$ ssh simonw@cori.nersc.gov
Use the home directory to store data, but don't run codes here as space is limited. See Compiling Codes at NERSC
Use the scratch directory to compile and run codes, but don't use scratch as longterm storage.
$ cd $SCRATCH
There is a long-term storage solution called HPSS: http://www.nersc.gov/users/storage-and-file-systems/hpss/
Save a batch script as a separate file (e.g. mybatch ). Submit with
$ sbatch mybatch
Before launching a multi-hour run, use a debug script to make sure you set everything up right. Only one debug run can be active at once, time-limit is 30 minutes.
Debug Script
#!/bin/bash -l #SBATCH -p debug #SBATCH -N 64 #SBATCH -t 00:20:00 #SBATCH -J my_job srun -n 2048 ./mycode.exe
Regular batch script
#!/bin/bash -l #SBATCH -p regular #SBATCH -N 2 #SBATCH -t 00:30:00 cd $SLURM_SUBMIT_DIR # optional, since this is the default behavior srun -n 64 ./my_executable
Use the account number m2560 (default acct) for runs.
sbatch mybatch_script
Check your job status
$ sqs
Cancel a job
$ scancel jobid
Alter options for currently queued job (does not work on running jobs)
$ scontrol update jobid=jobid partition=new_partition timelimiit=timelimit qos=normal