25.06.2015 Views

Administering Platform LSF - SAS

Administering Platform LSF - SAS

Administering Platform LSF - SAS

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Chapter 16<br />

Resource Allocation Limits<br />

Resource allocation limits and resource usage limits<br />

Resource allocation limits are not the same as resource usage limits, which<br />

are enforced during job run time. For example, you set CPU limits, memory<br />

limits, and other limits that take effect after a job starts running. See Chapter 26,<br />

“Runtime Resource Usage Limits” for more information.<br />

How <strong>LSF</strong> enforces limits<br />

How <strong>LSF</strong> counts resources<br />

Job slot limits<br />

Resource<br />

reservation and<br />

backfill<br />

MultiCluster<br />

Resource allocation limits are enforced so that they apply to:<br />

◆<br />

◆<br />

◆<br />

◆<br />

Several kinds of resources:<br />

❖ Job slots by host<br />

❖ Job slots per processor<br />

❖ Memory (MB or percentage)<br />

❖ Swap space (MB or percentage)<br />

❖ Tmp space (MB or percentage)<br />

❖ Software licenses<br />

❖ Other shared resources<br />

Several kinds of resource consumers:<br />

❖ Users and user groups (all users or per-user)<br />

❖ Hosts and host groups (all hosts or per-host)<br />

❖ Queues (all queues or per-queue)<br />

❖ Projects (all projects or per-project)<br />

All jobs in the cluster<br />

Combinations of consumers:<br />

❖ For jobs running on different hosts in the same queue<br />

❖ For jobs running from different queues on the same host<br />

Resources on a host are not available if they are taken by jobs that have been<br />

started, but have not yet finished. This means running and suspended jobs<br />

count against the limits for queues, users, hosts, projects, and processors that<br />

they are associated with.<br />

Job slot limits often correspond to the maximum number of jobs that can run<br />

at any point in time. For example, a queue cannot start jobs if it has no job<br />

slots available, and jobs cannot run on hosts that have no available job slots.<br />

When processor or memory reservation occurs, the reserved resources count<br />

against the limits for users, queues, hosts, projects, and processors. When<br />

backfilling of parallel jobs occurs, the backfill jobs do not count against any<br />

limits.<br />

Limits apply only to the cluster where lsb.resources is configured. If the<br />

cluster leases hosts from another cluster, limits are enforced on those hosts as<br />

if they were local hosts.<br />

<strong>Administering</strong> <strong>Platform</strong> <strong>LSF</strong> 259

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!