epr_beowulf is our local client that allows submission of image restorations to our compute cluster.
epr_beowulf creates a job file named after the user, followed by a period and then job id. The job file is initially copied to /storage/big1/epr/queues/standard.
A restoration daemon, eprd.UMMS, located on m13.umassmed.edu monitors the queue directory and will move jobs to /storage/big1/epr/queues/running to execute a restoration.
eprd.UMMS executes the script run_epr, which determines the correct pararemeters needed to submit the job to Torque running on m13.umassmed.edu.
All queues located in /storage/big1/epr/queues
- Starting queue for all job submissions
- Queue for all running restorations
- Queue for paused jobs.
- These jobs will be started before any jobs in the standard queue
- Put jobs here that you will want to start later
Any jobs found in the Running queue will be launched by Torque running on m13.umassmed.edu
The following commands are available
- Status of jobs or queues
- Manually submit a job
- Delete a job
- Show the compute nodes
- A user-friendly point and click interface to many Torque commands.q
We currently have 7 8-core machines.