Merger of Pleiades PBS Queues and Nodes on August 16th14 Aug, 2012
During dedicated time on Thursday, Aug. 16th, from 9 - 11 AM Pacific Time, NAS staff will merge the Pleiades server pbspl3 with pbsl1. With this change, plus simple script updates, you'll no longer have to keep track of which server to use and specify the server name in your qsub or qstat commands. During this time, no jobs will run on pbspl3, and existing queued jobs will move over to pbspl1. Once the merger is complete, pbspl1 will provide all PBS service for Pleiades, and the pbspl3 PBS server will be turned off. The merger will facilitate active management of the resources available to the devel queue, growing or shrinking in response to demand. Modify Your ScriptsIf you have scripts or other tools that submit jobs to (or get status for) jobs from pbspl3, please modify them to interact with pbspl1. For example, if you currently use the following command to get status on all your jobs:
after the merger you will use this command:
GPU Users Take NoteWhen you submit jobs for the various Pleiades GPU queues, the GPU nodes will now have the attribute model=wes_gpu. This means that if you have scripts for GPU jobs that specify the model type, you will need to update those scripts. Get More InfoFor more specific usage details, you can use the search terms "devel" and "gpu" and do customized searches in the user Knowledge Base. If you have any questions or experience any problems after the change, please contact NAS Control Room staff at: (800) 331-8737, (650) 604-4444 or support@nas.nasa.gov. |