Difference between revisions of "Ada:Batch Queues"
(→Queues) |
(→Queues) |
||
| Line 8: | Line 8: | ||
These settings control whether a job will lie idle in the queue or be dispatched quickly for execution. | These settings control whether a job will lie idle in the queue or be dispatched quickly for execution. | ||
| − | The current ( | + | The current (May 2015) queue structure. It is in flux. It should consolidate into something more |
permanent by, say, the end of March 2015. | permanent by, say, the end of March 2015. | ||
<pre> | <pre> | ||
| − | Queue Min/Default/Max Cpus Default/Max Walltime Compute Node Types | + | Queue Min/Default/Max Cpus Default/Max Walltime Compute Node Types Notes |
special None 1 hr / 36 hr All | special None 1 hr / 36 hr All | ||
devel 1 / 1 / 320 10 min / 1 hr All | devel 1 / 1 / 320 10 min / 1 hr All | ||
| Line 18: | Line 18: | ||
short 4 / 4 / 8000 1 hr / 5 hr 64 GB and 256 GB nodes | short 4 / 4 / 8000 1 hr / 5 hr 64 GB and 256 GB nodes | ||
medium 4 / 4 / 4000 5 hr / 24 hr 64 GB and 256 GB nodes | medium 4 / 4 / 4000 5 hr / 24 hr 64 GB and 256 GB nodes | ||
| − | long 4 / 4 / 2000 24 hr / 7 days 64 GB and 256 GB nodes | + | long 4 / 4 / 2000 24 hr / 7 days 64 GB and 256 GB nodes Maximum of 6000 cores for all running jobs in this queue |
xlarge 1 / 1/ 280 1 hr / 200 hr 1 TB nodes (11) 2 TB nodes (4) | xlarge 1 / 1/ 280 1 hr / 200 hr 1 TB nodes (11) 2 TB nodes (4) | ||
vnc 1 / 1 / 20 1 hr / 6 hr All 30 nodes with GPUs | vnc 1 / 1 / 20 1 hr / 6 hr All 30 nodes with GPUs | ||
Revision as of 16:48, 11 May 2015
Contents
Queues
LSF, upon job submission, sends your jobs to appropriate batch queues. These are (software) service stations configured to control the scheduling and dispatch of jobs that have arrived in them. Batch queues are characterised by all sorts of parameters. Some of the most important are: (1) the total number of jobs that can be concurrently running (number of run slots); (2) the wall-clock time limit per job; (3) the type and number of nodes it can distpatch jobs to; (4) which users or user groups can use that queue; etc. These settings control whether a job will lie idle in the queue or be dispatched quickly for execution.
The current (May 2015) queue structure. It is in flux. It should consolidate into something more permanent by, say, the end of March 2015.
Queue Min/Default/Max Cpus Default/Max Walltime Compute Node Types Notes special None 1 hr / 36 hr All devel 1 / 1 / 320 10 min / 1 hr All small 1 / 1 / 3 1 hr / 120 hr 64 GB and 256 GB nodes short 4 / 4 / 8000 1 hr / 5 hr 64 GB and 256 GB nodes medium 4 / 4 / 4000 5 hr / 24 hr 64 GB and 256 GB nodes long 4 / 4 / 2000 24 hr / 7 days 64 GB and 256 GB nodes Maximum of 6000 cores for all running jobs in this queue xlarge 1 / 1/ 280 1 hr / 200 hr 1 TB nodes (11) 2 TB nodes (4) vnc 1 / 1 / 20 1 hr / 6 hr All 30 nodes with GPUs
LSF determines which queue will receive a job for processing. The selection is determined mainly by the resources (e.g., number of cpus, wall-clock limit) specified, explicitly or by default. There are exceptions: (1) the the xlarge queue that is associated with nodes that have 1TB or 2TB of main memory; and, (2) the special queue which gives one access to all of the compute nodes. To use either one you must specify the -q xlarge option in your job file as well as have special permission for such access. Output from the bjobs command contains the name of the queue associated with a given job.
... pending ...
Public and Private/Group Queues
... pending ...
The Interactive Queue
... pending ...
