bsub

Enable VM job checkpointing by specifying an initial checkpoint time and recurring checkpoint interval. Dynamic Cluster creates the first VM job checkpoint in the specified number of minutes (initial_minutes) after the job dispatched. Dynamic Cluster creates subsequent checkpoints in the specified number of minutes (interval_minutes) after the previous checkpoint.

The job must be rerunnable, either by using the bsub -r option, or by submitting the job to a rerunnable queue.

VM job checkpointing cannot be used together with LSF checkpointing (bsub -k).

Specify the name of one or more Dynamic Cluster templates that the job can use, or use the any or none keywords. Separate names in the list with space. Using this option makes the job use Dynamic Cluster provisioning.

Specify -dc_tmpl none to use none of the templates and the job also ignores the -dc_mtype and -dc_vmaction options. This cancels the default Dynamic Cluster template configured by the default application profile (DC_MACHINE_TEMPLATES in lsb.applications) or esub to allow the end user to the choice of using only non-Dynamic Cluster hosts. (with -dc_tmpl)

When you define Dynamic Cluster templates on the command line, DC_MACHINE_TEMPLATES in lsb.applications is ignored.

Specify -dc_mtype vm if you want the Dynamic Cluster job to be a VM job.

Specify -dc_mtype pm if you want the job to run on physical machines.

By default, Dynamic Cluster jobs run on any machines. The syntax -dc_mtype any is supported to specify this explicitly (physical machines are selected before virtual machines).

When you define Dynamic Cluster templates on the command line, DC_MACHINE_TYPE in lsb.applications is ignored.

The following are a list of preemption actions that you can specify with this option:

• -dc_vmaction savevm: Save the VM.

  Saving the VM allows this job to continue later on. This option defines the action that the lower priority (preempted) job should take upon preemption, not the one the higher priority (preempting) job should initiate.

• -dc_vmaction requeuejob: Kill the VM job and resubmit it to the queue.

  The system kills the VM job and submits a new VM job request to the queue.

• -dc_vmaction livemigvm: Live migrate the VM (and the job running on it) from one hypervisor host to another.

  The system migrates the job to the destination host, then releases all resources normally used by the job from the hypervisor host. During this time, the job remains in a RUN state.

  RHEL KVM hypervisors do not support live migration with VM job checkpointing. Do not use -dc_chkpntvm with -dc_vmaction livemigvm.

The action defines the behavior only when this job is the lower priority (preempted) job, not the higher priority (preempting) job.

When you define the preemption action on the command line, DC_VM_PREEMPTION_ACTION in lsb.applications is ignored.

When using livemigvm to specify a live migration preemption action, you can also specify a second preemption action to trigger if the live migration action fails. Use a space to separate the two actions and quotation marks to enclose the two actions:

-dc_vmaction "livemigvm[details] second_action"

For example,

bsub -dc_tmpl rhel58_vm -dc_mtype vm -dc_vmaction \
"livemigvm[wait_trigger_time=100:livemig_max_downtime=0:livemig_max_exectime=1000] requeuejob" \
myjob

If the live migration fails (because the trigger time or execute time is exceeded), the requeuejob action triggers.