Page History

...

For example, user aaa777 would run

$ ssh aaa777@gadi.nci.org.au

...

In order to run graphical tools, you need to enable X Windowing system on local system before ssh. This can be done by running X-Server like XQuartz (Mac), MobaXterm (MS Windows), startx or similar (Linux). Then ssh to gadi: 

$ ssh -Y aaa777@gadi.nci.org.au

...

For example, to take a look at the snapshot of the usage by the project xy11 on the filesystem /scratch, run

$ nci-files-report -f scratch -g xy11

...

For example, aaa777 runs the following command line in the local terminal 

$ scp input.dat aaa777@gadi-dm.nci.org.au:/home/777/aaa777

...

If the transfer is going to take a long time, there is a possibility that it could be interrupted by network instability. For that reason, it is better to start the transfer in a resumable way. For example, the following command line allows user aaa777 to download data in the folder /scratch/a00/aaa777/test_dir on Gadi onto the current directory on their local machine using ‘rsync’.

$ rsync -avPS aaa777@gadi-dm.nci.org.au:/scratch/a00/aaa777/test_dir ./

...

To submit a job defined in a submission script, called for example ‘job.sh’, run

$ qsub job.sh

on the login node. Once the submission is accepted, the above command returns the jobID, for example, 12345678.gadi-pbs, which can be used for tracking its status.

...

To submit an interactive job, run ‘qsub -I’ on the login node. For example, to start an interactive job on Gadi’s gpuvolta queue through project a00 with the request of 48 CPU cores, 4 GPUs, 380 GiB memory, and 200 GiB local disk for 5 minutes on one gpu compute node, run

$ qsub -I -qgpuvolta  -Pa00 -lwalltime=00:05:00,ncpus=48,ngpus=4,mem=380GB,jobfs=200GB,storage=gdata/a00,wd

...

To submit a copyq job, called for example job.sh, define the tasks in the submission script, specify the queue to be copyq, and run

$ qsub job.sh

on a login node.

...

To look up the status of a job in the queue, run the command ‘qstat’. For example, to lookup the job 12345678 in the queue, run

$ qstat -swx 12345678

If the job is running, you would see something like

...

To see how much CPU and memory the job actually has been using, run the command ‘nqstat_anu’. For example, to look up the info of the job 12345678, run

$ nqstat_anu 12345678

The output shows the CPU utilisation rate in the column ‘%CPU’ and the peak memory usage in the column ‘RSS’ and ‘mem’.  The example output below says the job used only 23% of the compute capacity of the requested 48 cores in the elapsed 36 minutes and 47 seconds. Depending on the tasks running inside the job, this percentage may increase while the job proceeds further.  We normally recommend users aim for at least 80% overall CPU utilisation rate.

...