

 

batch jobs are submitted with the srun -bcommand. By

 

default, the output is written to

 

$CWD/slurm-SLURMjobID.outfrom the node on which the

 

batch job was launched.

 

Use Ctrl-C at any time to terminate the job.

Interactive batch job

A job submitted to LSF or SLURM that maintains I/O

 

connections with the terminal from which the job was

 

submitted. The job is also subject to resource availability and

 

scheduling policies, so it may pause before starting. After

 

running, the job output displays on the terminal and the user

 

can provide input if the job allows it.

 

By default, SLURM jobs are interactive. Interactive LSF jobs

 

are submitted with the bsub -Icommand.

 

Use Ctrl-C at any time to terminate the job.

Serial job

A job that requests only one slot and does not specify any of

 

the following constraints:

 

mem

 

tmp

 

mincpus

 

nodes

 

Serial jobs are allocated a single CPU on a shared node with

 

minimal capacities that satisfies other allocation criteria. LSF

 

always tries to run multiple serial jobs on the same node, one

 

CPU per job. Parallel jobs and serial jobs cannot run on the

 

same node.

Pseudo-parallel job

A job that requests only one slot but specifies any of these

 

constraints:

 

mem

 

tmp

 

nodes=1

 

mincpus > 1

Pseudo-parallel jobs are allocated one node for their exclusive use.

NOTE: Do NOT rely on this feature to provide node-level allocation for small jobs in job scripts. Use the SLURM[nodes] specification instead, along with mem, tmp, mincpus allocation options.

LSF considers this job type as a parallel job because the job requests explicit node resources. LSF does not monitor these additional resources, so it cannot schedule any other jobs to the node without risking resource contention. Therefore LSF

90 Using LSF

Page 90
Image 90
HP XC System 4.x Software Batch system scheduling policies, Batch job submissions typically provide instructions on I/O